LICENSED FOR DISTRIBUTION

Critical Capabilities for Business Intelligence and Analytics Platforms

Published: 02 March 2017 ID: G00303250

Analyst(s):

Summary

The BI market has shifted to more user-driven, agile development of visual, interactive dashboards with data from a broader range of sources. Data and analytics leaders should augment or upgrade traditional BI platforms to modern platforms that improve business value and speed time to insight.

Overview

Key Findings

  • Core capabilities from traditional BI vendors have largely caught up to data discovery vendors who initially disrupted this market, although differences remain at the subcriteria level and in the degree of excellence exhibited.

  • The next wave of disruption in the form of smart data discovery has begun, with larger vendors innovating first or acquiring startups.

  • Although this is a crowded market, significant differences remain in functionality, and in which products are most appropriate for a given use case.

Recommendations

Data and analytics leaders looking to modernize their business intelligence and analytics should:

  • Expand BI and analytics tool portfolios beyond traditional BI platforms, either by augmenting or by evaluating improved capabilities and product roadmaps of incumbent vendors.

  • Assess which BI and analytics products are best for your organization based on the use cases needed by each business function and the sweet spot for that product.

  • Establish the degree to which centralized IT teams can keep up with demand for new data sources and analyses, along with the skills levels and readiness of business units to perform more of their own data preparation and analytics.

  • Embrace easier-to-use, more-agile tools as greater responsibility for analytics shifts to lines of business.

  • Assess the measures that directly influence customer satisfaction with a BI and analytics vendor on top of an evaluation of functionality, integration and cost-of-ownership requirements.

Strategic Planning Assumptions

By 2020, smart, governed, Hadoop/Spark-, search- and visual-based data discovery capabilities will converge into a single set of next-generation data discovery capabilities as components of modern BI and analytics platforms.

By 2021, the number of users of modern BI and analytics platforms that are differentiated by smart data discovery capabilities will grow at twice the rate of those that are not, and will deliver twice the business value.

By 2020, natural-language generation and artificial intelligence will be a standard feature of 90% of modern BI platforms.

By 2020, 50% of analytic queries will be generated using search, natural-language processing or voice, or will be autogenerated.

By 2020, organizations that offer users access to a curated catalog of internal and external data will realize twice the business value from analytics investments than those that do not.

Through 2020, the number of citizen data scientists will grow five times faster than the number of data scientists.

What You Need to Know

This Critical Capabilities research is a companion to the 2017 "Magic Quadrant for Business Intelligence and Analytics Platforms."

The BI and analytics platform market has undergone a fundamental shift away from more IT-centric solutions to business-user-driven solutions. Although traditional ad hoc query tools allow power users to author reports, they often still require an upfront IT modeling effort, often via a semantic layer and data warehouse. In contrast, modern BI and analytic platforms often use a self-contained in-memory engine with minimal to no upfront modeling requirements. This architecture allows a wider range of business users to perform interactive analysis, without the need for advanced technical or data science skills. As demand from business users for pervasive access to data discovery capabilities grows, IT wants to deliver on this requirement without sacrificing governance — in a managed or governed data discovery mode.

The need for governed reporting using an agile centralized BI provisioning model to run businesses remains. However, there is a significant change in how companies are satisfying their governed reporting requirements: Companies are asking if the data discovery tool can be used to fulfill the full spectrum of BI and analytic requirements. They would like to leverage the higher ease of use, higher business benefits, and lower cost of ownership to deliver enterprise reporting requirements. Here we have seen vendors who started out as point solutions for individual analysts in a decentralized use case evolve their capabilities to handle enterprise governance features with better report distribution and KPI alerting.

As data discovery startups evolve their capabilities to meet a broader range of use cases, traditional BI vendors are striking back, hoping to be early in the next wave of disruption in the form of smart data discovery.

Smart data discovery leverages machine learning to prepare and cleanse data more intelligently, automatically generate the most important insights, and interpret charts via natural-language generation. Vendors are at different levels of product maturity.

The degree to which a customer must mix and match capabilities from multiple vendors to get the best of both worlds varies significantly, depending on the vendor and product in question. The modern BI platform includes the broad range of capabilities for agile, interactive visual exploration, as well as governance and report distribution. Customers may start with a decentralized analytics use case, and later look for governance and promotability. In other instances, customers may immediately start with a governed BI use case, essentially trying to replace the former IT-centric reporting platform with a more agile, modern solution.

Analysis

Critical Capabilities Use-Case Graphics

Figures 1 through 5 show aggregate product scores across the 15 critical capabilities that have been weighted for each use case. Each of the products/services has been evaluated on the critical capabilities (and their subcriteria) on a scale of 1 to 5:

1 = Poor or Absent: Most or all defined requirements for a capability are not achieved.

2 = Fair: Some requirements are not achieved.

3 = Good: Meets requirements.

4 = Excellent: Meets or exceeds some requirements.

5 = Outstanding: Significantly exceeds requirements.

Weightings have been applied to individual subcriteria to determine the score for each capability. Scores and weightings represent a combination of customer survey results and analyst opinion.

A definition of the critical capabilities and the subcriteria evaluated are described in the Critical Capabilities Definition and Use Cases sections. Capability weightings, scores by capability by vendor, and scores by use case by vendor are shown in Figures 6 through 8. Each vendor section details which platform product components were evaluated for each vendor to arrive at a composite score.

Although Gartner has provided recommended weightings for each critical capability and use case, individual customer requirements vary greatly. Customers are advised to use the web-based interactive version of this Critical Capabilities research to set their own weightings. Further, the rankings in Figures 1 through 5 may provide a useful prioritization, but customers should study the differences in capability scores in Figure 6 to assess each vendor's strengths, weaknesses and acceptable trade-offs.

Figure 1. Vendors' Product Scores for the Agile Centralized BI Provisioning Use Case
Research image courtesy of Gartner, Inc.

Source: Gartner (March 2017)

Figure 2. Vendors' Product Scores for the Decentralized Analytics Use Case
Research image courtesy of Gartner, Inc.

Source: Gartner (March 2017)

Figure 3. Vendors' Product Scores for the Governed Data Discovery Use Case
Research image courtesy of Gartner, Inc.

Source: Gartner (March 2017)

Figure 4. Vendors' Product Scores for the OEM or Embedded BI Use Case
Research image courtesy of Gartner, Inc.

Source: Gartner (March 2017)

Figure 5. Vendors' Product Scores for the Extranet Deployment Use Case
Research image courtesy of Gartner, Inc.

Source: Gartner (March 2017)

Scores in Figure 6 reflect a combination of analyst opinion and customer opinion based on products released before 15 January 2017.

Figure 6. Product/Service Rating on Critical Capabilities
Research image courtesy of Gartner, Inc.

Source: Gartner (March 2017)

Scores in Figure 7 represent analyst opinion only and specifically exclude customer opinions that can sometimes inflate as well as suppress product capability scores as customers work with different definitions and expectations. Gartner clients may also want to consult Gartner Peer Insights for additional customer opinions from customers in which vendors did not supply a list of customer references.

Figure 7. Product/Service Rating on Critical Capabilities (Analyst Opinion Only)
Research image courtesy of Gartner, Inc.

Source: Gartner (March 2017)

Vendors

Alteryx

Alteryx offers a workflow-based platform for data preparation and building of parameterized analytic applications. Alteryx Designer is a desktop application that can be used for stand-alone advanced analytics or for self-service data preparation that can then be output to partner applications such as Tableau, Qlik Sense or Microsoft Power BI. Alteryx Server enables data engineers to publish datasets for governance and sharing. Alteryx Analytics Gallery is a cloud-based application for sharing analytic apps and supporting browser-based interactivity within the dashboards.

Alteryx is on an annual major release cycle with minor releases throughout the year, typically quarterly. In the last year, Alteryx has added support for more cloud and big data sources, such as Amazon Aurora, Google Sheets and Adobe Analytics. Its location-based analytics, which were already strong, have further improved with support for more international regions. Version 10.6 is the focus of this evaluation.

Decentralized analytics is the primary use case for Alteryx (73% of surveyed customers), and 43% of customers use it for agile centralized BI provisioning.

Strengths

  • Self-service data preparation: Alteryx allows power users, such as citizen data scientists, to combine data from multiple data sources while also transforming and cleansing data. Surveyed customers report using an average of nine data sources per application, putting the product in the top third of the vendors included in this research for this metric. It provides connectivity to a broad range of data sources, including JSON, XML, direct HDFS, Spark, Impala, Google Big Query, and a broad range of relational databases. For data scalability, Alteryx supports push-down processing to a number of leading databases. Alteryx is in the top quartile for fastest time to create complex reports, and customers rate the ease of use for authoring complex reports as the easiest.

  • Advanced and location analytics: Alteryx embedded advanced analytics are rated outstanding overall. It supports forecasting and clustering via a menu-driven interface, along with more than 60 R-based functions, allowing these to be used either in the data preparation process or as output columns for an application. Models can also be output to PMML or R for refinement in other data science platforms. Alteryx has its origins with the U.S. Census Bureau, and supports spatial analytics using a range of maps down to street level for a number of world regions. It also supports drive time and radius geospatial calculations.

  • Scheduled reports: Alteryx allows formatted reports to be distributed in a variety of formats such as PDF, PowerPoint and Excel on a scheduled basis, with email notification. While this capability is typical in traditional BI platforms, it is lacking in many of the modern BI and analytic products. These schedules can be set based on system or business events, such as low inventory.

Areas of Improvement

  • Visual exploration for consumers: Alteryx rates only Fair for its visual exploration capabilities. The ability to manipulate data or author content is only supported in the desktop interface, not via the browser. In this regard, information consumers mainly interact with a highly parameterized dashboard, as opposed to performing more free-form exploration. Alteryx lacks the ability to automatically display numeric values as percentages, link multiple visualizations on a page, or create groups via a point-and-click interface. Particular chart types must be specified at design time, with no support for trellis or histogram charts.

  • No native mobile: Alteryx does not offer native mobile apps, nor specific support for mobility outside of generic, browser-based access. While there may not be high demand for mobile support for the development of data blending workflows, the Alteryx Analytics Gallery could benefit from improvements to its content-consumption experience through support for responsive design when creating content, or through the addition of native mobile apps.

  • Other gaps: Alteryx does not natively support dashboard layouts. Scores for this capability are still rated as Fair to Good because of its support for subcriteria related to mapping. Cloud capabilities are limited to AWS deployment for the Analytics Gallery, with lack of support for hybrid connectivity to on-premises data sources and no additional software security certifications. Within the publish, share and collaborate capability, Alteryx lacks discussion threads, storytelling and the ability to rate content.

Birst

Birst provides a full range of data management and analytic capabilities on multitenant cloud architecture through a software as a service (SaaS)-based delivery model. Birst Enterprise Cloud can be deployed in a public or private cloud or in a customer's data center; the same underlying product — branded as Birst Enterprise Virtual Appliance — is also offered for on-premises deployments.

In 2016, Birst added enhanced functionality for self-service data preparation, the ability to utilize Exasol as a high-performance in-memory MPP data store, and increased its use of responsive design techniques as part of a "design once, use everywhere" approach to multiple device types.

Birst delivers a major release every three months, with a minor release every two weeks. The focus of this evaluation is version 6.

Birst is most often deployed for the agile centralized BI provisioning use case (43%), followed by OEM or embedded BI (34%).

Strengths

  • Cloud native: Birst's multitenant, cloud-architected platform offers strong support for every aspect of the cloud BI critical capability, with the exception of a marketplace, which it is planning to offer in future. A particular strength is its ability to federate queries and support hybrid data connections between cloud and on-premises data sources in a way that is transparent to the end user. Birst also offers prepackaged applications called Solution Accelerators that bundle prebuilt connectors to cloud data sources (Salesforce, Marketo, NetSuite and Google Analytics, for example) with prebuilt metadata, transformations, and prebuilt reports and dashboards in an out-of the-box solution that customers are able to customize to meet their specific needs.

  • Functional breadth: As a result of its broad capabilities, Birst scored in the top quartile for four out of the five use cases assessed. Its highest scores were for the OEM or embedded BI and extranet deployment use cases. This was largely driven by its broad range of SDKs and APIs used to embed analytic content (where it rated Outstanding), and the capabilities of its cloud architecture. The work the company has done in the area of self-service data preparation in Birst 6 (launched November 2016) further bolsters its suitability for governed data discovery and decentralized analytic use cases. Birst's strength in metadata management within a two-tier architecture (an enterprise data tier and a user data tier) enables the definition and management of a semantic layer for central governance, while also enabling decentralized use in a controlled manner. The addition of what Birst labels "networked BI" instances builds on this core strength through connecting independent user objects and the centrally defined semantic layer enabling agile, user-driven growth and expansion of metadata.

  • Mobile: The Birst Mobile module capabilities are rated Excellent to Outstanding. Going beyond responsive design, Birst supports offline exploration, and mobile collaboration and interaction (expanding and sorting columns on a chart, filtering, revisualization, drilling, notifications, and annotations [text and drawings]). The only area of mobile functionality missing is full GPS integration, which is planned.

Areas of Improvement

  • Embedded advanced analytics: Gaps in the capabilities for embedded advanced analytics remain and Birst scored as Limited in this capability. While the platform does offer core statistical functions natively, it lacks the robust library of embedded advanced algorithms, functions and visualizations that are required for more-complex use cases. The majority of this functionality is instead enabled through its integration with R and Weka, which provides users with an option to build and run models that leverage the underlying Birst data model.

  • Smart data discovery: Like other vendors covered in this report, Birst has yet to address the growing need for smart data discovery in its platform, and lacks this functionality. Birst can (and was one of the first BI offerings to) automatically process data and generate dashboards with KPIs, charts and visualizations when new data is loaded. However, to be considered "smart," it must automatically generate advanced analytic visualizations (such as the ability to visualize correlations or clusters in a dataset, or display a decision tree) and automatically generate models (including forecasting, trends, predictions, clustering, segments, correlations and factor analysis). Neither has it yet addressed natural-language input or output.

  • Social and collaboration: The publish share and collaborate critical capability remains an area of relative weakness for Birst, where it scores as Good. While functionality for diverse output formats, content search, alerting, and printing is mature and complete, other areas are developing. With the release of Birst 6, the company added live discussion threads within the platform, which are shown on a timeline. However, Birst currently offers no support for user ratings of the value of BI content, and lacks system-created recommendations of BI content. It does provide integration with Salesforce Chatter, and customers can embed Birst inside Jive.

Board International

Board delivers a single, integrated system that provides BI, analytics and corporate performance management (CPM) capabilities in a single platform. The focus is to deliver a single and unified information platform as a basis for analytics, planning and budgeting, and consolidation. A key differentiation is the hybrid in-memory self-contained platform built on Board's Hybrid Bitwise Memory Pattern (HBMP) algorithms. Board's platform is available on-premises and in the cloud, including a public cloud service offering. Board provides its own proprietary library of advanced analytics functions, Board Enterprise Analytics Modeling (BEAM).

Board modernized its user interface with a mobile-first design ethos, and improved the storytelling and collaboration features on the platform. Board's current release cycle is one major release per year, and one minor release per quarter. The current version is 10. Board has started to invest in smart data discovery features and collaboration on its near-term roadmap.

The two most prominent use cases in the survey for Board are agile centralized BI provisioning (67%) and traditional IT-centric reporting (57%), followed by decentralized analytics (50%). The average deployment size increased over previous years to just over 1,800 users and is slightly above survey average now (1,182), indicating broader deployments across its customer base.

Strengths

  • Mature platform: With its long-standing legacy of developing a single platform, several critical capabilities are strengths for the platform. Platform administration features, such as authentication and authorization, scalability and performance, got Excellent scores. As a single platform, it also supports multiple workflow and allows to customize them. Board natively supports a broad range of relational and multidimensional data sources. Connectors also exist for several enterprise applications, be it on-premises or in the cloud. Other web sources, such as Twitter or Facebook, are supported through the OData connector. Board also supports Hadoop and NoSQL sources.

  • Self-contained in-memory platform: Board's hybrid data platform offers Good to Excellent capabilities for built-in data storage and data loading. In the capability for self-service data preparation, Board's platform is Good in supporting business user data modelling and data mashup, as well as supporting data inference and data enrichment. Data Fast Track enables business users to develop their own data models independent from the platform and to seamlessly promote them to the platform for reuse.

  • Analytic content creation and exploration: Board offers a comprehensive set of embedded advanced analytics functions through its proprietary library of advanced analytics functions, Board Enterprise Analytics Modeling (BEAM). Board continues to introduce new advanced algorithms and functions through BEAM. Interactive visual exploration is well-supported on the platform, for instance, with a broad range of chart types, global filters, binning or linking visualizations. Conditional formatting, color selection and features to enable color consistency are also supported.

Areas of Improvement

  • Embedded BI: Similar to last year, only 10% of surveyed clients indicated that they use Board for embedded analytics and product rates Fair to Good for this critical capability, which is not a focus point for Board's development. Several important software development kit (SDK) capabilities are currently not supported by the platform, such as creating, copying and deleting reports or analytic content, adding users, changing security settings, and performance management and monitoring. Portal integration is supported through iframes and a native Microsoft SharePoint Web Part.

  • Dimensional model: Board supports a wide range of data sources, but customers are limited by the vendor's "cube" concept. Board's core cube architecture is based on multidimensional online analytical processing (MOLAP) or relational online analytical processing (ROLAP), organized by facts and dimensions. A more-flexible data model is not supported.

  • Content authoring and analysis: Integration with R is not available, nor is PMML supported, so clients have to rely on the BEAM library and cannot leverage advanced analytics models developed by data scientists outside the platform. Board started to invest in developing smart data discovery features, such as its "cognitive search," but does not automatically generate insights or natural-language generation. Geospatial and location intelligence capabilities are still limited. Geocoding requires a third-party solution, but no out-of-the-box integration is provided. Only OpenStreetMap maps are fully integrated in BOARD, down to street level.

ClearStory Data

ClearStory Data is a cloud-based BI and analytics platform that allows for smart data preparation and integration, data storytelling, and collaboration in a single platform. It uses Spark-based processing to handle large data volumes. ClearStory is well-suited to business users that need to combine, harmonize, blend, and explore multiple and varied data sources, including personal, cloud, streaming and syndicated data.

