Finance could use artificial intelligence (AI) to cut time, costs and errors in many of its budget, forecast and planning activities, but few financial planning and analysis (FP&A) leaders do. The problem is partly data and partly people.
“AI can accurately analyze high volumes of data and identify patterns and apply decision rules much faster than people using traditional technology or spreadsheet-based solutions, but only 2% of FP&A functions have adopted AI,” says Richard Ries, VP, Advisory, Gartner. “Our research suggests the low adoption rate stems back to three main barriers. Leaders must remove those obstacles to realize the full potential of AI.”
Improve data quality
Data trains and runs AI models and drives the quality of the output. FP&A functions today can access masses of financial, performance and operational data, but its quality is uneven. Consistency is a challenge because of the way data is created, maintained and handled.
Financial and operational data is often captured, maintained and monitored without standard definitions, and stored in different locations across business and FP&A systems. To deliver accurate results, AI models require a high volume of data that is consistent and relevant enough to produce meaningful predictions and classifications. FP&A leaders must ensure that data is properly cleaned and standardized.
- Standardize business-critical data that is common and shared across the organization to address anomalies by identifying missing, incomplete and duplicative data. It may be helpful to create a data catalog to maintain an inventory of datasets. Define common data elements used for analysis to ensure the AI isn’t modeled and run around the wrong data elements.
- Create an organized data environment to store, maintain and handle data effectively within data lakes and warehouses for training AI models. Partner with IT to examine the ability of current FP&A data storage solution to support the need for well-defined data lakes and warehouses.
Learn more: Finance Analytics
Remove human bias from algorithms
Data scientists or data engineers in the FP&A function may be biased by past projections and experiences to select evidence that supports their preconceived approach.
If these biases bleed into the assumptions they program into AI models, the AI could misinterpret data or use inaccurate information to draw conclusions used for business decision making— thus exposing the organization to financial and regulatory risks.
- Ensure datasets are diverse so staff can run correlations and comparative analysis and spot right and wrong answers when AI produces a forecast or predicts business performance.
- Validate AI outputs in a larger group of diverse perspectives. For instance, the IT team can help validate the AI model’s ability to identify anomalies or exceptions that may disrupt the outcomes.
FP&A leaders will also gain business leaders’ trust in using AI outcomes when making strategic, judgment-based decisions by showing that they monitor and remove human bias from AI models.
Read more: Gartner Predicts the Future of AI Technologies
Reassure employees that AI isn’t a job killer
Employees may fear that AI will make their jobs redundant.
FP&A leaders must identify ways to reassure FP&A staff and build digital dexterity to complement AI with human judgment. Building digital dexterity in FP&A staff helps them to embrace AI in the workplace and improve decision making and process efficiency through AI.
- Build digital dexterity, or the ability and ambition of the staff to use technology to drive better business outcomes, in employees. As employee’s technology acumen grows they become more open to AI and agile, helping them effectively use, interpret and apply AI to FP&A activities such as forecasting and analytics.
- Train staff in analytical roles, such as FP&A analyst, data scientist, data engineer and forecasting analyst, to handle anomalies and exceptions and anticipate their occurrence based on factors extraneous to the trained AI model. This new dimension of their role will increase the efficiency and quality of their analysis.
- Train staff in business partnering roles, such as a finance business partner and business-aligned FP&A staff, to apply creativity and empathy when using AI-enabled analysis.
Read more: Add Data Science Skills to Corporate Finance Decisions