Most skills gap analyses don't change anything.
It's what L&D teams already know quietly, between themselves, over coffee. You run the audit. You present the findings. The leadership team nods. Three months later, the only thing that's actually changed is the date on the spreadsheet.
A skills gap analysis is a structured exercise to identify the difference between the capabilities your workforce currently has and the capabilities your business needs them to have — now and in the near future.
It's not a competency framework (that's the inventory of what "good" looks like for each role). It's not a training needs analysis (that's the shopping list of courses). It's not capability mapping (that's the broader strategic exercise that may include skills gap analysis as one input).
It overlaps with all of them. But the specific job of a skills gap analysis is to answer one question: where, exactly, can our people not do what the business needs them to do?
That's a useful question. The problem is that most analyses stop at the answer.
LinkedIn's 2026 Talent Velocity report found that 86% of organisations say they can't clearly see their current skills, mobilise talent, or keep pace with change. That's not a skills problem. That's a visibility problem — and most skills gap exercises are designed to surface the data, then leave it stranded in a deck.
A skills gap analysis fails for one of three reasons, usually all three:
Every step in this guide is designed to avoid those three failures. Treat the analysis as the first phase of a closed loop — not the deliverable.
Almost every skills gap analysis starts with the wrong question. Teams ask: what skills do our people need? That sounds reasonable, and it produces a wishlist that grows without limit.
The right question is: what is the business trying to do, and what would have to be true of our people for that to happen?
Run this conversation with the executive who owns the strategy. Sales targets, market expansion, a new product launch, a regulatory change, a digital transformation — whichever business outcome is actually on the table this year. Write it down in one sentence: "By [date], we need to be able to [specific business capability]."
That sentence is the anchor for everything else. If a skill doesn't connect back to that sentence, it doesn't belong in this analysis.
Common mistake: Letting the analysis become "all the skills everyone could possibly need." Stay narrow. You can run a different analysis next quarter for a different outcome.
Now translate the business outcome into roles. Not every role in the organisation. The roles that need to do something different, or do it better, for the outcome to land.
For each role in scope, list the capabilities needed. Capabilities are things people can do: "Can run a discovery call that uncovers a buying problem." Courses are how they might learn to do it.
Keep the capability list short. Five to eight capabilities per role is plenty.
Where possible, anchor capabilities to existing frameworks: the SFIA framework for digital roles, the SHL competency library, or whatever your organisation already uses. Reinventing the framework is a tax you don't need to pay.
Common mistake: Confusing seniority with capability. A senior salesperson and a graduate salesperson need most of the same capabilities — they just operate at different levels. Define the capability once, then define the level required for each role.
This is where the methodology choice matters — and where most analyses get optimistic.
You have four practical options:
The mature answer is: triangulate. Use two or three methods for the capabilities that matter most. Use self-assessment alone only for low-stakes capabilities or to scope the deeper investigation.
LinkedIn's 2025 Workplace Learning Report found that the L&D teams most successful at driving career development are 13 points more likely to use internal data to track skill gaps than their peers. Most teams still rely on perception. The ones doing it best rely on evidence.
Common mistake: Running a survey, calling it "the assessment," and treating perception as data. Surveys are a starting point, not the analysis.
You now have two columns: required capability level, and current capability level. The gap is the difference.
The temptation is to sort by size of gap and start with the biggest. Resist it. Sort by business impact instead.
A small gap in a critical capability matters more than a large gap in a peripheral one. A team that's "almost good enough" at the thing the strategy depends on is a bigger risk than a team that's mediocre at something the strategy barely touches.
Score each gap on two dimensions:
The gaps you prioritise are the high-impact ones, regardless of size. The low-impact ones, even if large, can wait. Or, often, they don't need closing at all — they just need acknowledging.
Common mistake: Trying to close every gap you find. The point of a skills gap analysis is to make choices, not to surface a longer to-do list.
For each priority gap, work out what would actually close it. Sometimes that's training. Often it's not — or it's training plus something else.
The 70-20-10 model (70% on-the-job experience, 20% coaching and feedback, 10% formal learning) is imperfect but useful as a sanity check. If your entire plan is in the 10%, you've designed a course, not a capability programme.
For each gap, decide:
Then design the learning to actually do those things. A scenario-based module beats a slide deck if the capability involves judgement. A simulation beats a quiz if the capability involves decision-making under pressure. A short reinforcement push two weeks later beats a one-off course every time.
If your authoring tool can't easily build scenarios, branching, or rich interactions, that's a constraint worth naming early. The cheapest course to build is rarely the one that closes the gap.
Common mistake: Defaulting to a course because a course is what L&D usually delivers. Start with what would actually change behaviour, then work backwards to the format.
This is the step that gets skipped the most often, and it's the one that turns a skills gap analysis into a closed loop instead of a one-off exercise.
Before any learning is built or delivered, decide:
If you can't answer these four questions, you're not ready to start building learning. You're ready to argue about which course to buy.
Common mistake: Treating measurement as a post-implementation step. By the time the programme is delivered, it's too late to design the measurement — you'll be reaching for whatever data happens to exist, which is usually completion rates and a feedback survey.
If you've done steps 1–6 well, you've identified the gap, designed something to close it, and set up the measurement.
Here's what then needs to happen:
The data has to come back. You need engagement data on the learning (who actually did it, where they dropped off, what they got wrong), behavioural data on the application (are they doing the thing differently at work?), and outcome data on the business result (did the metric you set in Step 6 move?).
Someone has to look at it. Not "have access to a dashboard." Actually look at it, at the cadence you set, with a decision attached. "At three months, if the discovery-call metric hasn't moved by half the target, we redesign the reinforcement layer."
You feed it back into the next analysis. The next skills gap exercise isn't a blank page. It's an iteration on the last one. What did you learn about how your people learn? Which capabilities were harder to close than expected? Which measurement methods told you what you needed to know, and which were noise?
This is the loop that turns skills gap analysis from a strategic exercise into a strategic practice.
It's also where infrastructure matters more than methodology. If your authoring tool, your delivery platform, and your analytics are three different systems run by three different vendors, the loop will break — because someone has to manually stitch the data together every quarter, and they won't.
You don't need to use Chameleon to do this well. (We'd obviously like you to, but that's not the argument here.) You do need a stack where the gap, the learning, and the measurement live close enough together that the data actually gets used. Otherwise you're back to running an audit every two years and hoping it sticks.
A few patterns we see often enough to flag:
If you're staring down your first skills gap analysis, don't try to do everything in this guide at once. Pick one business outcome, three roles, five capabilities each, and the measurement plan from Step 6. Get the loop working on a small scope before you scale it.
You can download our skills gap analysis template — it's the same structure we use ourselves, built around the six steps above and pre-set up for the measurement plan most analyses forget.
Download the Skills Gap Analysis Template here.