Insights

The Framing Problem

Walk into most large companies and you will find two conversations running side by side inside the talent function, and they almost never meet. In one room, the skills team is building a taxonomy — defining capabilities, mapping them to roles, laying the groundwork for a skills-based strategy. In another room, the leadership development team is designing programs — curriculum, competency models, who gets into which cohort.

I have spent most of my career in those rooms, and what strikes me is how little they talk. The skills team thinks in data and architecture. The leadership team thinks in behavior and learning. They share a few words at the edges. They almost never share a single piece of data.

That split is not an accident. It comes from an old assumption about how talent functions are organized: that managing skills and developing leaders are two different jobs that happen to sit under the same roof. The assumption was always a little wasteful. Right now it is doing real damage.

What the Skills Crisis Is Actually About

The skills conversation has gotten loud since 2022, and for good reason. AI is changing what work is faster than job descriptions can keep up, so gaps open quietly and stay invisible until they hurt. If you cannot see what your people can actually do, close to real time, you cannot make good calls about them.

The numbers back up the urgency. The World Economic Forum’s Future of Jobs 2025 report projects that 39% of existing skill sets will be transformed or outdated by 2030, and names skills gaps as the top barrier to transformation for 63% of employers.¹ Lightcast’s read of job-posting data found that 32% of the skills in the average job have already changed in just three years.¹

Mercer’s 2025/2026 Skills Snapshot Survey, one of the more thorough looks at enterprise skills maturity out there, found that 38% of organizations now keep a single enterprise-wide skills library, up from 30% in 2023, and 55% map skills directly to jobs, up from 47%. Even so, the average share of jobs with skills mapped sits at 72%.² That leaves more than a quarter of roles unmapped — inside companies that have already put real money into skills work.

Gartner’s October 2025 analysis put skills gaps near the top of the priority list and flagged four talent-management trends for 2026, starting with the fast decline in entry-level roles and the pressure that puts on companies to rebuild how early-career talent gets developed.³

A separate Gartner survey of 190 HR leaders in June 2024 found that 41% agree their workforce lacks the skills it needs, and 62% agree that uncertainty about future skills is a serious business risk.⁹ And even where the skills work is underway, the foundation is thinner than it looks: the same Gartner research found that only 8% of organizations have reliable data on the skills their people currently hold and on the capabilities that matter most to the business.⁹

All of that money and attention is going somewhere real. But there is a blind spot underneath it that almost nobody names. The capabilities walking out the door as the management layer gets cut were never in the taxonomy to begin with. They were never defined. Nobody is tracking them. And a system built to show you skills gaps cannot show you a gap in something it never mapped.

The Skills That Left With the Managers

When a company cuts a layer of management, the talk is all about the obvious work — coordination, reporting, oversight, moving information up and down. That work is visible. It is the work automation can pick up, and the work that shows up in the efficiency math.

What never shows up in the efficiency math, and never shows up in any taxonomy, is the developing that layer was quietly doing: seeing potential in someone junior and naming it before they can see it themselves; knowing which stretch assignment will actually grow a particular person; knowing when someone is ready and when they need more time; giving feedback in a way that builds someone up instead of knocking them down.

Those are leadership skills, and they are also skills-architecture problems. They never got written down because the org chart was deploying them on its own, without anyone having to name them. The org chart is changing. The skills are still undefined. And no system, however good, can flag a gap in something it never mapped.

Most HR tech stacks make this worse, not better. They are still a pile of disconnected systems where assessment results never make it into a person’s profile as usable data, and where slapping a skills tag on a record gets mistaken for building a skills architecture.⁴ Leadership is no different. Parking competency ratings in a system does nothing for succession unless someone deliberately wired that data into the succession process.

What Succession Planning Cannot See

This shows up most sharply in succession planning.

Skills-based succession is a real step up from the old way, which ran on relationships and gut. When you can see who has the skills next to a critical role, you make faster, cleaner calls and cut out a lot of the politics.

Gartner found that 75% of talent-management leaders have overhauled their succession process in the last 18 months, and almost half still think their technology is behind their strategy.⁵ That technology gap is really an architecture gap. These systems have gotten good at telling you who is next to a role. They are still blind to whether that person is ready to actually do it.

Here is the distinction that matters. Adjacency is about skills — who is close. Formation is about history — what someone has actually been through. A taxonomy can answer the first. It cannot answer the second, unless you have tied it to a leadership development system that tracks more than ratings: the experiences that built the person.

Two succession scenarios show the difference.

The gap between those two outcomes is not about technology. It is a decision about architecture. Scenario B only works if leadership development data — what someone has done, not just what they score on an assessment — gets captured, structured, and wired into succession. Most companies have never built that wiring.

