Skills gaps are now one of the biggest constraints on business transformation. But before an organization can close a gap, it needs a credible answer to a more basic question: which skills does each role actually require?
The World Economic Forum’s Future of Jobs Report 2025 found that 39% of workers’ core skills are expected to change by 2030, while 63% of employers identify skills gaps as the biggest barrier to transformation.
Those figures point to the same operational challenge. A year-long spreadsheet exercise, a 20-second query in a general-purpose AI tool, and a weekend spent prompting an LLM will each produce a different skills-to-role map. Yet none is automatically credible enough to withstand scrutiny from a promotion committee, workforce planning team, works council, or auditor.
This article explains what credible skills-to-role mapping requires in 2026, where the three most common approaches fall short, and six questions HR leaders can use to assess their current architecture.
In our conversations with HR leaders, three methods come up repeatedly:
Each approach solves part of the problem. But when an HR leader needs to defend the resulting architecture, the weaknesses quickly become visible.
The spreadsheet approach is the oldest of the three. HR teams run workshops with subject matter experts, gather job descriptions, competency frameworks, and leveling documents, then map roles one by one across a business unit.
“They went through this mapping exercise in Excel, and then they went through one business unit. It took them 18 months, and now all of the roles are outdated already, because it took them 18 months to map it.”
— Christine du Plessis, Head of Value and Skills Strategy at Fuel50
Speed is the central limitation, and it cuts two ways:
When different teams map skills in isolation, a second problem appears. Each business unit creates its own version of concepts such as problem solving, communication, and leadership. The organization is left with a fragmented skills vocabulary that must be reconciled before it can support decisions across functions.
Generic AI solves the speed problem but introduces a context problem. A skills list generated from a job title alone cannot reliably reflect the organization’s strategy, structure, or the difference between someone entering a profession and someone leading the function.
“AI output flattens exactly the distinctions the mapping is supposed to reflect. A senior software engineer and a junior software engineer receive the same twenty skills because the AI is matching on the title rather than the level of the work.”
— Christine du Plessis
The output may look credible at first glance. It becomes difficult to defend as soon as someone asks why a particular skill appears on a particular role profile and the only explanation is, “the model returned it.”
The newest approach is a do-it-yourself LLM workflow. A CHRO, People Analytics team, or technically fluent HR business partner feeds job descriptions into an LLM, prompts it to generate skills and proficiency levels, and produces role profiles within days.
The result can look impressive on a slide. Under governance scrutiny, however, six risks emerge:
Any one of these issues is inconvenient. Together, they create an enterprise governance problem that cannot be solved at prompt time.
A role profile that can withstand a workforce planning review, promotion decision, or enterprise audit must do five things well:
These requirements are interdependent. A profile that reflects strategy but ignores level is unreliable in a promotion discussion. A profile that distinguishes level but drifts from strategy is precise in the wrong direction. Credibility comes from holding all five together.
All three approaches can produce something that looks like a skills list. What turns that list into a usable skills architecture is governance. Three properties matter most.
Every credible architecture rests on one shared skills language. Fuel50’s Skills Ontology is an expert-curated, bias-reviewed, and continuously maintained library of skills, with descriptions, proficiency levels, and development actions built in.
Organizations typically use it in one of three ways:
Whichever path they choose, the objective is the same: one consistent language that makes role profiles comparable across the organization.
An auditor, works council, promotion committee, or legal team should be able to ask why a specific skill sits on a specific role profile and receive an answer that traces back to the business.
“Strategy always—or structure always—follows strategy. So you’ve got a strategy, you’ve got a structure, and your structure informs the skills and capabilities needed to deliver that.”
— Christine du Plessis
This echoes the argument Alfred Chandler set out in Strategy and Structure: organizational design follows strategy. By extension, the capabilities and skills required by each role should follow the work the strategy demands.
Fuel50’s AI is bias-tested and explainable by design, so the reasoning behind matches, recommendations, and role suggestions can be surfaced to the people affected by them. That traceability turns a skills architecture from a black box into something enterprise governance teams can assess and approve.
Strategies shift. Business units restructure. New skills enter the market and older skills decline in relevance. A governed architecture separates assumptions shared across roles from details specific to an individual role. When a shared assumption changes, the update can flow through every profile that inherits it.
Without that structure, every change becomes another rebuild. An organization that repeats a mapping exercise every 18 months is not managing a skills architecture; it is running the same one-time project on a loop.
Use these questions to test whether your current approach is credible, useful, and ready for enterprise scrutiny:
If your answers are inconsistent, the problem usually sits beneath the individual profiles—in the design and governance of the architecture itself.
Every downstream skills decision begins with a clear view of what the business needs from its workforce.
“If you don’t understand your demand, it doesn’t matter what your supply is.”
— Christine du Plessis
Employee skills data describes what the workforce currently offers. Role profiles describe what the strategy requires. Every meaningful comparison between the two—whether for gap analysis, redeployment, succession, or internal mobility—depends on defining that demand with enough rigor to make the comparison useful.
The organizations that make credible skills-based decisions over the next three years will treat role profiles as the foundation of talent strategy, define them against business priorities, and hold the architecture to a standard that can withstand enterprise scrutiny.
Want to see how Fuel50 applies these principles in practice? Join our upcoming Skills-to-Roles Mapping Framework webinar.