Forget the framework. Fix the problem.

Imagine you had it all. A perfect skills-to-job mapping. A complete framework. Clean data. Would any of it change anything in your business tomorrow?

That challenge sat at the heart of our World of Learning session on 6 October, led by Phill Lord-David, our Principal Product Manager, and Justin Chave, our Head of L&D.

The pattern is familiar: You map the skills, gather the activity data, assign the content, and track the completions. Then someone asks “so what?”… and the room goes quiet.

The real question isn’t whether you have the data. It’s whether your people have the skills to do the things that matter: growing revenue, driving innovation, and running teams more efficiently than the rest of the market.

Here are five ideas from the session that challenge how most organisations approach skills, and what to do differently instead.

1. Frameworks answer the wrong question

A skills framework tells you what skills exist on paper. It doesn’t tell you whether they’re being used, or whether they’re the ones your business needs this quarter.

We build them anyway because they feel safe. They look rigorous, they produce something tangible, and they’re what everyone else is doing. But roles now change faster than any framework can be updated, and skills move between roles rather than sitting neatly inside them. By the time the mapping is finished, it’s already a snapshot of last year.

The cost isn’t just time. It’s credibility. As Justin asked the room, how often has a skills gap report been met with silence from leadership? Silence usually means the data didn’t connect to anything they were worried about.

There are two genuine reasons organisations reach for a framework: execs want visibility, and employees want a clear sense of what’s next. Both are real needs, but neither is a business metric. The trick is to meet each one directly rather than building a giant structure in the hope it covers both.

Try this: before adding anything to a framework, ask “if this data changed tomorrow, what decision would we make differently?” If nobody can answer, it’s activity.

2. Start with the metric, not the skill

Most skills programmes are built forwards: identify skills, design learning, then hope a business result appears. Flip it, and everything gets sharper.

Start with the number the board already cares about. Then work backwards:

• The metric. What are we trying to change at board level?
• Why now. What business pressure makes it urgent? (If the answer is “L&D thinks it’s important”, keep looking)
• The behaviour. What do people need to do differently day to day for that number to move?
• The skill gap. What capability is stopping that behaviour?
• The intervention. Only now do you design the learning, and you judge it against the metric you started with

This order does something a framework can’t. It gives every piece of learning a reason to exist, and it gives you a result you can attribute.

It also tends to produce interventions that look small. Phill made the point that, on paper, the programme in the case study looked narrow and unimpressive. That’s a feature, not a flaw. A tight fix to one real gap is easier to launch, easier to measure and far easier to defend than a sprawling programme that touches everyone and moves nothing.

The principle underneath it all: outcomes, not activity.

3. Is your AI busy or useful?

Right now, most of L&D’s AI conversation is about content: generating courses, writing descriptions, authoring faster. That’s useful, but it’s also rapidly becoming a commodity. Free, general-purpose tools already do much of it well.

Which raises Phill’s question: “Why do you want your LMS provider spending six months of roadmap building what free tools already do?”

The value isn’t in generating more material. It’s in context. AI becomes genuinely useful when it can connect what it knows about a person (their skills, development, performance and goals) to what the organisation is trying to achieve. A free tool can’t do that, because it doesn’t have the data.

Take end-of-year reviews. Today, managers spend hours preparing, and the conversations often end up generic. Justin pointed to the hidden cost: lost time, uneven quality, and employees who can tell the conversation wasn’t really about them. Now imagine preparation taking minutes, with prompts grounded in that person’s actual progress and objectives. Same manager, better conversation.

Try this: when a vendor shows you an AI feature, ask “what does this know about my people and my objectives that a free tool doesn’t?” If the answer is nothing, it’s activity.

4. Decide how you’ll prove it’s working before you launch

Headline ROI takes months. Budgets and executive patience often don’t last that long. If you have nothing credible to show in the meantime, the programme gets judged on gut feel, or quietly cut.

The answer is leading indicators: signals that appear in weeks, not months. But they only work if they sit on the path between learning and the metric. If you started with the metric, then you’ve already mapped that path. Now put a signal on each step.

First, capability. Are people actually picking up the skill? Look at how quickly they engage and whether they come back, not just whether they finish.

Next, behaviour. Is anyone doing anything differently? This is where managers matter most, because they see the change in how work gets done long before it shows up in a report.

Then a proxy metric. Find an operational measure closely tied to the headline number that tends to move first. It’s the bridge that lets you say “this is heading the right way” with evidence rather than optimism.

Finally, the business metric itself: the number you started with, and the one the board will ultimately judge you on.

Phill drew a clear line between this and vanity data. Logins and completions tell you something happened. A good leading indicator tells you something is changing. It also keeps three ideas apart that often get blurred: measurement (did learning happen?), impact (did behaviour change?) and ROI (did it deliver business value?).

The strategic benefit is course correction. Spot a weak signal in week three and you can adjust. Spot it in month nine and you’ve spent the whole budget finding out.

5. When execs ask for a framework, they’re really asking for confidence

Any of this is easy to agree with in a conference room. The real test is the boardroom, where leaders often ask for exactly the thing we’ve been arguing against: the framework, the mapping, the dashboard.

Here’s the reframe. That request is rarely about the framework itself. It’s a proxy for something else: “show me we’re in control” or “prove this spend is worth it.” A framework is just the most familiar way they’ve seen that need met.

Pushing back still carries risk. Justin was candid that telling senior leaders “I don’t think that’s what we need” isn’t a comfortable conversation.

What changes the conversation is offering something better than the thing they asked for. A metric they already care about, and leading indicators that show progress long before the headline number moves. That’s more visibility, not less.

The same pressure exists on the product side. Phill described the constant pull to add more skills data points. The better filter, every time, is one question: does this move a number?

Try this: next time you’re asked for a framework, ask “what decision would it help you make?” Then solve for that decision.

The one question worth taking back to your desk

There’s no takeaways slide here, and that’s deliberate. A tidy list of principles would just be another framework.

Instead, it comes down to one question. You don’t need a perfect map of every skill in your organisation. You need to know the one thing that would actually move the needle right now. Do you know what that is?

If you do, you have permission to stop mapping and start measuring. If you don’t, that’s the place to start, not with another spreadsheet of competencies.

That’s the way out of The Activity Trap: the cycle of more frameworks, more content and more completions that keeps L&D busy but leaves the business asking “so what?” Busy isn’t the same as effective, and activity data can’t prove value on its own. Breaking that cycle doesn’t take a bigger framework or a smarter content engine. It takes the discipline to pick one outcome, measure what changes, and let everything else wait.

If you’re ready to move beyond measuring learning activity and start proving business impact, So What? is the newsletter for you. Created for L&D, HR, and people leaders, this newsletter brings together practical ideas, fresh research, and expert perspectives to help you connect learning with the outcomes that matter most. Subscribe today and join thousands of people leaders committed to measuring what matters.