As a cloud platform, ClearStory releases new products every three weeks. For major releases, customers can choose not to have an upgrade implemented in their tenant. Major new capabilities delivered in the last year include support for smart data discovery; insights can be generated automatically and include natural-language generation through optional integration with Narrative Science. A number of additional application connections with built-in data inference were added in 2016, including support for Google Analytics, Zendesk and Jira. ClearStory can also act as a data source to Tableau and Microsoft Power BI.

ClearStory is mainly used for decentralized analytics, with 63% of surveyed customers deploying for this use case, followed by 58% for governed data discovery.

Strengths

  • Smart data inference and harmonization: ClearStory was recently awarded a U.S. patent for its smart data inference and harmonization, which uses machine learning. ClearStory can ingest from traditional personal and relational data sources, but can also harmonize these data sources with Hadoop-based and other NoSQL data sources — including Google BigQuery and IBM BigInsights, log files, and streaming data sources. Data is processed using Spark for high levels of query and analytic performance on granular data. The smart data inference will recommend how best to blend and cleanse data but, in addition, the product will suggest other public and premium datasets that are mashable, a capability it refers to as "data you may like."

  • Ease of use: ClearStory Data received the highest customer reference score for ease of implementation and administration, as well as ease of content consumption. Across other ease-of-use drivers — for content creation and visual appeal — ClearStory Data scored in the top quartile. Harmonized datasets can be arranged into an interactive, visually appealing storyboard. In building storyboards, the platform intelligence only exposes functions based on what is in the data. For example, year over year will not be exposed if the data does not include time series. There are also smart recommendations attached to text alerts for some functions that guide users, such as "you did x, now try y." Furthermore, "smart visualizations" will automatically render data using the best-fit visualization.

  • Metadata management: The metadata catalog contains everything ClearStory learns about a source dataset, lineage, refreshes, and about the user activity in a story. As well, the accuracy of inferred dimensions and the accuracy with which multiple datasets can be combined are tracked. It also infers importance by tracking the types and frequency of questions users are asking, how insights are explored, and how users collaborate and augment their analysis.

Areas of Improvement

  • Mainly cloud: ClearStory Data is primarily a cloud BI and analytics solution. It lacks hybrid connectivity for live query of on-premises data sources, although customers can connect to on-premises data sources. This may make the product less suitable for customers with large-scale, on-premises data sources that do not want their data in the cloud. In addition, ClearStory Data relies on its own physical data centers; on a case-by-case basis, ClearStory will work with customers who want to deploy in Amazon Web Services (AWS) or on-premises. The data centers have a number of security certifications — such as SOC2, the Federal Information Security Management Act (FISMA) and ISO 2700 — but this approach limits its geographic reach and provides less flexibility than competitive products that will also allow customers to rely on other cloud infrastructure providers.

  • Embedded advanced analytics: ClearStory only scores Fair to Good for the embedded advanced analytics capability. It supports Spark-based statistical and machine-learning functions, but does not support the ability to call R functions or other third-party libraries. It also lacks native support for menu-driven forecasting and decision trees. Support for K-means clustering was added in the last year, and is supported from within the storyboard.

  • No native mobile: ClearStory uses HTML5 for content consumption and authoring on smartphones and tablets. It does not support native apps that would further allow for location-based analytics, annotations and offline interactivity. There is no out-of-the-box support for integration with third-party mobile device management platforms.

Datameer

Datameer specializes in big data analytics, targeting organizations investing in data lakes and other types of big data environments supporting analytics. The company offers a modern BI and analytics front end with the potential to solve complex problems, leveraging the native query engines for Hadoop and Spark, with support for an expanding range of connectors to other types of data (including SQL relational databases, several file formats, cloud storage platforms, NoSQL databases and web services ranging from enterprise applications to social media and consumer APIs).

Datameer customers report using the platform primarily for agile centralized BI provisioning (60%) and decentralized analytics (60%) use cases. Governed data discovery is less common with 45% of customers deploying in this use case.

Datameer 6 was a major release announced in May 2016, to enhance the user experience and further optimize the smart query engine. A single front end now provides access to multiple steps of the workflow including data access, preparation, analytics and visualization. Spark was also added to Datameer's smart execution engine, which automatically determines the best compute framework, or combination of frameworks for various big data analytics jobs across both small and large datasets. The product is evolving at a rate of one major and quarterly minor releases every year.

Strengths

  • Big data analytics: Datameer has strong capabilities in self-contained ETL and data storage, and native connectors to big data sources. One of the key differentiators of the platform is a patent pending smart execution framework, capable of identifying the right query processing engine for each analytics task — from Tez to Spark, and others, in a way that is transparent to the user. The platform can ingest and process data from multiple sources, but is optimized for big data. The tool also offers a breadth of data sources made available to analysts, including support for SQL-based data sources as well as more-complex data such as IoT and typical digital marketing datasets — including social media.

  • Complex analytics environments: Some of Datameer's strengths are related to the support of different types of challenging data sources (mostly big data related), Excellent self-contained ETL and data storage, and Good to Excellent capabilities to embed analytic content. The multiple performance enhancement features offered by Datameer (such as its smart execution engine, and smart sampling) enable it to address challenging data environments, and help position the tool as the go-to solution where other mainstream data discovery solutions might not meet data scalability and performance requirements.

  • Fast content development: Datameer customers report favorable development times for content across different levels of complexity when compared to the averages of products included in this report. This is particularly impressive given the complexity of the data and analysis done by content authors with Datameer — which often includes big data sources. This ability will appeal to roles such as the citizen data scientist that need to speed the exploration process and time to insight.

Areas of Improvement

  • Low ease of use and visual appeal: Despite a significant update in Datameer 6 to the user front-end experience in 2016, customers report below-average ease of use and low visual appeal, in the bottom quartile of vendors in this Critical Capabilities research. Although the composite rating is Good overall, this is not enough to drive buying in the modern BI and analytics market.

  • Capabilities gaps: Datameer has gaps in a number of key capabilities that are expected features of modern BI and analytics platforms. In particular, interactive visual exploration, analytic dashboards, mobile exploration and authoring, publishing, sharing and collaboration are weaker than with most other products, while self-service data preparation and metadata management show some limitations. A possible way to overcome these shortcomings would be to leverage Datameer's data environment and big data engine on the back end, while using a partner product for the visual exploration (such as Tableau or Microsoft Power BI).

  • Narrow focus on big data: Although Datameer is investing to expand the product's scope, it is still clearly targeting native functionality to support big data use cases (through specialized self-contained ETL and data storage) that rely on technologies such as Hadoop and Spark. Emerging capabilities that are driving innovative offerings in the market — such as smart data discovery, natural-language processing or more-automated data preparation — are not on the vendor's near-term roadmap.

Domo

Domo is a cloud-based business intelligence and analytics platform targeted at senior executives and business users, and is well-suited to management-style interactive dashboards. Domo enables rapid deployment through its native cloud architecture, an extensive set of data connectors and prebuilt content, and an intuitive, modern user experience. Strong social and collaborative features in DomoBuzz make it possible for users to discuss findings, follow alerts, collaboratively develop content, and rate dashboards from the web or mobile devices, including smartphones. Domo includes a web-based and business-oriented data preparation tool called Magic for combining cloud-based and on-premises data sources. Domo Workbench is a desktop tool for loading additional on-premises data sources into Domo datasets in the cloud. These datasets are stored in an OEM version of a cloud-based columnar database. The product runs on Amazon Web Services. Release cycles are continuous (as with most cloud vendors).

During the past year, Domo formalized its channel program and launched developer.domo.com on the newly branded Domo Business Cloud platform — an Appstore that allows Domo and its ecosystem of partners to sell vetted Domo connectors and apps. Also, Domo now supports live customer instances on Amazon Australia, Microsoft Azure and an Equinix colocation — in addition to Amazon US-East previously supported. Amazon Ireland is on the roadmap for 2017.

A higher percentage (74%) of Domo's customer references report using it primarily for decentralized use cases, more than most of the other vendors in this Critical Capabilities. This is consistent with how business people primarily use Domo — for management-style dashboards often deployed in the line of business with little or no support from IT.

Strengths

  • Rapid deployment of management-style dashboards and infographics: Domo offers business people an easy and intuitive interface in which to build interactive "cards" (views) and store them either in "collections" (a way to visually organize cards — instead of in folders) or assemble them into "pages" (Domo's equivalent of dashboards). Users can also build visually appealing infographics using App Design Studio that leverages Adobe Illustrator. Domo's reference customers rate it highly for its intuitive user experience when combining a large number of data sources into business-friendly dashboards. A large percentage of reference customers (in the top quartile) report selecting Domo for ease of use and, because it can combine a large number of data sources into business-friendly dashboards. Domo ranks in the top quartile for ease of use and visual appeal, with an overall Excellent rating for this capability.

  • Collaboration, alerting and scheduling: Domo's "design or assign" abilities enable collaborative content development between a content author and a content consumer. In the assign mode, a user can assign the creation of content to someone else with more skills, which automatically populates a template in both users' favorite pages. This way, the author and the consumer can iterate and discuss until the content is what the user wants. With DomoBuzz, users can participate in discussion threads or groups, and can follow other users. Users are also presented with recommendations based on the behavior of other users as a native part of the product workflow. Users can also rate dashboards and follow content created by particular users. Business-user-defined dynamic alerting and scheduling is extensive and intertwined with the platform's collaborative and social capabilities. Users can create their own scheduled reports as well as add and remove metrics that are shown in a "favorites" tab.

  • Native mobile with a focus on smartphones: Domo's native mobile applications for iOS and Android (Windows Phone devices are supported only through HTML5) optimally render content on both tablets and smartphones. They are well-integrated with DomoBuzz for chat, collaboration and alerts. However, integration with enterprise MDM security providers and offline analysis on a mobile device are not supported.

Areas of Improvement

  • Advanced data exploration and embedded advanced analytics: Domo's analyst-oriented visual exploration and user-based data manipulation features are limited. For example, users can only create reusable groups as new dimensions, or automatically bin data in the dashboard authoring and analysis environment, through a custom calculation. Moreover, while Domo's formula expression editor (referred to as "beast mode") lets users create their own calculations, there are limited automated suggestions that would make it easier for analysts to use the correct syntax. Finally, while business users can create forecasts in a card via a drag-and-drop feature, and analysts can integrate R and Python scripts into calculations, embedding advanced statistical functions in calculations requires the use of JavaScript libraries. There is limited support for other common drag-and-drop functions (such as for clustering and correlations). Users can create linked charts to filter and drill to detail, or drill across to a related card, but there is no way to centrally define, or automatically generate, or infer hierarchies (time or geography, for example) to support free-form drilling. Consistent with Domo's primary use for simpler management dashboards, reference customers scored Domo's complexity of analysis in the bottom third.

  • Self-service data preparation a work in progress: While Magic, Domo's self-service capability, provides a web-based point-and-click design process for accessing web-based and on-premises relational data sources to load data into Domo, its data inference and profiling capabilities are a work in progress. Users can promote, collaborate and reuse a dataset, but they can't reuse individual metadata objects. Workbench — Domo's desktop tool for administrators to load on-premises data into its cloud — doesn't have the same ease of use and visual appeal as its web-based dashboards and data-loading process. Workbench supports manual data mashups, creation of additional calculations, and some data transformations. It lacks a point-and-click graphical user interface to build a query or extract data. According to reference customers, IT and/or tech-savvy business users who are familiar with SQL can handle this part of the data loading process.

  • Cloud-centric approach: Domo's approach requires all data — whether from on-premises sources or cloud applications — to reside in its cloud for visualization and analysis, which may not suit organizations with primarily on-premises data sources. Domo has recently introduced an on-premises version (with limited adoption to date) when this is a requirement. Hybrid data connectivity to on-premises data is on the roadmap. Domo offers a desktop tool specifically built for admin users to load on-premises data into its cloud, but this tool is less business friendly than other components of the platform.

IBM (Cognos Analytics)

IBM Cognos Analytics is one of two product offerings provided by IBM that, together, offer a broad range of BI and analytic capabilities. Cognos Analytics is version 11 of the Cognos Business Intelligence product line and a much-improved and redesigned, modern product offering.

Over the past year, IBM has delivered on adapting its BI and analytics offerings to more-closely align with the market. Cognos Analytics was first released in December 2015. The product is on a continuous release cycle and averages one release per quarter. There were, however, five new releases in 2016 with version 11.0.5 becoming available in November 2016. The product combines both IT-authored content and content authored by business users within one platform. In addition, several modern design elements from Watson Analytics have been incorporated, resulting in an easier-to-use, more-visually-appealing experience. Cognos Analytics can be deployed both on-premises or via the IBM Cloud.

Cognos Analytics is most often used to support the agile centralized BI provisioning use case, as represented by 74% of survey references. It is also often used to support decentralized analytics (47%) and traditional IT-centric reporting (47%).

Strengths

  • Platform management: Robust BI platform administration, security and architecture capabilities provide a solid foundation for the IBM Cognos Analytics solution. The product enables scalability and performance via extensive load balancing features, tunables for managing resource availability, and performance optimizations such as function shipping to the database, multilevel caching and aggregate awareness. The platform supports a number of operating systems including AIX, Linux, Solaris and Windows, and supports a number of open standards. Users are authenticated by directly connecting to any v3 LDAP, including Active Directory. Cognos Analytics also provides an out-of-the-box audit solution to monitor when and how users interact with the platform.

  • Interactive visual exploration and analytic dashboards: Visual exploration and analytic dashboards are rated Good. The extensive chart library leverages IBM's RAVE visualization technology and includes standard chart types as well as heat maps, tree maps, packed bubble charts, geographic maps and more. Each chart is automatically interactive and, with Release 11.0.5, able to be animated. Release 11.0.5 also introduced enhanced mapping and geospatial analytics enabled via a partnership with Mapbox and Pitney Bowes. Users can create their own groups to form new dimensions, interactively display numbers as values or percentages, and easily rank values. Additionally, multiple charts on a page are automatically linked for brushing and filtering.

  • Visually appealing and easy to use: Borrowing from IBM Watson Analytics, IBM Cognos Analytics has a new clean and attractive user interface that enables access to both IT-authored content, and business-user-authored content, via one portal. In addition, several of the authoring interfaces in previous versions have been combined and streamlined in Cognos Analytics. Reports can be authored against existing Framework Manager models and, with release 11.0.5, an author has direct access to relational Framework Manager packages in dashboards. "Smart" capabilities including smart searches and joins, automatic inferences, representation of time and location data. Recommended tasks and visualizations further enhance the user experience and reduce time to insight.

Areas of Improvement

  • Deficient self-contained ETL and data source connectivity: The trend in analytic platforms is to not only extend analytic capabilities to include more-advanced analysis, but to also extend the ability to acquire, organize and store the data for analysis. IBM Cognos Analytics lacks many of the features needed to effectively move from ingesting data to providing insight. Neither does Cognos Analytics provide a true columnar in-memory data store, instead, relying on the creation of datasets using file caching. As a result, self-contained ETL and data storage rated only as Fair. IBM Cognos Analytics does not provide access to unstructured/semistructured data sources. In addition, it has no native connectivity to enterprise applications.

  • Gaps in collaboration: As analytics proliferates throughout the organization, the ability to share and collaborate around analytic content is paramount. The ability to create discussion threads, have real-time collaboration and integrate with other social platforms is currently unavailable. The ability to rate and recommend analytic content based on user ratings or usage patterns is also lacking.

  • Limited extension to additional analytic users and analytics in the cloud: Smart data discovery is a growing trend in the analytics space, allowing the use of the analytic tools by new types of users as well as enabling faster time to insight. Limited smart data discovery capabilities, as well as a lack of embedded advanced analytics, limit the extension of the platform. In addition, the limited ability to embed analytic content in other applications further hampers the ability to extend the use of analytics within both internal- and external-facing applications; embedding is limited to URLs within web portals, which was added in Release 11.0.5. Platform integration and workflow integration are a work in progress. Cognos Analytics' cloud offering is somewhat limited, minimizing the opportunity to extend analytics via the platform. Multitenancy is not yet supported, although it is on the roadmap for 2017. Lack of a marketplace and packaged content, and minimal self-service administration capabilities also minimize the ability to jump-start analytics in the cloud using Cognos Analytics.

IBM (Watson Analytics)

IBM Watson Analytics is one of two products that comprise IBM's BI and analytics offering. Watson Analytics continues to pioneer the next-generation, machine-learning-enabled user experience for analytics, including automated pattern detection, support for natural-language queries and generation, and embedded advanced analytics, via a cloud-only solution.

Over the past year, IBM has delivered on adapting its offerings to more-closely align with the market with its continuous release cycle, averaging one release per quarter. Integration with Cognos Analytics is currently limited to the ability to bring in a Framework Manager package, a list report or a dataset for deeper exploration, and appeals primarily to individual users and workgroups who need to perform smart data discovery. Watson Analytics is also a distinct product from the Watson cognitive solutions offered by IBM (such as Watson for Oncology and Watson Discovery Advisor). There is no integration between Watson Analytics and these later Watson systems.

IBM Watson Analytics is most often used for decentralized analytics as represented by 74% of the survey respondents. An additional 26% of survey respondents use the platform for governed data discovery.

Strengths

  • Easy to use and visually appealing: Watson Analytics provides integrated data access, exploration, dashboarding and exploration capabilities within one platform and enables interaction and exploration via natural-language dialogue, enabling an easy-to-use approach for working with the platform. It rated Good to Excellent for ease of use and visual appeal. In 2016, IBM developed 20 "storybooks" that provide analytic templates for addressing specific business problems. In addition, an expert storybook can be created by an author, which then guides a user by selecting analyses, questions and visualizations that lead to faster analytic insight. Expert storybooks are accessible via Watson Analytics' Analytics Exchange.

  • Smart data discovery and exploration: Watson Analytics is a pioneer in continuing to push the boundaries for smart data discovery. The platform automates many of the steps in the data-access and preparation process, including scoring the data on readiness for analysis and highlighting potential data issues, as well as providing semantic recognition of concepts such as time, place and revenue. In addition, Watson Analytics provides recommended starting points for analysis and targets for prediction by automatically detecting patterns in the data as it is loaded by determining strong correlations and associations. Statistical information about the models, and which algorithms were used, can also be viewed, enabling validation of the model.

  • Cloud-based architecture: Watson Analytics is fully cloud-enabled and accessed via a web browser. The data and content within the platform are hosted in the SoftLayer cloud and data is stored using IBM dashDB, which combines columnar, in-memory capabilities with embedded, in-database analytics. The cloud platform provides robust self-service administration and elasticity capabilities for monitoring, managing and scaling the solution as needed by the end-user organization. The administrator user interface allows monitoring of user licenses, data source connections, space utilized and space available.

