The Real Skills Gap Isn't the One You Think

 Photo by Farnoosh Abdollahi on Unsplash 

The skills gap you can’t see is the one in your data

Every executive team I talk to is asking the same question right now: does a skills gap cap our growth? Can we staff the work that matters, move people where the business is going, and compete in a market AI is reshaping faster than our org charts can keep up? 

Good question. Wrong place to start. 

Before you can say whether your workforce has the skills your business needs, you have to trust what your systems tell you about the skills it already has. Most organizations can’t. And that’s the actual problem, not the AI tool everyone’s testing. 

Your AI didn't mislead you, your data did

Candidate matching, internal mobility, career pathing, workforce planning — every one of these sits on top of two things: your job catalog and your skills data. The engine doesn’t know your business. It knows your data. Feed it good data and it’s a competitive edge. Feed it garbage and it hands you garbage back with the same confidence either way. A mobility tool matching against empty skill profiles isn’t malfunctioning. It’s doing exactly what you asked, with exactly what you gave it. 

So when the recommendations are bad, the instinct is to question the configuration, the application, the model. In my experience that’s almost never it. The foundation is the problem, and nobody wants to hear that, because the foundation is boring and the AI is not. 

Your Job Catalog may be worse than you think

Job Catalogs don’t collapse. They rot, slowly, through reorgs, acquisitions, hiring surges, and years of someone in HR making a “quick fix” that never got reconciled. The pattern is the same everywhere I look:

• The same job sitting under three titles across two families at different levels

• Profiles with no skills attached, or skills tagged once during implementation and abandoned

• Job Families built for the company you were five years ago

At a global medical technology company, one that considered itself ahead on skills, we found that 93.5% of workers had no skills listed at all. A talent leader at a global biotech told me her recruiters check LinkedIn to figure out what their own employees can do, because the system of record is useless for that question. Nobody planned for this. It’s just what happens when you treat data as a launch task instead of an asset you maintain. 

Which leaves you somewhere uncomfortable: you can’t make a credible skills-gap claim to your board, and you can’t trust what the AI tells you, on top of data nobody’s actually checked. 

Pick your stage, not your ambition

You don’t need more analysis. You need a next move — one you can stand behind.

• Stabilize: The Job Catalog itself is broken. Fix that first. Nothing downstream works until you do.

• Optimize: The foundation holds, the tool is licensed, and it’s sitting there while skeptics block the budget. Don’t leap. Run a contained proof — one or two Job Families, real data, a go/no-go inside weeks.

• Modernize: You actually want to run as a skills-based org. That’s not a data fix. That’s taxonomy, governance, configuration, adoption — built on purpose, not assumed into being.

The expensive mistake is skipping a stage. Fund a modernization program on a broken catalog and you’ve just bought yourself a very costly way to rediscover your data problem. 

Three questions for this quarter

1. What percentage of your workforce has current, validated skills data? If nobody can answer with a number, that’s your answer.

2. Would your Job Catalog survive a compensation-equity review? Bad architecture doesn’t just wreck AI — it’s a liability.

3. If your board asked for skills supply versus demand across your three most critical functions tomorrow, how long would it take — and would you actually stand behind the number? A number you can't defend isn't an answer, it's an exposure.

The real gap isn’t between your workforce and the market. It’s between what your systems claim and what’s true. That one’s fixable in weeks, not years. 

You can’t manage what you can’t see. Start there.