How to Measure a Skills Gap (Most Companies Don’t)
A practical guide to measuring skills deficits and surpluses when AI keeps moving the target.
"Close the skills gap" is one of the most repeated phrases in corporate learning, and one of the least measured. Most companies can tell you they have a gap. Very few can tell you, for a given role, exactly which skills are short, by how much, and for whom. You cannot close a gap you have never actually measured; you can only spend money in its general direction. And most companies never measure it: only 18 percent of organizations regularly measure their workforce's skills, according to Skillsoft's 2025 Global Skills Intelligence Survey of 1,000 HR and L&D professionals (Skillsoft).
AI makes this both harder and more urgent. The World Economic Forum estimates that 39 percent of workers' core skills will change or become outdated between 2025 and 2030 (WEF Future of Jobs Report 2025). When the definition of a role keeps moving, a skills inventory you took last year is already partly fiction. So measuring the gap cannot be a one-time audit. It has to be something you can redo cheaply as the target moves.
How to measure a skills gap, step by step
Here is a practical way to do it.
Step 1: Define the skills the role requires
Start with the role, not the person. Write down the five or six skills the role genuinely needs to be done well, not the job-description boilerplate, the real work.
This is usually called a skills taxonomy or a role-skills matrix, and it does not need to be elaborate. The trick is to pair each skill with what "good" looks like and how you would measure it. For an AI workflow lead, a simple version might look like this:
Step 2: Measure the current state, both ways
Now assess where your people stand against that list. Two findings matter, and most companies only look for one.
Deficits are the obvious target: skills the role needs that the person does not yet have. That is your reskilling list.
Surpluses matter just as much and get ignored. A person who is over-skilled for their current role is both a flight risk and your best internal candidate for a bigger one. In an AI transition, surpluses are also where you find the people who have quietly taught themselves the new tools on their own time. A gap analysis that only hunts for weakness misses half the point, and usually the more valuable half.
How you measure matters too. Self-assessment alone is unreliable, and the numbers bear it out: 91 percent of HR professionals believe employees overstate their own proficiency, especially in leadership, AI and technical skills (Skillsoft). Manager ratings drift and need calibration across teams. The most trustworthy signal is demonstrated skill: an assessment or a work-sample project where the person actually does a version of the task, scored against a clear rubric.
Step 3: Don't confuse activity with ability
Here is the trap that undoes most skills measurement. Companies reach for the number that is easy to collect, training hours completed or courses finished, and treat it as a proxy for skill. Completion is in fact the single most common L&D metric, used by 54 percent of teams (eLearning Industry). But completion measures attendance. It tells you a person was present, not that they can do anything new. You can finish a course and learn nothing, and most dashboards cannot tell the difference. What you actually want is mastery: can this person demonstrate the skill on a task that resembles the real one. (We go deep on that distinction in course completion vs. mastery.)
Once you can see the gaps, close the ones tied to revenue or risk first, not simply the biggest. A small gap in a critical role usually costs more than a large gap on the periphery.
The honest catch, and how it changes
There is a reason most companies stop at completion rates, and it is not laziness. Real measurement is expensive. Building assessments, designing work-sample projects and grading them against rubrics takes time and expertise that a completion percentage does not. When the choice is a free number you already have versus a costly program you have to build, the free number wins, and the gap goes unmeasured.
That math is what changes. After you have done your own skills-gap analysis, Honen can measure mastery at the level of the individual and the cohort. It generates assessments and rubric-graded projects from your own standards and materials, in minutes, and reports not just who finished but who actually demonstrated the skill, flagging the people who need another pass. The expensive part of real measurement becomes cheap enough to repeat as fast as AI keeps redrawing the role.
The bottom line
Skills-gap analysis is not a one-time spreadsheet exercise, and in an AI economy it especially cannot be. It is the instrument panel for everything else: who to reskill, into what, and whether it worked. Counting hours and trusting completion will confidently fund training that changes nothing while the roles move out from under you. Naming the real skills and measuring demonstrated mastery gives every dollar a target and a scoreboard.
Measure mastery, not minutes. Then you will know whether the gap actually closed, instead of hoping.
Frequently asked questions
How do you measure a skills gap? Define the five or six skills a role actually requires, assess each person against that list using demonstrated work samples rather than self-assessment, and look for both deficits and surpluses. Then track mastery, whether the person can perform the real task, not just course completion.
What is the difference between a skills deficit and a surplus? A deficit is a skill the role needs that the person does not yet have, which is your reskilling target. A surplus is a skill the person has beyond their current role, which marks them as a flight risk and a strong internal candidate for a bigger one.
Why isn't course completion a good measure of a skills gap? Because completion only records attendance. It is the most common L&D metric, but it tells you someone finished a course, not that they can do anything new. Demonstrated mastery on a realistic task is the measure that predicts performance.
How often should you measure a skills gap? Continuously, or at least far more often than once a year. The WEF estimates 39 percent of core skills will change by 2030, so a one-time audit is out of date almost immediately.
Want to see whether your people have actually closed a gap, not just finished a course?
By StudyFetch Staff. Honen turns the materials your team already has into real courses they'll actually finish, built in minutes and measured by mastery.