Areas of Improvement

  • Minimal self-contained ETL: A precursor to performing data discovery is the ability to extract, transform and load (ETL) the data easily within the platform in preparation for analysis. The self-contained ETL capabilities available within IBM Watson Analytics rated Poor to Fair, the weakest of the vendors included in this Critical Capabilities. The platform does not use existing analytic storage (such as the data warehouse or third-party in-memory engines) but instead requires loading the data into Watson Analytics' own data storage environment. Incremental data loads and parallel/multithread loading of data are also not currently supported, nor is scheduling and monitoring of active data loads.

  • Lacking comprehensive data source connectivity and metadata management: IBM Watson Analytics' data source connectivity capabilities continue to expand, but are still deficient relative to other vendors. Although the platform does provide access to flat files, many relational data stores, and Twitter data, it does not support OLAP connectivity or other data sources such as XML, RSS feeds or JSON. It also supports connectivity to Cloudant, and Apache Hive, Cloudera Impala and Hortonworks HDFS, but does not include Spark. Support for native connectivity to enterprise applications is also limited, with access only to Salesforce. The platform does not provide data lineage or impact analysis.

  • Gaps in sharing findings: The ultimate value of an analytic platform is its ability to share and collaborate throughout the analytic process. IBM Watson Analytics rated Fair for its ability to embed analytic content and its ability to publish, share and collaborate. SDKs are limited to data loading, administration and building connectors. There are no SDKs for printing, parameterization, building workflows, custom visualizations/analytic web applications, or create, copy and delete capabilities. The platform does not support white labeling, and provides portal integration solely via iframes. Chart extensions to support third-party chart libraries, and the ability to access all BI content as an embeddable report via an API, are also currently unavailable. The platform supports limited output formats (including PDF, PPT and image) and does not support scheduling or alerts. The platform also does not support discussion threads, real-time collaboration and timelines, ratings for content, or recommendations based on rating or usage patterns.

Information Builders

This report covers Information Builders' InfoAssist+ product. This is part of the company's integrated WebFOCUS business intelligence and analytics platform, but can be used stand-alone. InfoAssist+ is a combination of visual data discovery, reporting, rapid dashboard creation, interactive publishing, mobile content and the Hyperstage in-memory engine.

During 2016, Information Builders improved the self-service data preparation and visual exploration capabilities in InfoAssist+. It also made changes in packaging and distribution, and is now leading with InfoAssist+ as the introductory edition to all three editions of the WebFOCUS platform (the business user edition, application edition and enterprise edition).

Information Builders releases major new features annually with maintenance releases quarterly. The focus of this evaluation is on InfoAssist+ 8.2.

According to the reference customers surveyed, InfoAssist+ is most often deployed for the agile centralized BI and decentralized analytics use cases (both 52%), closely followed by governed data discovery (46%).

Strengths

  • Platform administration: BI platform administration is rated Excellent to Outstanding. This includes architecture, scalability and performance, and disaster recovery. WebFOCUS is Section 508-compliant, and the server runs on Linux, Unix, IBM (iSeries and System Z), and VMS. For the self-contained ETL and data storage critical capability, it is rated Excellent. InfoAssist+ has its own columnar, in-memory data store, which is included in the product. Unsurprisingly (given Information Builders' heritage, and the data integration capabilities of iWay), data source connectivity is rated Excellent to Outstanding. Connectivity to a range of relational data sources, enterprise applications, big data sources and personal data sources is a core strength. The product supports XML, RSS, SOA, REST, JSON, flat files, Excel and other data sources. In addition, it has native adapters for Facebook, Twitter and Salesforce to consume data for social/sentiment analysis and other reporting.

  • Embed analytic content: Information Builders was rated Excellent to Outstanding in the embed analytic content capability. Its web services REST API allows developers to call a broad range of functionality from another application. Reports and dashboards can be embedded within other portals including Microsoft SharePoint via Web Parts, JSR 168 portlets and iframes. This strength makes InfoAssist+ on WebFOCUS a good match with the OEM and embedded BI and extranet deployment use cases covered in this report, but relatively few customers surveyed use InfoAssist+ in this way, perhaps using IBM's WebFOCUS enterprise edition instead.

  • Mobile: Information Builders supports native mobile apps on a broad range of devices including iOS, Android and via HTML5 for browser-based access. The MobileFavs native app allows users to interact with dashboards and reports using touch and in offline mode. Information Builders supports interactive disconnected analytics via its patented Active Technologies.

Areas of Improvement

  • User-centric functionality: Although strong in IT-led areas, InfoAssist+ has less-capable functionality in the areas that are needed to meet the requirements of our decentralized analytics use case. It gained Good ratings for the interactive visual exploration and analytic dashboards capabilities, areas that are key buying requirements in this space. It has some gaps — for example, the fact that visually driven binning is not supported, or that only a small subset of the chart types available in analytic dashboards are available in InfoAssist+ when doing interactive visual exploration. While rated Limited to Good for the self-service data preparation capability, the company has made progress here in 2016, adding automatic generation of metadata and sample visualization, reports, interactive documents and formatted dashboards immediately from user's data. In the smart data discovery capability, InfoAssist+ lacks the level of automatic generation of visualizations and analytic models needed, and so scores as Limited. Information Builders has partnered with natural-language specialist Yseop, and at the time of writing was about to make natural-language generation functionality generally available.

  • Cloud BI: Although Information Builders has a number of cloud-based partners, adoption has been slow, and it rated Limited to Good only for this capability. The product is architected to be multitenant, but otherwise lacks a turnkey solution for cloud deployment or hybrid connectivity to on-premises data sources. During 2016, it has established new partnerships with Microsoft Azure, Amazon Web Services (AWS) and IBM SoftLayer, better positioning it to take advantage of growing cloud adoption intentions, although it still does not offer a SaaS model directly.

  • Less easy to use: Consistent with prior research, for ease of use, Information Builders' customers rated the vendor as Good overall, but in the bottom quartile relative to the rest of the products covered. In a market in which ease of use significantly influences buying, Good is often not enough. In part, the rating stems from the product's multiple authoring interfaces (for a dashboard, for a visualization, for a report, for a chart) with inconsistent capabilities in each. InfoAssist+ has a UI style like the Microsoft Office ribbon, which may also be viewed as outmoded by end users. In this regard, its visual appeal also ranked in the bottom quartile. The only area where the customers surveyed ranked InfoAssist+ above the bottom quartile for ease of use was in content development.

Logi Analytics

Logi Analytics' BI platform Logi Suite is composed of Logi Info, Vision and DataHub. Logi Analytics is best-known for Logi Info, which is commonly used to embed analytic content in websites and applications, and to enable end-user organizations to extend their BI access externally to customers, partners and suppliers. Logi Vision is the company's data discovery tool, which enables business users to prepare, analyze and share data. Logi's DataHub is a data preparation and columnar data store that ingests, blends and enriches data from multiple sources. Logi Info and Vision can both use DataHub for self-service data preparation.

In 2016, Logi Analytics added enhanced functionality to make self-service data discovery embeddable into analytical applications — for shared authoring of analytics and applications — and expanded data preparation to include more data joining and blending fidelity, as well as faster query performance.

Logi has two major and two minor releases per year. The focus of this evaluation is on version 12.2.

Logi Analytics is most often deployed for the OEM and embedded BI use case (49%), followed by traditional IT-centric reporting (46%).

Strengths

  • Embedded BI: Logi is deployed in an embedded use case by more of its customers than any other vendor surveyed. From a product perspective, Logi rates as Excellent to Outstanding for the embed analytic content critical capability. The Logi Embedded Reports API allows developers to embed content in other applications and call functions to create, add and delete objects. Individual visualizations and/or reports are fully interactive within the third-party application. Portal integration includes JSR 168-compliant portlets, iframes, Microsoft Web Parts and Oracle BPEL portlets.

  • Sharing, collaboration and visualization: Logi offers good functions to drive adoption. It scores Excellent in the publish, share and collaborate critical capability. The Logi Vision Info Board is a modern user interface through which users visualize their most important content based on usage statistics and ratings. Users can pin content to a board. A visual activity stream also shows who is producing new content and making changes. Integrated discussion threads allow users to collaborate on findings and reference users within a comment. The size of a particular visualization is automatically adjusted in the dashboard to reflect its usage and importance. Overall, Logi Analytics' interactive visual exploration is rated Excellent to Outstanding. Users can filter, sort, lasso and drill; and all common chart types are offered. Geographic mapping capabilities automatically interpret location based on names, and data does not need to be geocoded in advance by latitude and longitude. Data manipulation — such as binning and display as percentage variances — is intuitively supported. However, the ability to create new custom groups remains less intuitive.

  • Data connectivity: Logi's self-contained ETL and data storage capabilities are rated as Excellent. It offers a wide range of relational, OLAP, and other data sources (such as XML, RSS, JSON feeds) and connectors to apps. Logi DataHub is used for caching and performance, but customers can also use the in-database processing to run complex SQL in the data source. Its self-service data preparation capabilities are also rated as Good (handling user mashups, data modelling, joins and profiling).

Areas of Improvement

  • Smart data discovery: While Logi's embedded advanced analytics functions have improved (adding support for decision trees and menu-driven forecasting) and are now rated Good to Excellent, its smart data discovery functions are immature, lacking the ability to automatically generate forecasting, trends, predictions, clustering, segments, correlations and factor analysis on data load. While not supporting natural-language queries, Logi is working on natural-language generation with partner Yseop.

  • Metadata handling: Logi's concept of a metadata layer is different for each of its main interfaces and, as such, Logi scores Fair to Good on our metadata management critical capability. When working with Logi's Self-Service Reporting Module (SSM), report authors only need to decide what in the dataset (tables and columns, for example) they want to make available to the end user, as well as the types of joins they want to allow between those data sources. Within Logi's SSM, the Metadata Builder automatically introspects the database catalog and infers relationships between tables and columns. Logi Vision, meanwhile, does not use the same Metadata Builder. The degree that these models can be reused across applications and promoted by users is a work in progress. The optional Logi DataHub module can be used as a unified metadata layer.

  • Mobile: Mobile exploration and authoring could be improved with a better touchscreen experience associated with native mobile device support, and also offline exploration. Currently, all mobile access is via browser-based HTML5. The benefit of this approach, though, is that content can be authored, consumed and interacted with from any tablet or smartphone. Logi does support responsive design (which allows content to be smartly re-rendered depending on the screen dimensions).

Microsoft

Microsoft offers a broad range of BI and analytics capabilities with its Power BI suite, delivered via the Azure cloud. (Microsoft Reporting Services and Analysis Services are covered in our "Market Guide for Enterprise-Reporting-Based Platforms," as on-premises offerings.) Excel is also widely used for data analysis and, while it is not considered here as a BI and analytics tool per se, the integration with Power BI has continued to improve. A number of Excel add-ins that were part of earlier releases of Power BI are native and supported in Office 2016 (Power Query, Power Pivot, Power View and Power Map).

Power BI supports browser-based authoring and visual exploration for cloud data sources, but when authoring complex data mashups involving on-premises data sources, the desktop interface is required. Power BI offers data preparation, data discovery and interactive dashboards via a single design tool. The Cortana Intelligence Suite (reviewed in the "Magic Quadrant for Data Science Platforms" and its companion Critical Capabilities) includes Power BI.

Power BI Desktop can be used as a stand-alone, on-premises option for individual users as part of the decentralized analytics use case, for which 46% of survey customers have deployed it. Agile centralized BI provisioning is the predominant use case (61%). With this use case, central BI teams are modeling the data and publishing the dashboards and reports to the cloud-based Power BI server. Microsoft does support hybrid connectivity to on-premises data sources, but all dashboards and reports must be published to the Microsoft Azure cloud for sharing and collaboration. An option to publish Power BI reports to an on-premises SQL Server Reporting Services deployment is a major roadmap item due in 2017. The on-premises version of Power BI will not support the full range of data sources and features provided in the SaaS version.

Throughout 2016, Microsoft has continued on its monthly release cadence. Major improvements in the last year include the addition of a number of key data sources, embedded BI capabilities, and enterprise features that include row-level security and usage monitoring.

Strengths

  • Ease of use and visual appeal: Microsoft's ease of use rates Excellent overall and is ranked in the top quartile. Visual appeal is an aspect of this, and inquiry customers have frequently said that business users chose Microsoft in competitive proofs of concepts partly for its appealing first impressions. There are a number of specific capabilities that contribute to its ease of use, including that Power BI is a cloud-based product in which Microsoft takes care of the infrastructure for the data storage, processing and sharing environment. In addition, the product has particular features such as Q&As that have a searchlike interface for users to generate visualizations. In addition, Quick Insights is a basic form of smart data discovery that will automatically generate the most-meaningful charts. Despite the high ease-of-use scores, the time for users to create reports in dashboards is higher than the survey average for all report types (simple, moderately complex, complex) with the complex report development time ranking in the bottom third.

  • In-memory engine with data preparation: The Power BI in-memory engine has its origins in Microsoft SQL Server Analysis Services tabular data models. These provide both a flexible and high-performance analytic tier in the cloud. Microsoft limits storage to 10GB per user. Power BI also supports DirectQuery mode for the most-popular data sources, in which data is not replicated into the in-memory engine for greater data scalability. A robust self-contained in-memory engine allows users to mash multiple data sources together in a reusable dataset. The data preparation capabilities, which rate Good overall, allow data modelers to clean and transform the data as they load it.

  • Data sources and prebuilt apps: Microsoft scores Excellent to Outstanding for its breadth of data sources. Microsoft has continued to expand the range of relational data source supported, with SAP Hana added in the last year, and Informix in beta. It also natively supports the Hadoop File System with support for Spark (both third-party and Microsoft Azure HDInsight) also in beta. Prebuilt applications (called "content packs") include data connectors, models, metadata and out-of-the-box dashboards. Power BI users can quickly connect to their accounts in SaaS applications (for example, in Salesforce, Marketo, Zendesk, QuickBooks Online and Google Analytics) and see their data through prebuilt, live dashboards and interactive reports. Microsoft owns and delivers a number of these content packs, and its partner network provides others.

Areas of Improvement

  • Embedded advanced: Microsoft has a number of key ingredients for advanced analytic capabilities across several products. However, none of the advanced analytic capabilities are available out of the box with Power BI. Microsoft gives users the ability to install a local R instance with Power BI Desktop, and to call and embed an R script directly from within Power BI Desktop. However, there is not a straightforward way to perform basic forecasting via a menu-driven option in Power BI. More-advanced visualizations and analytics, such as decision trees and clustering, are not natively supported, but could be possible via extensions in the marketplace. The range of statistical functions natively supported is limited to Data Analysis Expressions (DAX).

  • Lacking important basics: Surprisingly, pivot tables continue to be lacking in Power BI. Even table displays with subtotals are still not supported. While several modern BI products lacked these capabilities in version 1 releases, Power BI Desktop has now been in the market for more than 18 months. For many customers, this may be a showstopper. Microsoft's work-around is for customers to create the tables in Excel and import the content into Power BI. However, this is less than an ideal workflow. The product also lacks a number of formatting options.

  • Disjointed platform components and workflow: The scale-up options for more data storage and processing — whether to HD Insight or SQL Azure Analysis Services — are not straightforward or clear. For the advanced analytics capabilities, the features may also be spread across Excel, Azure Machine Learning and R. While it's reasonable for a vendor to have a distinct product for data scientists, even basic advanced capabilities for business users are disjointed. Hence, Microsoft scores only Limited to Good for this capability.

MicroStrategy

MicroStrategy combines self-service data preparation, visual data discovery and big data exploration with enterprise BI. Version 10, a major release, added substantially enhanced interactive visual exploration capabilities, better promotability of user-built data models and content, improved self-service data preparation, and direct support for HDFS as well as a range of personal data sources. This range of capabilities, delivered in a single integrated platform, makes it better-suited to large-scale system-of-record reporting as well as governed data discovery deployments for larger and more-complex datasets than many other offerings.

MicroStrategy's front-end interface — MicroStrategy Web — is used for data preparation, visual-based exploration, and report and dashboard authoring deployed on MicroStrategy Server. MicroStrategy's Parallel Relational In-Memory Engine (PRIME) is an embedded, in-memory, column-oriented, distributed, analytic data store. MicroStrategy Cloud is a hosted service that includes MicroStrategy BI, an analytical database (which supports Actian Matrix, Microsoft SQL Server, Informatica, Netezza and Teradata among others), and data integration capabilities (such as Informatica PowerCenter, Informatica Cloud, and SQL Server Integration Services [SSIS]). MicroStrategy Cloud now runs on Amazon Web Services (AWS). MicroStrategy Mobile — part of the MicroStrategy Server — is a code-free environment for building native mobile apps for iOS, Android and Windows devices.

MicroStrategy is on a quarterly release cadence. Version 10 point releases in 2016 have introduced a new dossier client and "workstation" to simplify the creation, sharing and viewing of analytic dossiers and briefing books. Workstation capabilities also streamline the configuration and administration of enterprise deployments, including dynamic scaling of MicroStrategy Cloud on AWS. The new workstation is built on new REST APIs in an effort to make the platform more attractive for OEM and embedded use cases. A free desktop version of MicroStrategy 10, available since 3Q16, is well-suited for visual-based data discovery and gives users a risk-free way to try the product.

MicroStrategy is deployed primarily for traditional IT-centric BI (59%), followed by agile centralized BI (57%), and decentralized analytics (48%).

Strengths

  • Fully featured integrated product for all use cases: MicroStrategy has top-quartile product ratings for three of the given use cases. It is an enterprise-grade platform including security, scheduling and distribution with a strong capabilities in support of governed data discovery (including self-service data preparation of complex data models). It is well-suited to companies that need large-scale system-of-record reporting, mobile, dashboards and robust, business-oriented data discovery on large complex datasets in a single platform. Outstanding scores for BI administration, architecture, security, data source connectivity and platform workflow integration anchor this rating.

  • Extensive and widely deployed mobile: MicroStrategy has also been an early innovator in mobile BI, with some of the most-comprehensive, highly rated and widely adopted mobile capabilities. Customers choose MicroStrategy for mobile more often than most other vendors. MicroStrategy Mobile is a fully featured and native mobile development and consumption environment for iOS, Android and BlackBerry. It supports advanced and less-common features such as disconnected analysis, write-back, multifactor authentication, biosecurity, GPS and camera integration, although authoring from a mobile device is not supported. MicroStrategy Mobile can be deployed with the MicroStrategy platform, as a stand-alone mobile solution, or as a complement to other BI platforms. This is a differentiator versus the mobile solutions of other BI platforms.

