Why knowing isn't the same as doing

What makes data and AI learning stick at work?

Nearly nine in ten people who complete our data and AI education say it was interesting and insightful. That's good to know. But it's not the measure that matters most.

The harder question is what happens afterwards.

Do people ask better questions? Question an AI-generated answer when something doesn't look right? Push back when a dashboard doesn't answer the business question? Stop producing work nobody actually needs?

Because knowing more isn't necessarily the same as working differently.

What the research tells us

89 studies. Nearly 12,500 people. The same problem keeps appearing: knowing something doesn't mean you'll use it at work.

Research into training transfer found that knowledge matters, but so do motivation and the environment people return to after learning. In other words, you can teach someone perfectly and still fail to change what they do.

168 courses. More than 19,000 learners. Putting learning online wasn't the thing that made it work.

A major workplace meta-analysis found broadly similar outcomes between web-based and classroom instruction in several areas. The bigger gains came when digital learning supplemented other teaching, with practice and feedback also playing a role.

And the skills employers value are moving in the same direction.

The World Economic Forum ranks analytical thinking as employers' most important core skill, with curiosity and lifelong learning continuing to rise in importance.

Put those findings together and the implication is fairly straightforward: effective data and AI education can't just transfer information. It has to develop people's ability to question, judge and apply it, then give them an environment in which they can actually use it.

What that looks like in practice

That's why our education doesn't stop at explaining how something works.

Someone learning about AI needs to know how to use the tool, but also when to question its answer, what information might be missing and when AI isn't the right answer at all.

Someone learning about data needs the terminology and practical skills, but also the confidence to ask who needs the data, what decision it's supposed to support and whether they're answering the right question in the first place.

Our OnDemand learning builds that foundation at scale. Role and industry context make it relevant. Scenarios give people opportunities to practise their judgement. Live learning, applied activities and reinforcement help turn that learning into something people use and implement into their everyday.

Across more than 700 responses to our learner surveys, 89% reported greater confidence incorporating data tools into their daily work and 85% greater confidence identifying the KPIs that matter to their organisation.

One learner started asking colleagues who requested data what they needed it for before beginning the work. Twice in one week, they realised they didn't need the data after all. Applied across the team, the organisation estimates that simple change could save around 7,000 hours a year.

That's the outcome we're interested in.

Not simply whether someone completed the learning or enjoyed it, but whether they do something differently because of it.