xAPI

The skills crisis is a measurement crisis

Skills England says the pipeline alone won't meet employers' needs. It's right. But more training won't fix it either — not until we can see what our training actually builds.

The Skills England Annual Skills Report landed this week, and the headlines have gone where headlines go: 1.8 million more jobs needed by 2035, a youth unemployment crisis, whole sectors — clean energy, advanced manufacturing, social care, defence — short of the people they need. All true. All urgent.

But the number I can’t stop thinking about is a quieter one. Business investment in training has fallen 19% in real terms since 2011 — from £55.4 billion to £44.8 billion — even as the demand for skills has climbed. Employers are being asked to reskill more people, faster, with less. And the report is unambiguous about why it matters: “the education pipeline alone will be insufficient to meet employers’ needs.” The people you’ll need in 2030 are, for the most part, already on your payroll. You will have to grow them.

Here’s the part almost no one is saying out loud.

We can’t tell whether the training we already do works.

The report calls, rightly, for employers to demonstrate the return on their training investment. It’s a fair ask. It’s also, for most organisations, an impossible one — because we don’t measure learning, we measure attendance. Course completions. Hours logged. Modules ticked. None of which tells you whether a single person can now do something they couldn’t do last week. We are spending £44.8 billion a year on a black box, and then wondering why we can’t prove it pays off.

That’s the real crisis under the skills crisis.

It isn’t only that we invest too little. It’s that we invest blind.

And this is where I want to gently push back on the instinct the report will trigger, which is: spend more, buy more courses, push more content at more people. Because the learning science has been clear for half a century that this doesn’t work. Benjamin Bloom showed in 1984 that a learner given one-to-one, mastery-based tutoring performs two standard deviations better than one in a standard class — the gap between a middling student and a top one. Ebbinghaus mapped the forgetting curve a century before that. Every serious study since has said the same thing: people learn by doing, in a way that’s tuned to them, with feedback — not by consuming standardised material at scale. Scale and standardisation are precisely the enemies of how humans actually learn.

So pouring a shrinking budget into more generic e-learning isn’t doing more with less. It’s doing more of the thing that was already failing, and hoping volume fixes it.

What would actually move the needle isn’t complicated to name, even if it’s been hard to deliver:

Personalised. Meet each person where they actually are, not where the average learner is. This is Bloom’s two-sigma — and for the first time, AI makes it deliverable at a scale no human tutor ever could.

Active. Learn by doing and demonstrating, not by watching and clicking. Capability is built in the doing.

Evidenced. Capture what someone can now do — person by person, skill by skill — not whether they finished. Because the moment you can see capability, everything changes. You can point a reduced budget at where it genuinely moves people. You can spot the gaps before they become vacancies. And — to the report’s point — you can finally answer the return-on-investment question honestly, because the evidence is in front of you.

This is the reframe I’d offer any L&D leader reading the Skills England report this week: the skills gap you’re being asked to close is, first, a gap you cannot currently see. You can’t manage what you can’t measure. Before we spend another pound, we should be able to answer one simple question — is this person more capable than they were? — and most of us can’t.

The report is a genuine wake-up call, and I’m glad it exists. But the response it deserves isn’t “train more.” It’s “train in the way people actually learn, and finally measure the thing that matters.” Do that, and the 1.8-million-person mountain looks less like a hiring problem we’ve already lost, and more like a growing problem we can solve — mostly with the people we already have.

The pipeline was never going to save us. It doesn’t have to. We just have to get serious about capability: seeing it, building it, and proving it.