  • Enterprise-grade platform with modern capabilities: MicroStrategy 10 can support large-scale, trusted, self-service BI. It offers a seamless workflow for promoting business-user-generated data models and content to enterprise sources in MicroStrategy Web. When user data models are promoted to the enterprise, common dimensions are automatically remapped to inherit row-level security. These dashboards and datasets can then leverage other enterprise administration, scalability and distribution features in MicroStrategy Server. Advanced data manipulation capabilities support multisource self-service data preparation that recognizes geographical and time data, and is able to automatically generate the hierarchical elements not available in the source. From a visualization, users can build on-the-fly groups and hierarchies and do drag-and-drop forecasting. MicroStrategy natively connects to HDFS with queries executed in Hadoop or Spark engines, and then stages the data in memory for fast interactive visualization of large datasets and models that natively span modeled and un-modeled relational, personal and Hadoop sources. Version 10 supports Spark SQL and native connectors to big data platforms such as Amazon Elastic MapReduce and IBM BigInsights. Extensive geospatial capabilities are available in the platform (via an OEM version of Esri for free) — although specialized geospatial algorithms are not supported.

Areas of Improvement

  • Collaboration and smart data discovery: MicroStrategy is rated highly across most of Gartner's critical capabilities for this market; however, smart data discovery, and publish, share and collaborate score lower. Smart data discovery features — such as natural-language query, automated insight generation and integrated natural-language generation/narration — are missing in the current product, although recommendations based on user context, interest and usage are on the roadmap.

  • Gaps in cloud: MicroStrategy's single-tenant cloud solution lacks packaged domain and vertical content, and a robust content marketplace for customers and partners. Although MicroStrategy was early to invest in the cloud, it also has among the highest percentage of its reference customers reporting that they have no plans to consider deploying it in this manner.

  • Ease of use: Although an ongoing area of focus for MicroStrategy development, customers still rate its platform as more-difficult to administer and consider it builds content with a less-favorable visual appeal than competing products. For both ease of use and visual appeal, MicroStrategy's reference sores place it in the bottom quartile of vendors included in this research. Although the composite rating is Good to Excellent for this capability, the relative weakness impacts customer buying in the modern market. While a desktop deployment is relatively easy to download and use, an enterprise deployment still requires significant IT involvement. Similar feedback is reflected in Gartner's Peer Insights ratings and in client inquiries.

Oracle

Oracle offers a broad range of BI and analytic capabilities, both on-premises and in the Oracle cloud. Oracle Data Visualization (ODV), Oracle's modern BI offering, is available either as part of the Oracle Business Intelligence Cloud Service (OBICS) as a stand-alone cloud service, as a desktop offering, or as an optional component to Oracle Business Intelligence 12c (deployed on-premises). ODV (the focus of this Critical Capabilities assessment) offers integrated data preparation, data discovery (with advanced exploration) and interactive dashboards via a single design tool supporting both desktop and web-based authoring.

ODV is particularly attractive to organizations that have deployed Oracle applications and Oracle information management technology. Smart connectors inherit Oracle security, there are an extensive set of content packs for Oracle enterprise applications, and the platform offers semantic layer access to Oracle 12c and Oracle BI SaaS.

While Oracle was late to respond to the shift in the market toward modern BI and analytics, its modern BI and analytics components continue to gain traction and are now starting to appeal to the market — particularly within its own installed base. Oracle is also investing early in machine-learning-enabled smart data discovery, including automated pattern detection and integrated search/natural-language processing.

ODV is used for decentralized analytics (45% of customers surveyed) as well as agile centralized BI provisioning (44%). ODV is updated in four major releases a year.

Strengths

  • Packaged content for Oracle applications: ODV appeals to IT departments that have already implemented Oracle's traditional BI platform capabilities, and to lines of business that have deployed Oracle BI SaaS operational reporting on top of Oracle enterprise applications. Oracle packaged, domain-specific content packs include connectors, dashboards and KPIs for finance, HR, supply chain management and CRM. Users are also able to conduct "what if" and scenario analysis within OBICS or ODV Cloud Service via Oracle's Essbase Service.

  • Global and hybrid cloud offerings: Oracle BI can be deployed on-premises or in its global cloud, with the ability to directly query on-premises data from the cloud or migrate and extend on-premises data models and content to the cloud (and vice versa) using a common interface for content development across on-premises and cloud. Oracle's support for hybrid cloud deployments and data gives its on-premises BI customers a glide path to transition to the cloud. Oracle BI (including ODV) leverages the Oracle BI server's historical strengths in function shipping queries to the underlying database to support and optimize querying of data left in place.

  • Embedded advanced analytics and visual appeal: In addition to offering core visual exploration features for light interactive analysis, ODV supports advanced exploration and data manipulation with the ability to create custom groups while visualizing (as well as in the data preparation layer) drag-and-drop advanced analytic functions (such as forecasting, clustering, trending and outliers) via an easy-to-install plug-in. The ability to bin measures is also available as a function. An extensive set of statistical functions is natively available in the platform and through R integration, although Python is not yet supported. ODV's use of motion as part of the interactivity experience adds to the platform's modern visual appeal and contributes to its above-average scores.

Areas of Improvement

  • Self-service data preparation and data connectivity: ODV offers integrated self-service data preparation for harmonizing a range of relational and big data sources but lacks support for JSON and XML. Unlike competing products — where this basic data cleansing is done automatically — when ingesting an Excel or .CVS file, data must be hyper clean free from blanks, spaces, nulls and N/As. In the data preparation interface, analysts can create groups, build calculations, add filters and there are a number transforms available, but users cannot create custom hierarchies. Creating mashups from multiple sources is possible, but user-generated objects are not reusable — only the individual datasets are available to others. In addition, multiple tables can't be joined in one data source connection without writing SQL statements. Regarding inference, data types are often incorrectly inferred and must be manually adjusted. Join and hierarchy inferences — such as those for date and geography — are also not yet supported. While data lineage and impact analysis are strong in OBICS and 12C, these capabilities are still limited in ODV. Finally, packaged content including native connectors for cloud data sources is limited to Oracle (CRM, ERP, SCM, HCM, including acquisitions — NetSuite, Taleo, RightNow and Eloqua) and Salesforce applications, which is a gap when compared to most other cloud BI vendors.

  • Mobile, collaboration and alerts: Native mobile apps for iOS and Android are supported in OBICS and 12C, but with a different rendering experience and less interactivity for ODV content versus Answers and Publisher content; features such as disconnected analysis from a mobile device are not supported for ODV content. Similarly, alerting is not available in ODV — although basic scheduling is offered. Collaboration features such as discussion threads for discussing findings with others within a view or dashboard, and integration of discussion threads within social platforms, are also not supported.

  • Gaps remain for advanced interactivity and smart data discovery: There is no way to define a custom hierarchy in ODV — although drilling through a hierarchy created in the Oracle BI semantic layer and accessed as a data source in ODV is supported. Other interactive features such as display as a percentage must be defined as a calculation, and the ability to create parameters is not yet supported. Regarding geospatial capabilities, ODV does not yet support auto geocoding or out-of-the-box location and distance calculations. With respect to smart data discovery, ODV offers integrated search-based natural-language query as a way to generate views, but the ability to automatically generate insights once a variable is selected is minimal in the data preparation interface. More extensive features are on the roadmap.

Pentaho

The Pentaho Business Analytics platform offers a range of broad functionality across data preparation, self-service and advanced analytics, with a particular focus on big data access and integration. Mature data access and data transformation capabilities are provided by Pentaho Data Integration (PDI) and advanced analytic capabilities by its Data Science Pack. Pentaho is a Hitachi Group company. Lumada, Hitachi's IoT platform introduced in 2Q16, includes capabilities from Pentaho.

In November 2016, Pentaho released version 7.0, which combined its data integration and business analytics servers to support integrated analytic workflow, added visual data access and preparation, and enhanced its capabilities for accessing diverse data sources with improved governance to solve large-scale and complex analytic workflows.

Pentaho has one minor and one major release per year. The focus of this evaluation is on versions 6.1 and 7.0.

The reference customer group surveyed revealed that Pentaho is most often deployed for the OEM or embedded BI use case (45%). Customers also reported using Pentaho for agile centralized BI provisioning and decentralized analytics in more than a quarter of cases.

Strengths

  • Data to analytic scope: Pentaho rated Excellent to Outstanding for both embedded advanced analytics and for its ability to embed analytic content. The machine learning capabilities offered by its Data Science Pack offer a rich set of options for the development of BI applications with embedded analytic capabilities. Pentaho provides commercial licenses for Weka — an open-source, machine-learning framework with hundreds of analytic functions — and the ability to embed Weka, R, Python and PMML, into orchestrated processes. When used in concert, Pentaho's range of data handling and advanced analytics capabilities deliver a more-integrated experience than offered by the use of separate data preparation and data science tools. Version 7 extended this strength to governed data and analytic workflows, with visually enabled data access and preparation supporting developer-to-business, closed-loop data and analytic usage.

  • Data integration: PDI offers a modern user interface and is the key platform component offering access to data sources, data transformation capabilities, embedding of analytic models and BI outputs (such as reports). Pentaho received among the top scores for the data source connectivity and self-contained ETL and data storage critical capabilities, with an Excellent rating in both. Users can analyze a broad range of data sources that can be blended and transformed in PDI and analyzed in Pentaho Analyzer. Datasets can also be written back to a data repository for analysis. Unique capabilities in PDI include the possibility of pushing native queries to NoSQL databases, and an extensive array of data transformation functions using menus and formulas.

  • Scalable platform: Pentaho's open-source heritage, broad developer community and embedding expertise mean the platform has tested, mature, corporate-grade core capabilities. As such, it placed among the top scores for the admin, security and architecture capability. Scalability features — such as node clustering and load balancing for high availability — make Pentaho an excellent fit for large user deployments. Pentaho placed in the top quartile for user deployment size, with an average deployment of almost 4,000 users, well above the 1,182 average deployment size of surveyed customers.

Cautions

  • Ease of use: Customer reference survey data reports ongoing issues with ease of use. Pentaho customers ranked it last overall for ease of use in administration and implementation, content development and visual appeal, and in the bottom quartile for end-user content consumption ease of use. When used as an enabler for OEM or embedded BI, ease-of-use limitations may be less of an issue; however, they are more important for other use cases that are driving much of the new buying activity.

  • Data discovery: Pentaho's functional capabilities remain less-suited for decentralized analytics and governed data discovery use cases (only 5% of the customers surveyed used Pentaho for this type of usage). Pentaho rates as Good for interactive visual exploration and analytic dashboards, and is rated as Fair in the publish, share and collaborate capability. To be specific: in interactive visual exploration, it lacks custom groupings and bins, and natural-language search are not supported; in publish, share, and collaborate, it misses out-of-the box functionality for data storytelling, discussion threads, integration with social platforms, real-time collaboration via shared sessions, timelines and content rating and recommendations. These capabilities drive user engagement and adoption.

  • Emerging functional areas: Pentaho scored as Fair in the capabilities that Gartner sees as the emerging buying drivers for modern BI. In the cloud space, Pentaho does not provide its own cloud hosting service, instead adopting a strategy of "bring your own license." Customers can deploy the platform in AWS, Microsoft Azure or other cloud providers, which is then managed in the same way as an on-premises solution, but running on a remote (cloud) server in a virtual machine. For the smart data discovery area, it does not offer the capability to automatically generate advanced analytic visualizations or models, or to operationalize auto-generated models.

Pyramid Analytics

Pyramid Analytics offers a modern BI and analytics platform with a broad and balanced range of analytics capabilities, including ad hoc analysis, interactive visualization, analytic dashboards, mobile, collaboration, automated distribution and alerts. The solution is well-suited to governed data discovery through features such as BI content watermarking, reusability and sharing of datasets, metadata management, and data lineage.

Continuing a long-term partnership, Pyramid Analytics remains highly integrated with Microsoft's BI offerings. The platform offers an enterprise analytics front end to Microsoft SQL Server Analysis Services (SSAS), while Microsoft Power BI can publish to the Pyramid BI Office server to deliver Power BI content on-premises.

Pyramid's customers report deployments across a diverse range of use cases. Agile centralized BI provisioning and decentralized analytics are both mentioned by 56% of its customers, and 55% use the platform for traditional IT-centric reporting, while 51% reference governed data discovery.

In March 2016, Pyramid Analytics released major version 6, featuring additional connectors to new data sources, improved data preparation features, user experience enhancements on data discovery, and evolution of storytelling, mobile and publication capabilities. Since then, the product has evolved to version 6.33, with new functionality in several areas and problem-correction updates. A good example of functionality released throughout the year is the certified SAP connector, opening access to SAP Hana and SAP Business Warehouse content. Product release cadence stands at one major release per year, three minor versions and bug fixes as required.

Throughout 2017, Pyramid Analytics is focused on rearchitecting its platform to become partner agnostic. This may prohibit innovation in leading-edge capabilities such as smart data discovery, natural-language processing, native big data querying or higher integration with advanced analytics solutions.

Strengths

  • Integration with Microsoft: Pyramid Analytics offers tight integration with the Microsoft BI stack (including Microsoft Analysis Services) and the more-recent Power BI offering. BI Office has the highest percentage of deployments on top of Microsoft-based enterprise data warehouses, at 74% of its customer references. This is even higher than Microsoft's own result (47%). It is also one of the top platforms used with Microsoft's ERP and CRM solutions. Pyramid does offer a tight and extensive integration with the Microsoft environment and should therefore be assessed when that is a top requirement, although the stated roadmap is to make the platform agnostic and available to any data warehouse.

  • Broad range of integrated capabilities: Pyramid received among the highest scores for platform and integrated workflow capabilities, rated Outstanding, derived from a broad set of features delivered under a single product and unified user experience. The user experience follows look and feel of Microsoft Office, contributing to an easier learning curve.

  • Rapid content development: Content development using Pyramid Analytics is quicker than on most other tools covered in this report. Regardless of complexity — simple, moderate and complex content — the platform delivers results in less time than most of its competitors. Excellent ratings on critical capabilities such as data source connectivity, interactive visual exploration, and publish, share and collaborate explain these results.

Areas of Improvement

  • Dependence on the Microsoft stack: What was already described as one of the tool's strengths — a tight integration with Microsoft products — can also be seen as an area for improvement. Pyramid's dependence from Microsoft is strong and organizations leveraging databases and BI components from other vendors will find it difficult to justify using the platform. Being aware of that, and knowing that Microsoft will soon offer the ability to publish Power BI content on-premises without needing Pyramid Analytics, the product roadmap is clearly headed at making the platform more agnostic, to increase its appeal to non-Microsoft shops.

  • Modernization required to improve ease of use and visual appeal: Pyramid's customers rate the product's ease of use below average relative to other products in this research, although it is rated close to Excellent (3.8) in aggregate terms. Ease of use, for business consumers, was the lowest rated in our survey, and this group is arguably the most-important segment for users. Visual appeal also ranked in the bottom quartile.

  • Limited cloud offering: Although offering a cloud BI option that includes hybrid connectivity to on-premises and cloud-based data, Pyramid Analytics has gaps where it lags behind competitors. The solution has limited packaged content and does not offer a marketplace with partner's or Pyramid's add-ons to BI Office. Also, the offering is mainly a Microsoft Azure marketplace option. There is no optimized version for Amazon AWS — the leading cloud provider — although customers could decide to operate on a bring-your-own-license model to deploy it there.

Qlik

Qlik offers governed data discovery and analytics either as a stand-alone application or (increasingly) embedded in other applications. Qlik Sense is the vendor's lead product and is sold to most new customers, while QlikView continues to be enhanced and makes up a larger portion of the company's installed customer base. The Qlik Analytics Platform is the product that developers can use for embedded BI and is the platform on which Qlik Sense has been developed.

The in-memory engine and associative analytics allow customers to build robust, interactive applications and to visualize patterns in data in ways that are not readily achievable with straight SQL. NPrinting, which provides report scheduling and distribution, was added to Qlik Sense in 2016 (previously only available for QlikView), enabling Qlik to provide interactive visual discovery and also Mode 1 BI in an agile way.

In the last year, Qlik has improved the data preparation process, making it more visual, and launched Qlik Sense Cloud Business for up to 50 users in the cloud, with an enterprise edition planned for 2017. The support of NPrinting to Qlik Sense (previously only supported in QlikView) was an important enhancement in 2016, giving Qlik the ability to provide scheduled, formatted report distribution as well as visual exploration in a cohesive product. In this way, Qlik supports both Mode 1 and Mode 2 style of analytics. Qlik Sense 3.1 (the focus of this evaluation) was released in 3Q16. QlikView continues to be supported and enhanced but, as it is primarily offered to existing (rather than net new) customers, it is not assessed in this note.

Qlik Sense is used primarily for agile centralized BI provisioning, with 64% of customers deploying this way, followed by decentralized analytics (48%).

Strengths

  • Robust applications: Qlik Indexing Engine (QIX) allows customers to use Qlik Sense as a pseudo data mart that supports multiple data sources, complex calculations and robust applications. There is an in-database option, referred to as direct discovery, for customers who have invested in an analytic database, but this is less widely used. The in-memory, associative engine supports complex data models such as multiple fact tables. It also provides data scalability and in-memory compression, with Qlik ranking in the top third of Critical Capabilities vendors for data volumes from HDFS.

  • Ease of use and visual appeal: Ease of use is a key buying requirement in the modern BI and analytics market that includes the ease of implementation, but also ease of building content. While both QlikView and Qlik Sense both support rapid implementation, the ease of use and visual appeal is markedly better in Qlik Sense. Qlik Sense rates Excellent for this critical capability. The subcriterion for visual appeal ranks in the top quartile. Qlik's "smart search" contributes to the ease and power of the application; a user can enter a search term and Qlik Sense will automatically present a list of dimensions and measures to filter the current dashboard by these keywords. Values with no association to the currently displayed dataset are grayed out; in this way, a user can see, for example, which products are no t selling in a particular region.

