Research from McKinsey suggests that in-house analytics academies are better than outsourced training. But what do you do when your in-house resources are limited? Do you think it is better to start with outsourced training for a broad group, or just start with a very small in-house training program?

2.5k views3 Upvotes4 Comments

Senior Director, Data & AI in Travel and Hospitality, 10,001+ employees
IMHO, I would suggest that work with HR to define a proper training plan (both technical and non-technical), start with data literacy workshop, cloud vendor free training, Udemy for business, request budget for certification, that's a start, then scale to a train the trainers and learning by doing modality.
Chief Executive Officer in Software, 51 - 200 employees
When in-house resources are limited, it can be a challenge to establish a comprehensive in-house analytics academy. In such cases, it may be more practical to start with outsourced training for a broad group. This allows for the dissemination of foundational knowledge and skills across the organization, ensuring a basic level of competency in analytics. While outsourced training may not provide the same level of customization and depth as an in-house program, it can serve as a starting point to build awareness and establish a common understanding of analytics principles.

Starting with a small in-house training program can also be a viable option, especially if there are specific needs or areas of focus that require specialized training. This approach allows for targeted skill development within a select group of individuals who can then act as champions and internal trainers for future training initiatives. It may be more resource-efficient and tailored to address specific organizational requirements.

Ultimately, the decision between outsourced training for a broad group or a small in-house training program depends on the organization's unique circumstances, available resources, and strategic objectives. Both approaches have their merits, and the choice should be based on a careful assessment of the organization's needs, capacity, and long-term goals.

I hope this will help. 
Director of Data, Self-employed
We started by building a community that was self-training/self-inspiring. So even with limited "official" resources it's easy to just set up a breakfast or coffee group around training specific skills. There's tons of free self-paced learning out there (MSFT and other vendors,...)... we started with those and then just had groups meet once a week to discuss and keep them engaged. We're now doing "Python Mondays"... next to this we have official trainings with external trainers at regular intervals (once a month) but we grew them as the need grew. So early on we only had beginner sessions... as our beginners got better we introduced intermediate trainings... you can step it up together with the ROI you're seeing. So there is no need for a massive up front investment. Just start small and use each step as the business case for the next one. Drive it yourself ;)
Chief Data Officer in Media, 2 - 10 employees
It’s important to define training tiers and objectives for each one before anything else. Employees are separated into tiers based on their job function’s usage of data and analytics. Some need basic data literacy, while others need more advanced model literacy. Develop a training timeline based on when each tier must be fully trained.

Self-service tools training should also be considered. There are opportunities to leverage external tools vendors for supporting training. Many have free resources and low-cost advanced training options.

Evaluate the rest of the curriculum and decide on internal vs. external with the big picture laid out in advance.

Content you might like




90.4k views253 Upvotes71 Comments

CTO in Software, 201 - 500 employees
Without a doubt - Technical Debt! It's a ball and chain that creates an ever increasing drag on any organization, stifles innovation, and prevents transformation.
Read More Comments
39.9k views130 Upvotes318 Comments

Data mesh44%

Data fabric39%


Don’t know6%