Skills Matrix: How to Build One That Isn't Guesswork (2026)

Most organisations already have a skills matrix somewhere. It is usually a spreadsheet, it was built with real enthusiasm about eighteen months ago, and nobody has opened it since. The reason is rarely the format. It is that the numbers inside it were never trustworthy, so nobody was willing to make a decision on them.

That is a shame, because a skills matrix is one of the few management artefacts that earns its keep. It is a grid (people down one side, skills across the top, a proficiency rating in each cell), and when the ratings are sound, it answers two questions at a glance that most leaders otherwise answer by feel: what is this team actually able to do, and what happens if a particular person leaves.

This guide covers what a skills matrix is, how it differs from the terms it gets muddled with, exactly what belongs in the template, how to build one step by step, and (the part most guides skip) how to keep the ratings honest enough to spend money on.

Why the Skills Matrix Matters More in 2026

Precision about skills has just acquired a price tag.

The Growth and Skills Levy is now the mechanism through which levy-paying employers fund training in England, and the Department for Education has confirmed a set of changes that bite from 1 August 2026. Three of them matter here:

• Levy funds get more perishable. New funds entering an employer's apprenticeship service account from 1 August 2026 will expire after 12 months, rather than the 24 months that funds entering on or before 31 July 2026 still enjoy.
• The 10% government top-up stops applying to new funds from that date.
• Co-investment gets significantly more expensive. When levy funds run out, the employer contribution rises from 5% to 25% for apprenticeship and apprenticeship unit starts from 1 August 2026, with the government covering the remaining 75%.

At the same time, the levy has become more flexible. Apprenticeship units (short courses of 30 to 140 delivery hours, run over one to sixteen weeks, for employed staff aged 19 and over) can now be funded from levy accounts. Three of the first ten units cover AI leadership (AI strategy and opportunity; AI adoption, procurement and governance; AI delivery and organisational transformation); the other seven are hands-on technical trades: battery manufacturing, EV charging point installation, electrical fitting, mechanical fitting, permanent modular building assembly, solar photovoltaic installation and mechanised welding.

Put those together and the strategic picture is clear. Employers are getting more ways to spend levy money, less time to spend it, and a bigger bill if they get it wrong. The question "which skills, for which people, first?" is no longer an HR nicety: it is a budgeting decision with a deadline attached. A skills matrix is the artefact that answers it, and a matrix built on guesswork is about to become an expensive thing to own.

What Is a Skills Matrix?

A skills matrix is a grid that maps people against skills, with a proficiency rating in each cell. That is genuinely all it is. Its power comes not from the format but from the fact that it can be read in two directions.

Read down a column and you are looking at organisational risk. How many people hold this skill? At what level? If the column for "manages the reporting dashboard" has one person rated 4 and everyone else on 0 or 1, you have found a single point of failure, and you found it before that person resigned, went on leave, or was hit by the proverbial bus.

Read across a row and you are looking at an individual. Where is this person strong, where are they thin, and what is the most sensible next step in their development? A row is the beginning of an honest development conversation, because it is specific.

That dual reading is why the matrix survives while more elaborate talent tools come and go. It puts team-wide risk and individual development on the same page.

A note on the "will/skill matrix"

If you search for skills matrices you will quickly hit the will/skill matrix, which is an entirely different thing: a coaching model that plots someone's motivation against their capability to decide whether to direct, guide, excite or delegate. It is a useful management tool, but it is not what this guide is about. When someone asks for "the skills matrix", they nearly always mean the people-against-skills grid.

Skills Matrix, Skills Audit, Gap Analysis and Competency Matrix

These four terms get used interchangeably and they should not be. Knowing which one you are actually doing keeps the work focused.

Term	What it is	Its job
Skills audit	The whole exercise: defining what skills the organisation needs and inventorying what it has	The process
Skills gap analysis	The comparison between required level and current level, ranked by priority	The maths
Skills matrix	The grid of people against skills, with a rating in each cell	The picture
Competency matrix	Broadly the same grid, but rating competencies: behaviours and applied capability, not just discrete skills	The picture, drawn wider

In practice, a skills audit is the umbrella process; the matrix is what you produce partway through it, and the gap analysis is the interpretation you do on top. If you have been asked for "a skills matrix" and you produce a grid with no required levels and no conclusions, you have produced a decoration. The grid is the input to a decision, not the decision.