  • Embedded BI: Open APIs were a key design tenet of Qlik Sense and for the embed analytic content capability, the product rates Excellent to Outstanding. Partners can "white label" the product in an OEM arrangement, or can embed it in custom applications. Partner Host Analytics in the CPM market uses Qlik as its dashboard and visual exploration front end. In addition, Qlik was one of the first to showcase natural-language generation with partner Narrative Science. Here, the open APIs allowed Narrative Science to develop an extension that customers can easily install. The Narrative Science extension then appears automatically in the dashboard design interface to allow an author to add a text box to the page. A textual explanation of a chart automatically writes itself as a user filters within the chart. This is just one example of many extensions that the partner network and customers themselves have developed.

Areas of Improvement

  • No predictive: Qlik scores only Fair for its embedded advanced analytics. Increasingly, business users expect menu-driven ability to do simple forecasting and clustering within a BI tool, neither of which are supported in Qlik Sense. While there are some statistical functions as part of the Qlik Sense function library, there is no ability to call out to an R script. This is on the product roadmap. Likewise, advanced chart types such as decision trees are not natively supported but could be added via an extension. Smart data discovery, in which users want insights generated automatically from the software, is not supported.

  • Mobile limitations: Qlik Sense supports the consumption of content on mobile devices via HTML5. This approach supports authoring as well as optimal rendering of content across different device types via responsive design. However, Qlik Sense does not support native apps, a point of difference from QlikView and key competitors. The lack of a native app approach means that capabilities such as offline, push notifications and location awareness, and intuitive native iOS selectors and charts are not supported. These are roadmap items.

  • Cloud and other gaps: Qlik launched its cloud solution in 2015, initially positioned for individuals. Qlik Sense Cloud Business (for small to midsize organizations and up to 50 users) was released end of January 2017. The current offering does not support hybrid connectivity to on-premises data sources and lacks a number of administrative and security features that enterprises require. An enterprise release is planned. Prebuilt content for cloud-data sources is a subcriterion that Qlik also lacks. While Qlik scores well at the capability level for visual exploration and publishing, it lacks specific features in these categories. For example, there is no out-of-the-box support for trellis charts, display as percentages, or binning. Collaboration in the form of discussion threads is also not supported. While ease of use is excellent overall, the product still lacks a point-and-click interface for building expressions, although a type-ahead feature was introduced in the last year.

Salesforce

Salesforce is in the process of integrating its Wave Analytics offering with its recent acquisition (September 2016) of BeyondCore. Wave is a platform for creating point-and-click interactive visualizations, dashboards and analysis with integrated self-service data preparation. BeyondCore, a Visionary on last year's Magic Quadrant, is a market disruptor that uses machine learning under the covers to automatically find, visualize and narrate important findings in data, without requiring users to build models or write algorithms. Salesforce Wave is sold as a stand-alone platform and also as the foundation of packaged, closed-loop, front-office analytic applications for sales, marketing and service. The platform is natively mobile and offers collaboration through integration with Salesforce Chatter. BeyondCore, now rebranded as Salesforce Analytics Cloud Smart Data Discovery, will also be sold stand-alone and as an optional component to Wave, Wave apps and to Salesforce-branded Einstein applications. Ultimately, the plan is for BeyondCore's automated insight and narrative generation to be a seamless part of the Einstein-enabled Wave platform, applications and experience. Given that this work is already partially implemented, the capabilities are considered jointly rather than as separate product evaluations.

Salesforce continues to primarily cater to its installed base.

Salesforce Wave has three major releases per year in October, February and June. Wave is used primarily for the decentralized analytics (37%) and OEM or embedded BI use cases (37%). BeyondCore is used primarily for governed data discovery (33%) and decentralized analytics (33%).

Strengths

  • Well positioned for the next BI and analytics disruption: Embedded advanced analytics and smart data discovery are clear strengths for Salesforce receiving among the top ratings for these critical capabilities. Salesforce has also earned among the highest scores for ease of use, and the highest for visual appeal from its reference customers. BeyondCore is used to automatically evaluate every data combination to identify meaningful patterns and areas of potential interest that warrant further exploration. BeyondCore automatically creates a deterministic piece-wise regression model that finds complex relationships that are difficult to express with traditional regression models. For example, affinity analysis is used in diagnostic graphs and in the narrative text of descriptive and diagnostic graphs. Time series data is automatically identified, and bottom-up forecast analysis is conducted. BeyondCore conducts statistical tests on its findings and insights to ensure that they are statistically significant and actionable before generating graphs and corresponding narratives. It then automatically generates a narrative (textual as well as voice) explaining the key insights in each graph and relationships between graphs. The underlying R code for diagnostic and predictive analysis can be exported as an R model for data scientists to validate and extend the models as needed. Initial integration between Wave and BeyondCore allows Wave users to access datasets using BeyondCore connectors (not currently available in Wave) and smart data preparation features. Users can also generate a Wave app from insights automatically generated by BeyondCore.

  • Optimized for Salesforce: Wave packaged applications for sales, marketing and service are a key differentiator and a major reason why customers buy the platform. Salesforce business consumers gain integrated, contextualized insights from within the Salesforce Application workflow (particularly when using the packaged Wave-based analytics apps). Wave is natively integrated with Salesforce security, collaboration and metadata, including simplified access to Salesforce application tables through an intuitive wizard. Users can invoke Salesforce actions from within Wave (such as data quality, new campaigns and targeted outreach) and can collaborate using Chatter. Due to a focus on Salesforce optimizations and customer-facing Wave-enabled apps, Wave continues to appeal primarily to the Salesforce installed base who have most of their data in the cloud, particularly in Salesforce, and want to augment it with on-premises data.

  • Extensibility and embeddability: Salesforce Analytics Wave has a broad partner ecosystem that includes many ETL, predictive analytics vendors, and system integrators. Wave exposes services via REST-based APIs, which can be consumed by other distributed services and used to create and extend new applications based on the Wave platform. Its developer marketplace, AppExchange, provides a platform for independent software vendors and developers to build and sell custom content (including datasets, lenses, metadata and applications). It also provides a market for developer skills, making them widely available despite Wave's recent entry into the market. Salesforce also offers OEM-specific packaging and licensing for Wave.

Areas of Improvement

  • Advanced interactive data exploration and manipulation and analytic dashboards: Over the past year, Salesforce has added a number of good interactive features including quick calcs for common types of analysis (display as a percentage, for example) and good visual linking, and both Wave and BeyondCore automatically generate best-practice visualizations including colors and sorts, which is a positive step. However, advanced interactive exploration — such as grouping, binning in the visual analysis and dashboard environment, and extensive geospatial capabilities — is still limited. Users can interact with maps by country, region, state and city, but automatic geocoding is only available for Salesforce data, and there are no out-of-the-box geospatial algorithms. Currently, this is only possible with Salesforce SOQL Geolocate API. Some chart types are missing out of the box (trellis, for example), with some on the 2017 roadmap or possible via extensions.

  • Self-service data preparation: Over the past year, Salesforce added a new self-service data preparation interface — Dataset Designer — to Wave, which provides basic capabilities for accessing and manipulating data for analysis. Data lineage is available via the Wave bulk API, including derived fields and the new data recipe feature, which self-records transformations to the data, and is exposed in an information panel in each widget on a dashboard. While improved, there are limited features for data profiling and manipulation, building calculations, groups, bins, hierarchies and full join operations are not yet supported. BeyondCore can automatically discover data types such as date, text, and measures, even on datasets with data quality issues. During the "lookup data" (join) process, BeyondCore recommends the join key if it can detect a match. This can be leveraged in Wave.

  • Data source connectivity: Data is loaded into Wave's proprietary data store for analysis — hybrid data connectivity and direct query of on-premises data sources left in place are not yet supported in Wave. It will be supported in Wave via the integration of BeyondCore, which provides this capability. Connection to non-Salesforce enterprise application and other data currently requires third-party partner tools. Native Wave connectors to a range of other enterprise applications sources will be delivered through an OEM relationship. Informatica connectors are planned for 2017. Regarding personal data sources, connectors are limited to .CSV and Microsoft Excel files; XML, JSON or RSS feeds are not yet supported. BeyondCore can be used to access most major databases, as well as big data sources through Hive and partner sources such as SAP Hana.

SAP (BusinessObjects Cloud)

SAP delivers a broad range of BI and analytic capabilities for both, large IT-managed enterprise reporting deployments and business-user-driven, data discovery deployments. Companies often choose SAP as their enterprise BI standard, especially if they also standardize on SAP applications.

SAP BusinessObjects Cloud, introduced in October 2015 (formerly called SAP Cloud for Analytics), is a purely cloud-based deployment, built on SAP's Hana cloud platform. The focus of this evaluation is on version 2016.25.

SAP BusinessObjects Cloud combines data discovery, predictive analytics and planning in an integrated, cloud-based product running on the SAP Hana Cloud Platform, SAP's platform-as-a-service offering. SAP's new Digital Boardroom solution is built on the SAP BusinessObjects Cloud platform. The Digital Boardroom includes stories from SAP BusinessObjects Cloud and additional capabilities, most notably the ability to display stories on a three-panel, wall-size display that is touch-enabled. In addition, there are unique capabilities such as an agenda builder, value driver tree, and simulation or what-if analysis. As pure cloud products, all data modeling, administration, and authoring of content are done via a browser. The introduction of guided machine discovery represents a remarkable improvement toward smart data discovery as the next wave of disruption in the BI and analytics platform market.

The predominant use case for SAP BusinessObjects Cloud is decentralized analytics (73%), followed by agile centralized BI provisioning (46%). SAP BusinessObjects Cloud is currently updated on a biweekly cycle.

Strengths

  • Modern BI platform with smart data discovery: Guided machine discovery is a capability on the cloud platform, enabling business users to gain insights automatically, such as key influencers, classification and regressions. The insights are visualized with automatically selected graphs and a narrative to explain the most important drivers of a particular metric. A further option exists to integrate the SAP Hana text analytics and natural-language services.

  • Part of a comprehensive cloud platform: The cloud platform also provides planning and predictive analytics components (at extra license cost), and is the core platform for the Digital Boardroom solution. SAP offers a unified platform for analytics, predictive and planning capabilities, which has so far been a niche segment. The platform leverages Hana capabilities, as it is built on the Hana Cloud Platform (HCP). Live access for in-database queries is available for on-premises SAP Hana and SAP Hana Cloud Platform, with hybrid connectivity to Business Warehouse and S/4HANA planned for 1Q17. Incremental data loading is available for other SAP and cloud data sources, such as SAP BPC, SAP BusinessObjects Universes, Google Sheets, SAP SuccessFactors, SAP ECC and Salesforce.

  • Ease of use and data preparation: SAP BusinessObjects Cloud achieved a Good to Excellent score for the self-service data preparation critical capability. Several subcriteria in this critical capability are fully supported, such as enabling business user joins, data mashup, data modelling and data enrichment. It provides a modern, visually appealing user interface, and scores in the top third for all products. Survey respondents rated the platform in the top quartile for ease of use regarding administration and content creation; and in the top third for content consumption, with an overall Excellent rating for this critical capability.

Areas of Improvement

  • Modern and new, but less mature: One out of five survey respondents indicated absent or weak functionality as the primary platform problem for SAP BusinessObjects Cloud. For the critical capabilities interactive visual exploration as well as analytic dashboards, this product is less mature compared to Lumira, and other modern BI platforms. Several subcriteria are not fully supported, such as information visualization, binning, advanced chart types or chart formatting options, animation and playback, or disconnected exploration.

  • Mobile: Mobile exploration and authoring is currently the weakest critical capability for SAP BusinessObjects Cloud. Native support for iOS and Android mobile devices are roadmap items, unlike SAP BusinessObjects Lumira, which already offers full support here, including offline consumption. The current iOS app can only be used for collaboration with colleagues, viewing events and tasks, and notifications, but not to explore or even author analytics. Responsive design, a critical feature to adapt to different screen sizes for generic mobile device usability, is a roadmap item.

  • Embedding: Embedding analytical content is rather weak, with only Poor to Fair scores for SAP BusinessObjects Cloud in this critical capability. Software development kits for printing, parametrization, creation, deletion, copying, portal integration, administration, among others are currently not available. Several of the mentioned features are on SAP's roadmap and planned for near-term updates. Given SAP's frequent update cycle, clients wanting to embed analytics should check regularly for these updates.

SAP (BusinessObjects Lumira)

SAP delivers a broad range of BI and analytic capabilities for both, large IT-managed enterprise reporting deployments and business-user-driven, data discovery deployments. Companies often choose SAP as their enterprise BI standard, especially if they also standardize on SAP applications. SAP BusinessObjects Enterprise and Lumira are primarily for on-premises deployments.

Similar to last year, several of SAP's BI and analytic components are described in the "Market Guide for Enterprise-Reporting-Based Platforms" such as SAP BusinessObjects Design Studio, SAP BusinessObjects Dashboards, SAP Crystal Reports, SAP BusinessObjects Web Intelligence and SAP BusinessObjects Analysis for Office.

The following components were considered for the Magic Quadrant and this Critical Capabilities:

  • SAP BusinessObjects Enterprise with the following components:

    • SAP BusinessObjects Business Intelligence platform (current version is 4.2).

    • SAP BusinessObjects Lumira, server for BI platform (current version is 1.31).

    • SAP BusinessObjects Lumira desktop (current version is 1.31).

    • Lumira Edge, a small-to-medium size enterprise (SME) offering (current version is 1.31).

SAP BusinessObjects Lumira currently has a release cycle of one major release every 18 months and one minor release every quarter. As of 3Q15, SAP discontinued future updates to Lumira server on Hana, which is replaced by Lumira server for BI platform, and Lumira Cloud, which is replaced by SAP BusinessObjects Cloud. End of maintenance for Lumira server on Hana and Lumira Cloud was 30 September 2016.

The most-significant effort with respect to SAP BusinessObjects Lumira is certainly the announced merger of the two products, SAP BusinessObjects Lumira and SAP BusinessObjects Design Studio, while SAP continued to improve robustness and maturity of SAP BusinessObjects Lumira. Advances in SAP BusinessObjects Lumira were made with web-based authoring of SAP Hana-based stories and support for offline mobile consumption.

For SAP BusinessObjects Lumira, survey responses indicate an almost equal distribution across three use cases. It is equally often used for agile centralized BI provisioning (63), decentralized analytics (60%) and governed data discovery (58%).

Strengths

  • Information portal and analytics workbench scenarios: SAP BusinessObjects Lumira's product strength across the critical capabilities lies in the BI platform administration, and its interactive visualization and analytic content creation capabilities. The former is based on features of the mature SAP BusinessObjects BI platform, with integration to the metadata layer (universe). SAP has continued to enhance the product's interactive visual exploration capabilities to include the most important chart types, custom groups and hierarchies, applying filters, drilling down and up, binning, and sorting.

  • Visual exploration and further simplification: Early versions of SAP BusinessObjects Lumira lacked some of the core chart types and data manipulation features that have continued to improve with each release. The product now supports the most-popular chart types out of the box, including grouping, binning and display as percentages. Version 2 will merge the two components of SAP BusinessObjects Lumira and SAP BusinessObjects Design Studio into one product with two authoring interfaces positioned for business user authors, and IT developers, respectively. It is planned to start ramp up in April 2017 and a general availability release is planned in late 2Q17 or early 3Q17. This could further strengthen the interoperability between centrally provisioned and decentrally developed analytics, and minimize confusion over which authoring tool to use for analytic dashboards.

  • Self-service data preparation and mapping: SAP BusinessObjects Lumira was one of the first data discovery products to include menu-driven self-service data preparation. It achieved a Good score for our self-service data preparation capability, with full support for subcriteria, such as enabling business user joins, data mashup, data modelling and data enrichment. Basic geospatial capabilities are well-supported, leveraging the Esri partnership and Navteq geo database, geocoding and reverse geocoding is available, based on latitude/longitude data. Clients can build geographic hierarchies, and drawing point, bubble, pie and chloropleth visualizations are supported.

Areas of Improvement

  • Smart and advanced analytics: SAP BusinessObjects Lumira offers less features in the critical capabilities of embedded advanced analytics and smart data discovery (which are stronger in SAP BusinessObjects Cloud). It only achieved Poor to Fair scores for embedded advanced analytics, and Fair scores for smart data discovery. For embedded advanced analytics, some features are not available, such as advanced analytics visualizations or advanced predictive analysis. Menu-driven options for influence analysis and forecasting are supported. However, clustering is not natively supported, nor is integration with third-party statistical languages such as R, with only limited built-in statistical functions. For smart data discovery, the guided machine discovery features are not available, nor is natural-language Q&A or natural-language generation available as a feature.

  • Limitations in platform integration and workflow: Despite the common name, the interoperability between SAP BusinessObjects Enterprise and SAP BusinessObjects Cloud is limited. While there is some interoperability on data source and data access level, content created by one platform cannot be promoted to or reused by the other. Clients should regard both as distinct platforms, even requiring different licenses. Further, even within the SAP BusinessObjects Enterprise platform, Lumira does not inherit all of the platform capabilities. For example, scheduling and alerting are not supported.

  • Product maturity still work in progress: Clients continue to raise concerns about this product's maturity. One out of three clients using SAP BusinessObjects Lumira identified absent or weak functionality as the two most-significant platform problems. Software quality was the most significant limitation for one out of five clients in the survey. Lumira achieved only a Fair score in the publishing, sharing and collaboration critical capability, and only a Fair to Good score in data source connectivity.

SAS

SAS offers a range of BI and analytic capabilities with SAS Visual Analytics providing interactive discovery, embedded advanced analytics, and dashboards for mainstream business users as well as data scientists.

SAS Office Analytics includes SAS Enterprise Guide and Microsoft Office integration with Excel, PowerPoint, Word and Outlook. SAS Office Analytics allows Visual Analytics content to be dynamically refreshed and interacted with via PowerPoint and other Office tools. SAS Enterprise Guide is a desktop product that allows power users to perform self-service data preparation and advanced analytics that can then be published to SAS Visual Analytics server.

SAS Visual Analytics can be deployed on-premises or, via the cloud, although only a small percentage of surveyed customers have deployed in the cloud. According to SAS, Visual Analytics is deployed by over 13,000 organizations. As a server-based product, Visual Analytics is most often used for decentralized analytics (64% of customers surveyed), with 40% using it for agile centralized BI provisioning.

SAS Visual Analytics runs on the SAS LASR Analytic Server, an in-memory offering that uses Hadoop for its persistence layer, and that has the ability to handle large datasets.