The Investment Gap

Look at where the money goes and the disconnect gets concrete. Mercer found that 91% of companies see AI reshaping their workforce and are planning around it, and skills-architecture spending is climbing right along with that.² The leadership development picture looks nothing like it.

DDI’s Global Leadership Forecast 2025 — more than 12,000 people across 50 countries — found that only 22% of HR teams invest in the skills leaders themselves name as their biggest gaps: setting strategy and managing change.⁶ The money is not going where leaders say they need it. The same study found 77% of CHROs are not confident in their bench for critical roles, and 40% of stressed leaders have thought about leaving their roles altogether.⁶ So companies are pouring money into skills infrastructure while the leadership pipeline thins out underneath them. Those two things are not unrelated.

This is the framing problem again. Skills spending and leadership spending sit in separate budgets, under separate owners, measured in different ways, justified in separate business cases. The thing that would make both pay off — connecting them — never gets built, because nobody owns building it. And when a round of cuts comes, the person who might have owned it is often the first one gone.

You can put a number on what that costs. EY’s 2025 Work Reimagined Survey — 15,000 employees and 1,500 employers across 29 countries — found AI can deliver up to 40% more productivity when it lands on solid talent foundations, and that only 28% of organizations are set up well enough to capture that.¹⁰

What an Integrated System Looks Like

Building these as one system is not a tech project. It is an architecture call, and it comes down to being clear about what each side does, where they connect, and what data has to pass between them. Three connections do most of the work.

None of the methods here are mysterious. Real leadership capability gets built through hard assignments, peer learning, mentorship, cross-functional work with actual stakes, and time to reflect — far more than through a classroom.⁷ But the method is only half the question. The other half is whether what someone learned in that experience lands somewhere the talent system can actually use.

The Compounding Return

The reason to connect them is not only to avoid the cost of leaving them apart. It is that the connected version produces something neither side can produce alone.

Skills architecture on its own tells you what your people can do. It cannot tell you which of them are ready to lead, or what would get them there.

Leadership development on its own grows people in whatever direction the available programs point — which may or may not be where the business actually needs them. It cannot prove its worth in terms the business respects. It produces completion rates and satisfaction scores that never connect to the decisions they were meant to inform.

Put them together and you get the thing every company is really after: not “what skills do we have” and not “how did the leadership program go,” but the harder question — do we have the right people, grown the right way, ready to lead through what is coming?

Deloitte’s 2024 Global Human Capital Trends research — 14,000 business and HR leaders across 95 countries — found that 73% say it matters to keep human capabilities moving in step with the technology, while only 9% say they are making real progress.⁸ That 64-point gap is not a motivation problem. It is an architecture problem. Companies have not built the connected system that would let leadership development be steered by skills data and judged on business outcomes. The payoff from a connected talent system does not come from skills architecture by itself, or leadership development by itself. It comes from wiring the two together.

From Diagnosis to a System You Can Build

Here is the part I want to be clear about: this is fixable. It takes intent, some transparency, the courage to work across functions that have spent years happily just being adjacent, and a real plan. Concretely, three things have to connect — leadership competencies defined and leveled inside the skills architecture; development paths built from the skills data, so growth follows what the business needs instead of whatever program happens to exist; and formation evidence captured in the HCM system, so succession can pull from the whole organization rather than one corner of it.

This is the work I do through Talent Development Innovations, and I built three programs to do it. CapabilityOS™ lays down the skills foundation and the architecture that holds the rest together. Leadership: The Human Imperative™ builds the human side of leadership — the part no system can fake and AI cannot copy. AI Fluency: Responsible Practice™ builds the responsible AI capability the augmented workplace now needs. Run together, they are a practical way to a sane, usable skills practice, built on the tech stack you already own. When that is working, the talent function stops being overhead. It starts doing the job it was always meant to do.

Conclusion

For a decade the talent function has been trying to earn its seat at the table by showing that people decisions can be made with the same rigor as money decisions. Skills architecture has been the best argument for that, and fairly so. The move to skills-based talent management is a real improvement on what came before.

But skills architecture solves the visibility problem. It does not solve the development problem. And with the management layer that used to grow leaders being stripped out, the development problem is the one that will decide what organizations can do over the next ten years.

Build skills infrastructure without leadership development and you have a diagnosis with no treatment plan. Build leadership development without skills infrastructure and you are treating without a diagnosis. Both are half-measures, and neither is enough for what companies are actually facing.

The idea is not complicated. The execution is hard, because it means tearing down a wall between two functions that have run on their own for a long time. The companies willing to do it will end up with something that works like a real advantage: a talent system that knows what its people can do, grows them in the direction the business needs, and makes the right leadership call the obvious one.

That is not a talent-management goal. It is a business goal. And it means treating the talent operating system as infrastructure, not paperwork.