SAS Visual Analytics 7.3 was released in August 2015 with no new releases in 2016 as SAS works to rearchitect the product to run on the new SAS Viya architecture. SAS Viya is an open, cloud-ready, microservice-based architecture. The user interface has been completely redesigned. Existing SAS Visual Analytics customers will be able to upgrade to version 7.4 on their current SAS 9.4 platform or eventually upgrade to SAS Visual Analytics running on Viya. The first release of SAS Visual Analytics to run on Viya will be version 8.1, due in early 2017. As this product is not yet released, it is not within the scope of this product evaluation.

Strengths

  • Embedded advanced analytics: With the company's origins in advanced analytics, it is not surprising that SAS rates between Good and Excellent for its embedded advanced analytics capabilities. SAS supports multiple forecasting models via a point-and-click interface. Clustering and decision trees are also differentiators, as well as a number of advanced visualizations including Sankey diagram, network analysis and correlation matrix. Although SAS has high scores for this capability, the automated generation of analytics and natural-language generation are lacking. These are roadmap items.

  • Interactive visual exploration: SAS Visual Analytics Explorer interface rates Excellent to Outstanding for interactive visual exploration. As users drag and drop elements onto a canvas, Visual Analytics automatically renders the data using the optimal chart display and color settings. Users can choose to redisplay measures as percentages, time period variances or period growth, all via menu options, and without needing to create specific expressions. Binning for time periods is supported, but not for numeric values. Users can create custom categories for new groups of dimension values.

  • Broad BI platform capabilities: SAS competed in the traditional BI platform market with its Enterprise BI Server product and introduced SAS Visual Analytics in 2012. Often, newly introduced products have not inherited the full platform capabilities. Although SAS Visual Analytics was released as a new product, it supports the full platform capabilities of enterprise-grade administration, scheduling and publishing, and support for mobile.

Areas of Improvement

  • Less easy to use, administer and deploy: SAS Visual Analytics scores in the bottom quartile for ease of use and visual appeal, but still rating Good overall. Of customers surveyed, 17% cited difficulty in deploying the platform. In authoring, ease of use suffers from SAS Visual Analytics having multiple design interfaces for exploration and formatted reports. There is interoperability between the two, but the workflow is not as seamless as in competitive products.

  • SDK and APIs: SAS Visual Analytics has Fair scores for embedded analytic content, and only 11% of customers are using it for the OEM or embedded BI use case. There are no separate SDKs. Individual charts and dashboards cannot be published directly to third-party portals. SAS does support integration of Visual Analytics reports (a particular content type) as a Microsoft SharePoint Web Part. The SAS Visual Analytics viewer can be embedded in an iframe. Many of these limitations should be addressed via the new open architecture.

  • Self-service data preparation not cohesive product/workflow: Self-service data preparation is an important characteristic of a modern BI and analytics platform as users must be able to access and manipulate data from new data sources that may not have been curated by a data warehouse team. SAS scored between Good and Excellent for this capability as it is possible to prepare data via a number of options including the SAS Enterprise Guide and the browser-based Data Builder. However, there is room for improvement in its smart preparation capabilities, data profiling and a streamlined workflow. Chart types are not consistent between the visualize and report interfaces. These are all areas of improvement planned for SAS Visual Analytics 8.1 on Viya.

Sisense

Sisense offers a single platform with a self-contained, in-chip, columnar database engine that allows for visual exploration of web-based dashboards. Data modeling and preparation is via a desktop interface, while dashboards are authored via a browser-based interface. Security and administration is also browser-based.

The vendor's in-chip approach allows for data scalability and high performance. It has introduced a number of innovative capabilities in the last year, the most noteworthy being the voice integration with Amazon Alexa and BI bots. Users can pose analytic queries using voice, while Alexa will answer with key metrics and trends. In addition, integration with LIFX lightbulbs is able to light a room or a desktop lamp in particular colors depending on how key metrics are performing.

More fundamental improvements in the product include the ability to leave data in place, making it now more suitable for customers that have already invested in a high-performing analytic database. Scheduling and alerts were also added in 4Q16.

Sisense has a major release every quarter with minor releases monthly. The focus of this evaluation is on version 6.6.

Sisense is most often deployed for the OEM or embedded BI use case (43%), followed by agile centralized BI provisioning (33%).

Strengths

  • In-memory, in-chip and in-database: ElastiCube is Sisense's in-memory and columnar storage engine that supports analysis of complex data models, coming from multiple data sources. While many of the product's in this Critical Capabilities research use in-memory and columnar storage, Sisense further differentiates itself on in-chip processing to provide even faster performance and data scalability. With in-chip processing, large datasets are broken into smaller sets of data and cached at the CPU level, giving more scalability than if all data had to be loaded in-memory only. In addition, the engine supports multiple fact tables, with joins dynamically performed at query time, on the fly, to minimize the upfront data modeling. For customers that have already invested in high-performing databases, as of version 6.5 released in 4Q16, Sisense now also supports in-database processing for several leading platforms including Amazon Redshift and HPE Vertica.

  • Embedded analytics: Sisense rated Excellent for the embed analytic content critical capability. The platform is open with an SDK for a broad range of platform capabilities including mobile, print and security. Dashboards and data in the ElastiCube are accessible via a REST API. Sisense includes its own chart library that can be further extended with access to D3 visualization libraries. As well, Sisense charts can be rendered as JavaScript to embed in any third-party applications, with full support for interactivity such as dashboard filtering. Iframes are also supported. OEM and extranet customers can white label the product and apply their own color schemes and logos.

  • Elements of smart data discovery: Natural-language query and natural-language generation are two aspects of smart data discovery that Sisense newly supports. With the BI bots, users can engage in conversations to ask questions and receive a natural-language answer and recommended action. These bots can be embedded within third-party collaboration tools such as Slack. Sisense also has integration with Narrative Science for generating a text explanation of results. Pulse is a new capability that supports anomaly detection and alert notification using machine learning.

Areas of Improvement

  • Central model: The ability to mash data from multiple data sources is a responsibility of the cube designer, thus limiting the flexibility to individual users and the decentralized analytics use case. From a governance perspective, data lineage in Sisense is a two-step, disconnected process. On the ETL/data integration level, all data sources and how they are accessed can be seen. On the dashboard level, the underlying data model can be seen. Impact analysis of changes is not supported. Watermarking is also not available for datasets, nor is the ability to outline and manage the certification or sanctioning of datasets or content according to the clients' governance policies.

  • Embedded advanced analytics not out of the box: Sisense added support for R in 2015 and uses extensions to provide advanced analytics. Extensions may be built by the vendor or contributed to by the community. For example, for menu-driven forecasting or clustering, the administrator must download the extension to the server to be able to invoke it. A similar approach is used for advanced chart types such as decision trees and network charts that are not native to the product.

  • Cloud limitations: Sisense is offered in the Amazon Marketplace or can be hosted in a single-tenant option that the vendor maintains in Amazon Web Services (AWS). The vendor supports a bring-your-own-license approach for customers to deploy in other cloud platforms such as Microsoft Azure or Rackspace, but does not offer a SaaS option itself. Further, the software has not been certified for the cloud from a security perspective. Access to cloud-based applications is a subcriterion within this capability, and here, Sisense provides a number of prebuilt industry templates with data models and dashboards. However, these templates are not specific to particular business applications (such as Salesforce for CRM).

Tableau

Tableau delivers an easy-to-use, visualization-based analytic workflow experience to business users to access, prepare and analyze their data without IT support through three primary products: Tableau Desktop, Tableau Server and Tableau Online (Tableau's cloud offering). Tableau Desktop is a desktop-based design tool. Tableau Server and Tableau Online are used to share content and scale deployments to a broader range of users. Tableau Server represents the on-premises option. Tableau Online is the cloud-based option, which is hosted in Tableau's data center but is now also running on AWS. Additionally, Tableau Server can be hosted on any of the major public cloud platforms: AWS, Microsoft Azure and Google Cloud Platform. In both deployment models, data model authoring and full formatting options are available within Tableau Desktop and then published to the server or the cloud. From within a browser-based environment, users can interact with published dashboards and create new dashboards and sheets.

Tableau 10, released in August 2016, added data federation capabilities and makes data mashups more reusable. Drag-and-drop advanced analytics for clustering and segmentation are added to existing capabilities for forecasting and trending. Tableau 10 now has improved JavaScript and REST APIs. Also new in Tableau 10 is the ability to author a dashboard and workbook-level format from within the browser in Tableau Server. Tableau is deployed primarily for decentralized analytics (50%), followed by agile centralized BI provisioning (41%).

Strengths

  • Intuitive interactive visual exploration and dashboards: Tableau's core product strengths continue to be its intuitive interactive visualization and exploration and analytic dashboard capabilities for almost any data source, as confirmed by its Excellent rating for these capabilities. Tableau enables rapid advanced exploration and content creation for core users by automating routine tasks, such as geocoding and the creation of time hierarchies (month, quarter or year, for example) on data fields and adding type ahead for formula building. Users can also create and analyze data by custom geographic territories by lassoing or clicking on marks on a map. For advanced analytics, new drag-and-drop clustering in Tableau 10 builds on existing advanced analytics functions for forecasting, and trends that are available from simple menus make insights from advanced analytics available to business users. R and newly added Python integration are also supported for incorporating additional algorithms into Tableau calculations. Tableau's reference customers score its ease of use among the highest of all these vendors.

    • Breadth of accessible data sources: Tableau allows business users to interact with a broad range of data sources by using an extensive set of data connectors with both in-memory and direct query access for larger datasets. Data source connectivity is a strength of Tableau, and an area that is further improved in version 10. It natively supports a broad range of relational databases, Hadoop distributions, NoSQL sources, personal files and statistical package output formats (including IBM SPSS, SAS and R data files). A web data connector provides access to web services via available APIs. New in Tableau 10 is support for Marketo, SAP Hana MemSQL, QuickBooks Online, Oracle Stored Procedures and Google Sheets, to name only a few of the additions.

  • Broad enterprise mobile support with responsive design: Tableau content can be rendered natively with full interactivity on iOS and Android tablets and phones. Moreover, while offline analysis is not yet supported, improvements to Tableau 10's mobile features address enterprise security needs, with capabilities to integrate with third-party mobile device management platforms, such as those of VMware (AirWatch) and MobileIron, while also enhancing responsive design for all mobile screen sizes.

Areas of Improvement

  • Basic data models and multiple fact tables: Each data source is constrained to a single star schema; multiple fact tables within the same physical data source are not supported and must be created elsewhere when needed. Tableau's overall metadata management rating is impacted by limitations in its ability to reuse data-source-specific metadata objects (such as calculated measures, custom groups, hierarchies) across workbooks and underlying data extracts. It is not well-suited to more-difficult data mashups with inconsistent codes in join fields. This also affects Tableau's ability to offer support for data lineage and impact analysis. When preparing data, Data Interpreter will make its best guess at unpivoting cross-tabs and removing empty cells to create a tabular dataset from spreadsheet data, and users can do some data cleansing tasks through formulas as they load data, but extensive, point-and-click or automated data profiling and transformations are not supported. Tableau has announced its plans to release a stand-alone, self-service data preparation tool, codenamed Project Maestro, in 2017 to address its customers' challenges with large and complex data.

  • Gaps in enterprise features and collaboration: Tableau is prioritizing enterprise features to close current gaps — such as adding support for Linux, and improving APIs for better embeddability and extensibility. Event-based scheduling, conditional alerting, printing to PDF and PowerPoint, and collaboration and social platform integration are also works in progress. Collaboration capabilities for threaded discussions are not supported in Tableau 10; there is no integration with third-party collaboration platforms. There is also limited capabilities for users to rate (like) reports, but recommendation of content based on those ratings is on the roadmap.

  • Limited scale and variable performance of Tableau data extracts: When Tableau first came on the market, it primarily connected to data sources live — a key benefit for those customers with an analytic database that want to replicate data in a proprietary, in-memory solution. The Tableau in-memory engine, in the form of Tableau Data Extracts, was added in 2010 as an optional performance enhancement. This part of the product is less-mature and scalable than competing in-memory engines. Poor performance for large in-memory extracts often requires modeling in a separate data repository that is directly queried from Tableau. In March 2016, Tableau acquired HyPer to improve the engine's scalability. HyPer is a high-performance in-memory database developed as a research project at the Technical University of Munich (TUM) that was not launched as a commercial product. Tableau plans to use HyPer to replace the Tableau data extracts in order to support larger datasets.

ThoughtSpot

ThoughtSpot is a visual analytics platform whose main differentiator is its search-based interface, with a high-performance in-memory, columnar database deployed as an appliance. Users explore data via a type-ahead feature that will also recommend the most popular search terms, existing datasets, charts and dashboards. Once a chart is generated, users can assemble them into dashboards (or "pinboards" as the vendor refers to them). ThoughtSpot's software can be deployed on a preconfigured hardware appliance, as an on-premises virtual machine with VMware, or optionally in the cloud via Microsoft Azure or Amazon Web Services.

Although ThoughtSpot's origins are in search-based analytics, the vendor introduced smart data discovery branded as A3 (or Auto Awesome Analytics, if you prefer). With A3, visualizations are generated automatically with a short textual explanation about the most important outliers and trends. The vendor also recently added support for scheduling, discussion threads (first seen in version 4 released in 4Q16), and a native mobile app (version 4.1, 1Q17). ThoughtSpot has a major release annually and minor releases each quarter.

ThoughtSpot is most often deployed for decentralized analytics (61%) followed by agile centralized BI provisioning (52%).

Strengths

  • Ease of use via search: ThoughtSpot uses a search-based interface that allows users to explore and visualize data with an overall rating of Excellent. Machine learning algorithms provide a type-ahead ability to make it easier for business users to find the most relevant search term. Rather than having to drag and drop data elements onto a page, as is the predominant query flow in most BI tools, users simply type in words, such as "sales by state" or "sales by product if color is red." The data modeler can add a range of synonyms for each measure or dimension. After a visualization is automatically generated from the search-based query, a user can pin it to create a dashboard.

  • In-memory engine: The product uses a columnar, in-memory engine that indexes all of the searchable data to ensure fast performance on large datasets. The engine is deployed as an appliance on commodity hardware. Of customers surveyed, 24% are analyzing 1TB or more of data in a single application, ranking ThoughtSpot in the top third for this metric. The in-memory engine supports loading of data from multiple data sources. An OEM arrangement with Informatica as part of ThoughtSpot "Data Connect" module allows data to be transformed and cleansed as it is imported.

  • Scheduling and discussions: As of the version 4 release, users can schedule a dashboard to be distributed via PDF or .CSV. As this capability is lacking in many modern BI products, it provides a useful differentiator. However, the product does not yet support business events alerts, although this is a roadmap item. In addition, support for discussions was recently added within the dashboard or optionally via integration with Slack, a third-party collaboration platform.

Areas of Improvement

  • Lack of core exploration and advanced analytics: ThoughtSpot presents data in the most popular chart formats but lacks many of the advanced charts and data manipulation capabilities that power users and citizen data scientists want. For example, trellis charts and candlestick charts are not supported; geographic mapping is limited, without automatic geocoding or distance calculations. Features such as visual grouping, display as percentages, and binning would have to be created via a formula rather than a menu or visual point-and-click approach. Drill down is essentially a drill anywhere option as hierarchies are not supported. There are no advanced analytics charts, nor is there a simple forecasting feature. It is not possible to call to third-party algorithms such as R, although this is on the 2017 roadmap. Prior to the 4.1 release, content could be consumed via a browser only. Support for a native iOS app has now been added, but security features and offline capabilities are not part of the near-term roadmap.

  • Data replication: ThoughtSpot achieves high performance by replicating data into its in-memory engine. However, customers who have already made investments in a high-performance analytic engine (such as SAP Hana or Amazon Redshift) would prefer not to move data, thus limiting the range of data for which ThoughtSpot is ideally suited. This also may be why the data volumes accessed by surveyed customers are below the survey average for data originating in Hadoop, as well as in relational data sources. The creation of a reusable data model is not automatic or intuitive; a data connection is first defined and tables imported into ThoughtSpot, with limited ability to prepare the data during the ingestion process. A modeler can then create a shared "worksheet," but this model does not support hierarchies, grouping and visual cues of dimensions and measures, or complex joins; only left outer joins are supported. There is support for multiple fact tables, but fan traps are not handled.

  • Emerging smart data discovery: ThoughtSpot recently introduced several elements of smart data discovery. By selecting A3 analysis, ThoughtSpot will generate a number of charts that describe outliers in the data. There is a short textual explanation. Although these are important innovations, the statistical relevancy of the charts generated does not identify the most-important drivers of a particular metric; the charts generated are more descriptive and do not include cluster analysis or segmentation.

TIBCO Software

TIBCO Spotfire offers extensive capabilities for analytics dashboards, interactive visualization and data preparation in a single design tool while offering flexible processing options either in-memory or in-database. TIBCO has continued to expand its feature set to include advanced analytics, streaming and location intelligence.

In recent years, TIBCO moved to agile release cycles for Spotfire where functionality is first released to the cloud and then on-premises. It does not make its exact release cadence publicly available. TIBCO released Spotfire 7.7 in October of 2016.

TIBCO has also sharpened its product development focus and investment in Spotfire with an emphasis on streamlining all aspects of the analytics content development and exploration workflow, enhancing scalability and extensibility, and expanding connectivity to a range of data sources — including integration with Spotfire's real-time charts (with StreamBase). TIBCO rates Good plus to Excellent across all use cases, and 75% of Spotfire reference customers said they use it for decentralized deployments; this is the highest percentage of any vendor in this Critical Capabilities research.

Strengths

  • Analytics dashboards and advanced interactive exploration: TIBCO's end-to-end user experience centers around dashboards, where users can interact, filter and analyze information, while building dashboards that will be shared and consumed throughout the organization. A single interface serves both purposes — exploring information and designing dashboards — simplifying the user experience. Over the past year, the data preparation and harmonization experience has been more embedded into the exploration experience for iterative analysis. The ability to filter blended datasets, the option of linking different visualizations, extensive visual customization, and automatic recommendations of visualizations further enhance the dashboard design and consumption processes. Excellent interactive visual exploration within the dashboard environment combines a broad range of visualizations with highly interactive exploration of data, and point-and-click interactive features such as grouping, including top N versus rest, binning, displaying values as percent-to-totals, and year-over-year comparisons.

  • Embedded advanced and geospatial analytics: Within Spotfire, analysts have access to an extensive library of embedded advanced analytic functions. One-click calculations for descriptive statistics such as correlations, similarity analysis, clustering, regression modeling, classification, best fit as well as forecasting will appeal to advanced users, analysts and citizen data scientists. Reference customers say they select and use Spotfire for its ease of use in conducting advanced and complex analysis. They also score the platform as above average for these metrics. TIBCO's location analytics capabilities (acquired through Maporama Solutions in 2013) are a good complement to the analytics offering, and a strong differentiator. While most other vendors can only use maps to visualize location-based data, TIBCO provides advanced geospatial functionality such as geocoding, multilayering, geographic clustering, geofencing and several other location-related capabilities. This includes geospatial algorithms for best route/distance between data points on the map, and integrated access to TIBCO's optional data science runtime engine for the analytic language, R, called TERR, which also gives users access to hundreds of R functions, when more-advanced data modeling is required. Because of TERR, TIBCO has one of the strongest embedded advanced analytics offerings in the modern BI and analytics market.

  • Streaming and multistructured data source connectivity: TIBCO Spotfire supports a range of sources (relational, big data and cloud data sources) with new support for Amazon Redshift, OData, Salesforce and Google analytics added this year. Support for streaming data through Spotfire's bidirectional integration with TIBCO's StreamBase, which gives users insight into current streaming data with the ability to drill into history, is a differentiator versus most of the vendors included here. TIBCO's optional component, based on technology licensed from Attivio, supports recommendation-driven blending of structured and unstructured data and the natural-language query and exploration of this multistructured data from within Spotfire.

Areas of Improvement

  • Advanced self-service data preparation and aspects of security: Overall, TIBCO's self-service data preparation features are rated as Good due to limited impact analysis, watermarking of certified data sources, data masking, and recommendations for fixing or enriching data. In terms of security and user administration, to implement row-level security, an administrator must define personalized information links that leverage the source system security. As a result, row-level security is largely administered outside of the TIBCO platform.

  • Following rather than leading on smart data discovery: While TIBCO has invested in some features of smart data discovery — such as support for search-based query generation and exploration via an optional component — and there is early API-level integration for natural-language generation/narration with Automated Insights (a company owned by TIBCO's investor, Vista Partners), it has not leveraged its extensive embedded advanced analytics expertise to lead the market in next-generation machine-learning-enabled automation of the analytics workflow. Automatic generation of new insights and crowdsourced recommendations to the user are on the roadmap, and should close the gap with early disruptors.

  • Gaps in cloud and broad mobile support: TIBCO has been investing in a cloud-first strategy and hybrid cloud-direct query access to data left in place on-premises. However, its limitations around cloud-based packaged content, lack of a robust content marketplace, and gaps between the desktop and cloud authoring environments have resulted in only a Good rating for Spotfire's cloud capabilities. Regarding mobile, native support is limited to iOS devices including native iPhone support added over the past year — native support for Android and Windows Mobile devices is not provided. Offline exploration and support for out-of-the-box integration with mobile device management providers, often an important enterprise feature, are not supported. Responsive design was enhanced over the past year, but there is still only partial support. For example, a KPI chart is responsive but for others, the content author has to manually adjust the sizing to render it properly on each device. A less out-of-the-box, scripted approach to integration with the GPS on a mobile device to automatically filter reports based on location also contributed to TIBCO's Good mobile rating.

Yellowfin

Yellowfin delivers a single, web-based BI and analytics platform with a tightly integrated set of components. Yellowfin supports collaboration, storytelling, mobile and dashboard creation. Yellowfin established a marketplace for clients, which provides, for instance, geopacks, visualizations and prebuilt content. Yellowfin represents the modern design of a central enterprise BI platform and is well-suited for embedded analytics.

Yellowfin has a major release every nine months, with minor releases monthly. Yellowfin's platform is currently on version 7.3. In 2016, Yellowfin continued to deliver out-of-the-box connectors for third-party applications, and introduced new workflow capabilities. Agile centralized BI provisioning is the most-popular use case in this survey (37% of customers), followed by OEM and embedded analytics (29%).

Strengths

  • Leading in collaboration: Yellowfin achieved Excellent scores for the critical capability publish, share and collaborate. Collaboration and social BI can be regarded as the gold standard among the vendors in this Critical Capabilities. Discussion threads, comments and annotations can be seen in a timeline. Users can add reports, dashboards and storyboards as favorites and like or dislike them. Users can mark comments as insightful, and users can "follow" other users in Yellowfin's timeline and see what they are working on. They can also create and share tasks with other users.

  • Single modern platform: Yellowfin got Good to Excellent scores along the business-user-driven analytic workflow. Data inference, data lineage, data modelling and enrichment are examples of strong capabilities on the platform for self-service data preparation. The critical capabilities for interactive visual exploration and analytic dashboards also rated Good to Excellent, with support for the most-popular chart types, bins, and display as percentages (added in 7.3). Yellowfin provides good support for geospatial capabilities and also offers geopacks on its marketplace. In addition, Yellowfin's platform got an Outstanding score for workflow support within the platform.

  • Mobile support and workflow features: Yellowfin supports mobile BI generically with its web-based architecture through the mobile HTML5 browser interface. In addition, iOS and Android smartphones and tablets are natively supported with specific apps. Offline support is provided along with built-in device-based security if a device is lost. In addition, Yellowfin supports third-party MDM providers such as AirWatch and MobileIron.

Areas of Improvement

  • Cloud BI: Yellowfin currently does not offer a public cloud solution. A cloud deployment is available through partners — in line with its partner-centric strategy. As a consequence, some features and aspects of the cloud BI critical capability are not available (such as security certification for software, terms and conditions regarding data ownership, and hybrid connectivity to on-premises data sources). Absent or weak functionality was cited as the most-significant platform problems in the survey by 18% of reference customers.

  • Self-contained in-memory engine: Yellowfin connects directly to a data source or data warehouse. There is a caching mechanism that supports, appends and unions, but at a query level, rather than within the broader dataset. In this regard, Yellowfin is appealing to customers that have already invested in a high-performing analytic database, but less appealing when data is coming from multiple data sources and in need of further data preparation. Poor performance was reported as a barrier to adoption by 15% of reference customers.

  • Less-advanced analytics and smart capabilities: Embedded advanced analytics and smart data discovery were two critical capabilities where Yellowfin achieved only Fair scores. Yellowfin provides time-series analysis and a basic level of scenario modelling using Yellowfin's what-if analysis. Other advanced algorithms — such as decision trees, neural nets and vector machines — are not supported. Advanced analytics visualization requires plugging in third-party visualizations. Smart data discovery capabilities (including automated insight generation, natural-language Q&A, and natural-language generation) are increasingly important capabilities, but not supported.

Zoomdata

Zoomdata supports fast interactive analysis, visualization and dashboards for big and streaming data. It uses micro queries and Spark to push down query processing to underlying big data sources, while estimating results and making them immediately available for interactive analysis as queries process.

Zoomdata content authoring and data management are via a modern, web-based architecture. Zoomdata includes features for administration and security, creation of dashboards and visualizations, and includes a smart connector framework for building custom connections, and prepackaged connectors for a number of common application data sources. With the Data Fusion capabilities in Zoomdata, a business user can join data from multiple data sources, both structured and unstructured, into a single virtualized data source for federated exploration and analysis. Zoomdata's Data DVR capabilities allow users to rewind, fast forward, analyze and compare historical data with real-time streams.

Zoomdata is well-suited to business users and data scientists that need real-time insights from streaming data across a range of big data sources, or for developers that need to embed these insights in applications (32% of customers deployed for this use case). It has innovative capabilities for streaming analytics based on a modern, distributed architecture that harmonizes multiple large data sources into real-time interactive dashboards without moving data. Zoomdata can be deployed on-premises or purchased with a one-click deployment through Amazon Web Services, Google Cloud Platform or Microsoft Azure. Customers looking for traditional reporting or interactive analysis against a data warehouse, for example, would not look to Zoomdata as their best option.

Strengths

  • Native streaming and big data support via a hybrid data architecture: Zoomdata is best-suited to organizations needing to perform real-time interactive analysis on large datasets or streaming data. The platform is natively optimized for a range of big data sources, including Hadoop, NoSQL databases, streaming data, search index data (including Apache Solr and Elastic [formerly Elastic Search]), and cloud data sources (such as Google BigQuery, Google Cloud Dataproc, Microsoft HDInsight, Amazon Redshift, EMR and S3, among others) with support for in-database processing and in-flight data harmonization. Reference customers select Zoomdata for its ability to support large data volumes, data access and integration, and ease of use for business users. They also report among the highest data volumes for queries. Zoomdata's fast processing is achieved through micro-querying data left in place while leveraging the processing power of underlying big data repositories. Instead of one big SQL statement, Zoomdata executes parallelized micro-queries and performs calculations on the data as it streams in from the data source. Data is only pulled into Zoomdata's own Spark instance, or into an external one, as a last resort when this is the best approach for interactive analysis.

  • Analytic dashboards supporting complex types of analysis: Zoomdata offers embedded statistical capabilities for building custom calculations analytic dashboards (rated Good) and can use third-party models such as R and Python for advanced analytics. It can also be integrated with Jupyter Notebook to fit into a data science workflow. Zoomdata's reference customers report that an above-average percentage of data scientists/citizen data scientists use the platform for complex types of analysis. The platform is often chosen for its advanced types of analysis and its support for complex data types. Zoomdata is tied for first place among all the vendors in this Critical Capabilities for complexity of analysis supported by the range of data sources and breadth of usage.

  • Attractive embeddability: With roughly 40% of its 2016 revenue coming from OEMs, Zoomdata launched a developer network to serve the needs of the analytic app market and to continue the expansion of this part of its business. Zoomdata SDKs and extensive REST APIs support the customization and embedding of most aspects of the platform, including administrative functions, connectors and visualizations. Zoomdata visualizations work with any base map supported by the Leaflet JavaScript library, including OpenStreetMap, MapQuest, Mapbox, and any other provider that uses a compatible tile map server URL. Although out-of-the-box distance calculations are not supported, Zoomdata's geospatial analysis has the ability to zoom to the lowest layer available from the map provider, such as ZIP Code, city or street. Customers can create custom map visualizations using either the embedded Zoomdata Chart Studio, or through external applications using the Zoomdata SDK.

Areas of Improvement

  • Mobile, collaboration, scheduling and alerting: For mobile devices, HTML5 is supported, but Zoomdata does not offer native applications for specific mobile operating systems. Zoomdata is certified and supported to run in Safari browsers on Apple iOS tablets, but not on iPhones. Similarly, because Zoomdata is a web-based platform, there is no offline support or integration with MDM security providers. Storytelling and integration with social platforms are also not supported. Regarding scheduling and alerting, the product does not provide Zoomdata content to end users on a scheduled basis (user-defined or otherwise).

  • Self-service data preparation: Data lineage and impact analysis are not supported, nor is inference beyond the basic automatic inference of data types. Advanced features, such as those automatically suggesting or inferring relationships for joins or hierarchies, are not supported. In terms of data transformations, a user can rename a data source and other data transformations, but cannot yet combine and split columns or replace values — these are on the roadmap. Most of these gaps are considered basic and offered by most other modern BI platforms.

  • Advanced interactive exploration: Some aspects of advanced interactive exploration are limited. For example, Zoomdata dashboards have a global filter feature where users can apply a common filter to all charts from the same data source, but not across disparate data sources. Common exploration features (such as display as a percentage) are supported via a toggle out-of-the-box, but the creation of custom groups and bins is not available with the visualization environment — instead they must be defined in the self-service data preparation layer via custom calculations, which limits seamless interactive analysis. Conditional formatting beyond that which can be defined with calculations is not yet supported — more extensive and out-of-the-box conditional formatting features are on the vendor's roadmap.

Context

This Critical Capabilities research evaluates products included in the 2017 "Magic Quadrant for Business Intelligence and Analytics Platforms" on 15 critical capabilities in support of the five main use cases for BI and analytics platforms.

Product/Service Class Definition

Gartner's view is that the market for modern BI and analytics platforms will remain one of the fastest-growing software markets. The 15 critical capabilities defined in this research represent mainstream buying requirements for customers to modernize their BI and analytic platforms.

The modern BI and analytics market grew 64% in 2015 on a constant currency basis, with a projected growth rate of 30% in 2016 (see "Forecast: Enterprise Software Markets, Worldwide, 4Q16 Update" and "Forecast: Enterprise Software Markets, Worldwide, 2013-2020, 4Q16 Update" ).

Much of net new BI buying is driven by the following market activity:

  • Traditional BI and modern BI in a single platform: As traditional BI platform vendors initially lacked agile and data discovery capabilities, customers who were early to adopt data discovery typically augmented their traditional BI platforms with what were then specialty products. In the last two years, modern BI solutions from traditional BI vendors have matured to the point that it is possible to address both Mode 1 requirements and Mode 2 requirements in a single platform. Conversely, modern BI platforms have continued to advance their capabilities in governance and distributed reports. Customers would like to support both use cases (agile centralized BI provisioning for Mode 1) as well as governed data discovery or decentralized analytics (for Mode 2) in a single platform. The degree that this is possible (and "best in class" for both modes) is highly dependent on the particular vendor.

  • Expansion of data discovery dominates new investment: Continued investment in decentralized analytics and large, governed data discovery deployments is expected to continue. Current IT-centric vendors will continue to shift the focus of their new-product investments and platforms to more-modern capabilities, with more-frequent releases. Data-discovery-oriented, modern BI platforms will increasingly marginalize IT-authored static reporting approaches. IT-authored system-of-record reporting will not disappear, but will gradually account for a smaller percentage of overall analytics use. At the same time, a larger percentage of data discovery deployments will expand overall user adoption and easy-to-use, centrally deployed BI platforms based on modern architectures and broader capabilities will be key drivers of market growth.

  • Self-service data preparation and enrichment address a high-value data discovery challenge: The shift toward business-user-driven data discovery has highlighted the need to address the significant challenges of data preparation to enable broader and more-governed use. Self-service data preparation capabilities are emerging that extend beyond the current data mashup capabilities of most data discovery tools that help users prepare their data for analysis, but can be very time-consuming. Self-service data preparation platforms enable business users to reduce the time and complexity of preparing data for analysis in a governed and reusable way. They feature capabilities like visual data flow building and automation, semantic autodiscovery, intelligent joins, intelligent profiling, hierarchy generation, data lineage, and data blending on varied data sources, including multistructured data and enrichment. Many of these platforms also feature automated machine-learning algorithms in the background that visually highlight the structure, distribution, anomalies and repetitive patterns in data, with guided business-user-oriented tools to suggest how to resolve issues and enhance data. The intent of these tools is to make the data integration process accessible to business analysts — in addition to traditional IT users — to address the ongoing and high-value problem of data preparation.

  • Smart data discovery will extend data discovery to a wider range of users and enhance insights and interpretation: These emerging capabilities facilitate discovery of hidden patterns in large, complex and increasingly multistructured datasets, without building models or writing algorithms or queries. This goes beyond data discovery, because business users and business analysts can benefit from advanced analytics (to highlight and visualize important findings, correlations, clusters, predictions, outliers, anomalies, linkages or trends in data that are relevant to the user), with user interaction and exploration via interactive visualizations, search and natural-language query technologies. Some tools also interpret results for the user with natural-language generation of text to highlight patterns and explain insights. This will also reduce the time to insight, as well as the time and expertise needed for manual data exploration and modeling. Smart data discovery does not replace advanced analytics or the data scientist; it complements them, by adding a class of citizen data scientists that can develop hypotheses that can be explored in more detail and validated by data scientist.

  • Marketplaces will extend capabilities and analytic maturity: A handful of vendors in this report provide marketplaces through which developers can publish and sell extensions, data, and applications to improve the out-of-the-box capabilities provided solely from the vendor. Algorithms that detect patterns, recommend associated products, and optimize prices will be a point of differentiation for digital businesses, and will become an important content type for these marketplaces.

  • Cloud BI will continue to grow as data shifts to the cloud: Adoption intentions for cloud BI and analytics have accelerated. About 51% of respondents to Gartner's companion Magic Quadrant survey (compared with 47% in 2016 and 41% in 2015) said they either have already deployed, or plan to deploy their BI in either a private, public or hybrid cloud during the next 12 months. New capabilities are often delivered first and sometimes only via the cloud. Amazon's entrance with QuickSight (released in 4Q16) could further accelerate cloud BI adoption. With increasing support for hybrid cloud connectivity — which many BI vendors now support or have on their roadmaps — customers have greater flexibility and a glide path to cloud BI. However, larger BI platform vendors (such as IBM, Oracle, SAP or Microsoft) often rely predominantly (or entirely) on their own cloud data centers, while other vendors give customers greater flexibility in cloud infrastructure.

  • Streaming data: Much of the last decade of analytics investment has centered on analyzing internally generated transactional data to understand customer behavior and internal processes. The next 10 years will be driven by investments in applications that use the Internet of Things (IoT). The fastest-growing kinds of data will come from real-time event streams, sensors and machine data, and events generated by devices. These new use cases, combined with insights from other new (multistructured) data types (together with new types of analysis), will generate the next major wave of analytics investment and business transformation. This will enable companies that have historically competed on physical assets to compete on information assets.

  • Multistructured data analytics: Expanded investment in new types of analysis on a variety of structured and unstructured data will deliver new insights that drive business value and transformation. This may include external and public datasets, combined with internally generated data. The desire to analyze information in relational data sources has expanded to include JSON, personal data sources, Hadoop and NoSQL.

  • Embedded BI: Organizations will invest in embedding BI content (reports and dashboards), interactive analysis, predictive and prescriptive analytics in applications and business processes — all of which will deliver optimized recommendations and courses of action to nontraditional BI users at the point of decision or action (increasingly mobile) — to further extend the pervasiveness and benefits of BI and analytics.

  • Customer-facing analytics and data monetization: Companies will increasingly invest in capabilities that transform analytics from a cost center to a profit center as they find new ways to productize the data assets they have (or can assemble) to improve customer relationships, create new business models, and generate new sources of revenue.

  • Collaboration and social capabilities: Together with the crowdsourcing and sharing of BI content and analysis, these may drive a more-pervasive use and higher business value from BI investments.

Critical Capabilities Definition

Vendors are assessed according to the following 15 critical capabilities. Changes, additions and deletions from last year's Critical Capabilities are listed in Note 1. Subcriteria and detailed functionality requirements are included in a published RFP document (see "Toolkit: BI and Analytics Platform RFP" ).

Admin, Security and Architecture

Capabilities that enable platform security, administering users, usage monitoring, auditing platform access and utilization, optimizing performance and ensuring high availability and disaster recovery. This also includes the ability to run on multiple operating systems.

Data Source Connectivity

Capabilities that allow users to connect to structured and unstructured data contained within various types of storage platforms, including personal data sources, relational, NoSQL and direct HDFS. The ability to access business applications and ERP systems is included.

Cloud BI

PaaS and SaaS for building, deploying and managing analytics and analytic applications in the cloud based on data both in the cloud and with hybrid connectivity to on-premises data sources. Marketplaces and prebuilt content to cloud-data sources are included.

Self-Contained ETL and Data Storage

Platform capabilities for accessing, integrating, transforming and loading data into a self-contained performance engine with the ability to index data and manage data loads and refresh scheduling of loaded data.

Self-Service Data Preparation

Drag-and-drop cleansing, modeling, and blending of multiple data sources and creation of analytic models.

Advanced capabilities include machine-learning-enabled semantic autodiscovery, intelligent joins, intelligent profiling, hierarchy generation, data lineage and data blending on varied data sources, including multistructured data.

Metadata Management

Tools for enabling users to leverage the same system-of-record semantic model or for creating a semantic model automatically.

Modelers should be able to search, capture, store, reuse and publish metadata objects such as dimensions, hierarchies and measures, as well as conduct impact analysis for changed objects.

Embedded Advanced Analytics

Enables users to easily access advanced analytics capabilities that are self-contained within the platform itself, or through the import and integration of externally developed models.

Smart Data Discovery

Automatically visualizes the most important findings such as correlations, exceptions, clusters, links and predictions in data that are relevant to users without requiring them to build models or write algorithms.

Users explore data via visualizations, autogenerated voice or text narration, search, and natural-language query technologies. Forecasting and clustering should be menu-driven. Support for advanced visualizations such as decision trees should be out-of-the-box.

Interactive Visual Exploration

Enables the exploration of data via an array of visualization options that go beyond those of basic pie, bar and line charts, to include trellis, heat and tree maps, scatter plots, and other special-purpose visuals.

These tools enable users to analyze and manipulate the data by interacting directly with a visual representation of it to display as percentages bins and groups.

Analytic Dashboards

The ability to create highly interactive dashboards and content with visual exploration and embedded advanced and geospatial analytics to be consumed by others. Support for offline dashboards should be included.

Mobile Exploration and Authoring

Enables organizations to develop and deliver content to mobile devices in a publishing and/or interactive mode.

Takes advantage of the native capabilities of mobile devices, such as touchscreen, camera and location awareness. Device-based security and integration with third-party MDM solutions should be supported.

Embed Analytic Content

Capabilities including a software developer's kit with APIs and support for open standards for creating and modifying analytic content and visualizations, and embedding them into a business process, and/or an application or portal.

Publish, Share and Collaborate

Capabilities that allow users to publish, deploy and operationalize analytic content through various output types and distribution methods with support for content search, scheduling and alerts. Enables users to share and rate content via discussion threads, chat and storyboards.

Platform and Workflow Integration

This capability considers the degree that the capabilities are offered in a single, seamless product or across multiple products with little integration.

Ease of Use and Visual Appeal

Ease of use to administer and deploy the platform, create content, consume and interact with content as well as the visual appeal.

Use Cases

Agile Centralized BI Provisioning

Supports an agile IT-enabled workflow from data to centrally delivered-and-managed analytic content using self-contained data management capabilities of the platform.

Agile centralized BI provisioning enables an information consumer to access their KPIs from an information portal — whether on a mobile device or embedded in an analytic application — to monitor and measure the performance of the business. In a modern BI and analytics platform, interactivity is often supported out-of-the-box and automatically. This is in contrast to traditional reporting-based platforms in which interactivity is limited to what is designed in by the content author and in which a data warehouse must first be built.

The highest-weighted capabilities in this use case include:

  • Admin, security and architecture

  • Self-contained ETL and data storage

  • Metadata management

  • Analytic dashboards

  • Mobile exploration and authoring

  • Publish, share and collaborate

  • Ease of use and visual appeal

Decentralized Analytics

Supports a workflow from data to self-service analytics for individual business units and users.

On the analytics spectrum, users of platforms that excel at the decentralized analytics use case can explore data using highly interactive descriptive analytics ("what happened" or "what is happening") or diagnostic analytics ("Why did something happen?" "Where are areas of opportunity or risk?" or "What if?").

Increasingly, because of the embedded advanced analytics offered by many vendors, users can extend their analysis to some advanced descriptive analysis (for example, clustering, segmenting and correlations) and to a basic level of predictive analytics (for example, forecasting and trends). They can also prepare their own data for analysis, reducing their reliance on IT and time to insight. As decentralized analytics becomes more pervasive, the risk of multiple sources of the truth becomes a concern and decentralized analytics may evolve to governed data discovery over time and as a deployment grows.

The highest weighted capabilities in this use case include:

  • Data source connectivity

  • Self-contained ETL and data storage

  • Self-service data preparation

  • Embedded advanced analytics

  • Interactive visual exploration

  • Analytic dashboards

  • Ease of use and visual appeal

Governed Data Discovery

Supports a workflow from data to self-service analytics to system-of-record, IT-managed content with governance, reusability and promotability of user-generated content.

Capabilities that govern, promote and widely share content are what most differentiate governed data discovery from decentralized analytics. With the success of data discovery tools in driving business value, many organizations would increasingly like to use data discovery capabilities for a broader range of analysis and an expanded set of users than was previously addressed by IT-centric enterprise reporting platforms. Governed data discovery enables users to access, blend and prepare data, then visually explore, find and share patterns with minimal IT support, or technical and statistical skills. At the same time, it must also satisfy enterprise IT requirements for business-user-generated model promotability, data reuse and governance. In particular, users should be able to reuse sanctioned business-user-created data or datasets, derived relationships, derived business models, derived KPIs, and metrics that support analyses.

Governed data discovery enables pervasive deployment of data discovery in the enterprise at scale without proliferating data discovery sprawl. The expanded adoption of data discovery also requires BI and analytics leaders to redesign BI and analytics deployment models and practices, moving from an IT-centric to an agile and decentralized (yet governed and managed) approach. This would include putting in place a "prototype, pilot and production" process in which user-generated content is created as a preliminary model. Some of these would need to be recurring analysis and are promoted to a pilot phase, and others are promoted to production and operationalized for regular analysis as part of the system of record. Alternately, governance can be implemented after broad sharing of content as centralized experts proactively monitor usage.

The highest weighted features in this use case include:

  • Admin, security and architecture

  • Data source connectivity

  • Self-contained ETL and data storage

  • Metadata management

  • Interactive visual exploration

  • Ease of use and visual appeal

OEM or Embedded BI

These capabilities are used to create and modify analytic content, visualizations and applications and embed them into a business process, and/or an application or portal.

They support a workflow from data to embedded BI content in a process or application, as well as extending out-of-the-box capabilities. They can reside outside the application, reusing the analytic infrastructure, but must be easily and seamlessly accessible from inside the application, without forcing users to switch between systems. The ability to integrate BI and analytics with the application architecture will enable users to choose where in the business process the analytics should be embedded.

The highest-weighted capabilities in this use case include:

  • Admin, security and architecture

  • Data source connectivity

  • Analytic dashboards

  • Embed analytic content

  • Publish, share and collaborate

Extranet Deployment

Supports a workflow similar to agile centralized BI provisioning for the external customer or, in the public sector, citizen access to analytic content.

In addition, capabilities for embedding and cloud deployment are typically required for extranet deployments.

The highest-weighted capabilities in this use case include:

  • Admin, security and architecture

  • Cloud BI

  • Metadata management

  • Analytic dashboards

  • Embed analytic content

Vendors Added and Dropped

Added

ThoughtSpot, Datameer, Oracle and Zoomdata were added to the Magic Quadrant and hence to this Critical Capabilities this year as they met all of the inclusion criteria and were ranked in the top 24 of assessed vendors based on an evaluation of their modern BI product offerings against the current set of critical capabilities and other inclusion metrics defined for the Magic Quadrant.

Dropped

Platfora was acquired by Workday and is no longer being sold as a stand-alone BI platform.

BeyondCore was acquired by Salesforce and is included in the Salesforce assessment.

Datawatch and GoodData were excluded because they shifted their market emphasis.

Inclusion Criteria

Vendors included in this research also appear in the 2017 "Magic Quadrant for Business Intelligence and Analytics Platforms."

The number of vendors on this year's Magic Quadrant was limited to 24. We ranked vendors that met all the inclusion criteria below.

Modern BI and Analytics Platform Assessment

This was evaluated by Gartner analysts and was determined by the extent of IT involvement that is considered to be mandatory before the platform can be used by a business analyst/information worker to analyze data, without IT assistance. Products that require significant IT involvement, either internal or external to the platform, in order to load and model data, create a semantic layer, build data structures as a prerequisite to using the BI platform or are IT developer-centric platforms focused on building analytic applications, do not meet the criteria of a modern BI and analytics platform and were not evaluated further for inclusion. Products that met the modern criteria were evaluated for inclusion in the Magic Quadrant based on a funnel methodology where requirements for each tier must be met in order to progress to the next tier. Tiers 1 to 3 are evaluated at the vendor level; Tiers 4 and 5 are evaluated at the product level.

Vendor-Level Criteria

  • Tier 1. Market Presence — A composite metric assessing both the interest of Gartner's client base and that of the broader market, through internet search volume, job postings and trend analysis, was conducted for each vendor.

  • Tier 2. Revenue* — For those vendors meeting the market presence criterion (Tier 1), BI and analytics revenue for each vendor was assessed and evaluated. For this assessment, two common license models were assessed and revenue from each was combined (if applicable) and evaluated against the three revenue inclusion levels (shown below) for qualification:

  1. Perpetual License Model — Software license, maintenance and upgrade revenue (excluding hardware and services) for calendar years 2014, 2015 and 2016 (estimated).

  2. SaaS Subscription Model — Annual contract value (ACV) for year-ends 2014, 2015 and projected ACV for year-end 2016, excluding any services included in annual contract. For multiyear contracts, only the contract value for the first 12 months should be used for this calculation.

  • Revenue inclusion levels are as follows:

    • $25 million 2016 (estimated) combined perpetual license revenue + 2016 (estimated) ACV, or

    • $15 million 2016 (estimated) combined perpetual license revenue + 2016 (estimated) ACV with 50% year-over-year growth, or

    • $10 million 2016 (estimated) combined perpetual license revenue + 2016 (estimated) ACV with 100% year-over-year growth

  • * Gartner defines total software revenue as revenue that is generated from appliances, new licenses, updates, subscriptions and hosting, technical support, and maintenance. Professional services revenue and hardware revenue are not included in total software revenue (see "Market Share Analysis: Business Intelligence and Analytics Software, 2015" ).

  • Tier 3. Magic Quadrant Process Participation — Participation in the Magic Quadrant process requires the following input:

    • Completing and providing documentation for an RFP-style questionnaire of detailed critical capabilities.

    • Completing an online questionnaire around market presence, growth, go-to-market strategy and differentiation.

    • Submission of a video up to one-hour long that demonstrates how included products deliver on the predefined analytic scenarios defined by Gartner (we only look at the first hour; anything beyond that is not considered).

    • Verification of final BI and analytics revenue for 2014, 2015 and 2016 (estimated).

    • Providing references for an online customer and OEM survey.

    • Providing a vendor briefing to the Magic Quadrant authors.

    • Providing access to evaluation software.

    • Providing factual review of sections in the Magic Quadrant research.

Product-Level Criteria

  • Tier 4. Breadth of Coverage — The vendor must demonstrate breadth across vertical industries and geographic regions, as specified by Gartner.

  • Tier 5. Product Assessment — Products that progressed to this final tier were assessed by Gartner analysts using the information provided by each vendor in the data collection exercise outlined above. The final step involved narrowing down the field to 24 vendors for inclusion in the Magic Quadrant.

  • Gartner has full discretion to include a vendor on the Magic Quadrant regardless of their level of participation in the Magic Quadrant process, if the vendor is deemed important to the market. This discretion was not applied this year as all vendors fully participated in the process.

Table 1.   Weighting for Critical Capabilities in Use Cases

Critical Capabilities

Agile Centralized BI Provisioning

Decentralized Analytics

Governed Data Discovery

OEM or Embedded BI

Extranet Deployment

Admin, Security and Architecture

10%

5%

10%

10%

10%

Data Source Connectivity

5%

10%

10%

15%

0%

Cloud BI

0%

5%

5%

0%

25%

Self-Contained ETL and Data Storage

10%

10%

10%

5%

5%

Self-Service Data Preparation

0%

13%

8%

5%

0%

Metadata Management

20%

0%

10%

0%

10%

Embedded Advanced Analytics

0%

10%

5%

0%

0%

Smart Data Discovery

0%

5%

5%

0%

0%

Interactive Visual Exploration

0%

15%

10%

0%

5%

Analytic Dashboards

15%

10%

5%

10%

10%

Mobile Exploration and Authoring

10%

0%

5%

0%

5%

Embed Analytic Content

0%

0%

0%

40%

25%

Publish, Share and Collaborate

15%

5%

5%

10%

0%

Platform and Workflow Integration

5%

2%

2%

0%

0%

Ease of Use and Visual Appeal

10%

10%

10%

5%

5%

Total

100%

100%

100%

100%

100%

As of January 2017

Source: Gartner (March 2017)

This methodology requires analysts to identify the critical capabilities for a class of products/services. Each capability is then weighed in terms of its relative importance for specific product/service use cases.

Critical Capabilities Rating

Table 2 shows the product/service scores for each use case. The scores, which are generated by multiplying the use-case weightings by the product/service ratings, summarize how well the critical capabilities are met for each use case.

Table 2.   Product Score in Use Cases

Use Cases

Agile Centralized BI Provisioning

Decentralized Analytics

Governed Data Discovery

OEM or Embedded BI

Extranet Deployment

Alteryx

2.82

3.11

2.98

2.69

2.59

Birst

3.80

3.31

3.59

4.15

4.18

Board International

3.24

3.22

3.27

3.05

2.97

ClearStory Data

3.86

3.62

3.77

4.04

3.84

Datameer

2.74

2.84

2.86

3.15

2.92

Domo

3.30

3.12

3.19

3.31

3.29

IBM (Cognos Analytics)

2.82

2.71

2.81

2.56

2.69

IBM (Watson Analytics)

2.47

2.69

2.64

2.32

2.55

Information Builders

3.55

3.14

3.36

3.93

3.58

Logi Analytics

3.42

3.62

3.42

4.14

3.51

Microsoft

3.32

3.46

3.54

3.54

3.56

MicroStrategy

3.97

3.65

3.83

3.59

3.56

Oracle

3.17

3.05

3.15

3.65

3.62

Pentaho

3.20

3.14

3.19

3.81

3.34

Pyramid Analytics

3.38

3.37

3.32

3.52

3.37

Qlik

3.36

3.29

3.34

3.86

3.63

Salesforce

3.42

3.27

3.39

3.87

3.95

SAP (BusinessObjects Cloud)

2.75

2.92

2.79

2.38

2.50

SAP (BusinessObjects Lumira)

2.96

2.81

2.94

2.99

2.94

SAS

3.38

3.53

3.48

2.94

3.09

Sisense

3.46

3.25

3.35

3.54

3.39

Tableau

3.53

3.55

3.52

3.32

3.37

ThoughtSpot

3.30

2.82

3.05

2.61

2.79

TIBCO Software

3.62

3.71

3.63

4.06

3.76

Yellowfin

3.61

3.17

3.25

3.59

3.34

Zoomdata

2.81

2.77

2.77

3.36

3.08

Source: Gartner (March 2017)

Acronym Key and Glossary Terms

ACV annual contract value
AWS Amazon Web Services
BI business intelligence
ETL extraction, transformation and loading
HDFS Hadoop Distributed File System
IoT Internet of Things
KPI key performance indicator
PaaS platform as a service
SaaS software as a service

Evidence

Gartner's analysis, the ratings and commentary in this report are based on a number of sources including:

  • Customer perceptions of each vendor's strengths and challenges (as gleaned from their BI-related inquiries to Gartner).

  • An online survey of vendors' reference customers (which was conducted during October 2016 and yielded 1,931 responses).

  • A questionnaire completed by the vendors.

  • Vendor briefings (including product demonstrations, strategy and operations).

  • An extensive RFP questionnaire inquiring about how each vendor delivers the specific features that make up our 15 critical capabilities (see "Toolkit: BI and Analytics Platform RFP" ).

  • A prepared video demonstration of how well vendor BI platforms address specific functionality requirements across the critical capabilities.

  • Access to evaluation software from each vendor.

Rankings refer to where a product or vendor is positioned relative to other vendors based on a combination of customer reference survey and analyst opinion. Ratings refer to capability scores in Figure 6.

Critical Capabilities Methodology

This methodology requires analysts to identify the critical capabilities for a class of products or services. Each capability is then weighted in terms of its relative importance for specific product or service use cases. Next, products/services are rated in terms of how well they achieve each of the critical capabilities. A score that summarizes how well they meet the critical capabilities for each use case is then calculated for each product/service.

"Critical capabilities" are attributes that differentiate products/services in a class in terms of their quality and performance. Gartner recommends that users consider the set of critical capabilities as some of the most important criteria for acquisition decisions.

In defining the product/service category for evaluation, the analyst first identifies the leading uses for the products/services in this market. What needs are end-users looking to fulfill, when considering products/services in this market? Use cases should match common client deployment scenarios. These distinct client scenarios define the Use Cases.

The analyst then identifies the critical capabilities. These capabilities are generalized groups of features commonly required by this class of products/services. Each capability is assigned a level of importance in fulfilling that particular need; some sets of features are more important than others, depending on the use case being evaluated.

Each vendor’s product or service is evaluated in terms of how well it delivers each capability, on a five-point scale. These ratings are displayed side-by-side for all vendors, allowing easy comparisons between the different sets of features.

Ratings and summary scores range from 1.0 to 5.0:

1 = Poor or Absent: most or all defined requirements for a capability are not achieved

2 = Fair: some requirements are not achieved

3 = Good: meets requirements

4 = Excellent: meets or exceeds some requirements

5 = Outstanding: significantly exceeds requirements

To determine an overall score for each product in the use cases, the product ratings are multiplied by the weightings to come up with the product score in use cases.

The critical capabilities Gartner has selected do not represent all capabilities for any product; therefore, may not represent those most important for a specific use situation or business objective. Clients should use a critical capabilities analysis as one of several sources of input about a product before making a product/service decision.