The competency matrix distinction is worth pausing on. A skill is usually something discrete and demonstrable: can this person build a pivot table, write a SQL query, use our CRM. A competency is broader and usually behavioural: does this person exercise sound commercial judgement, coach others well. Both can live in a grid. Skills are far easier to measure objectively; competencies almost always require structured evidence and human judgement. Many organisations run one matrix with both, and simply accept that the two halves are rated by different methods. That is fine, as long as you are honest about which is which.

What Belongs in a Skills Matrix Template

A good template is boringly simple. Complexity is what kills these things.

The rows are people (or, for a larger organisation, roles with a headcount attached). Keep one matrix per team. A single matrix covering four hundred people across nine departments will be abandoned within a quarter.

The columns are skills. Be ruthless here. A column only earns its place if a gap in it would change a decision, and in most teams that is ten to fifteen skills, not forty. Each one needs a one-line definition. "Excel" is not a skill; "Excel: can build and maintain a model using lookups and pivot tables" is a skill, because two different people will rate it the same way.

The cells hold the current proficiency rating.

Beyond the grid itself, a template that is actually useful also carries:

• A required level for each skill, per role. Without this, the matrix tells you what people can do but not whether it is enough. This single column is the difference between a picture and a plan.
• A criticality rating for each skill. How much does this skill matter to the work? A big gap in a skill nobody needs is not a problem.
• The gap (required level minus current level), calculated, not eyeballed.
• A source or date column for each rating. This is the column nobody includes and everybody wishes they had. It records how a rating was arrived at (assessment, manager review, self-rating) and when. It is what allows the next person to know whether to trust the number.
• Planned action, owner and review date for the gaps you decide to close.

That is the whole template. If you want the grid to sit inside a wider process (objectives, prioritisation, costed plan), that is the skills audit, and it is worth doing properly.

How to Create a Skills Matrix: Step by Step

Step 1: Set the scope

One team, one purpose. Decide what decision this matrix is going to inform (succession risk, a training budget, readiness for a new system, levy spend before the funds expire), because the purpose determines which skills belong in the columns. A matrix built for no particular decision will be built for no particular use.

Step 2: Define the skills, and define them properly

List the skills the work actually requires, not the skills in the job description from 2019. Write a one-line, observable definition for each. Then set a required level for each skill in each role. Do this before you look at any people. Otherwise you will unconsciously set the bar wherever your current team happens to sit.

Step 3: Choose a rating scale and publish what each number means

Most teams use 1–5. The scale is not the hard part; defining the levels is. Something like:

• 0: None. No exposure.
• 1: Awareness. Understands the concept; cannot yet do it unaided.
• 2: Working with support. Can do it with guidance or checking.
• 3: Independent. Does it competently, unsupervised. This is usually "good enough" for most roles.
• 4: Advanced. Handles complex or non-standard cases; others come to them.
• 5: Expert. Sets the standard, coaches others, improves how it is done.

Write those definitions at the top of the sheet. An unlabelled 1–5 scale means five people will use five different scales, and your matrix becomes a collection of opinions in a trench coat.

Step 4: Measure the current level, and mind where the numbers come from

This is the step that decides whether the whole exercise is worth anything, and it is covered properly in the next section. In short: measure what can be measured, and use structured evidence for what cannot.

Step 5: Fill the grid, then read it in both directions

Populate the cells, then actually interrogate the result. Down the columns: which skills are thin across the team, and where is a critical skill held by only one person? Across the rows: who is close to the required level and just needs a nudge, and who has a genuinely large gap?

The pattern to hunt for is the wide, shallow gap: a modest deficit shared by most of the team. It rarely feels urgent, because no individual looks like a problem. But it is usually the cheapest thing on the board to fix, and exactly the shape of gap that one well-chosen short course can close for everybody at once.

Step 6: Turn it into action, then re-measure

Assign an owner, an action and a date to the gaps worth closing. Then (and this is the step almost everyone skips) re-measure afterwards using the same method. A matrix that is only ever filled in once is a snapshot. A matrix that is re-measured is evidence that your training money worked, which is the argument you will need next time you ask for a budget.

The Honesty Problem: Why Most Skills Matrices Are Fiction

Here is the uncomfortable part. In my experience most skills matrices are populated by asking people to rate themselves, or by asking a manager to rate them from memory. Both methods are fast, free, and unreliable.

Self-assessment tends to drift from reality in both directions. People over-rate skills they use every day but have never had checked against a standard: familiarity feels like competence. And they under-rate skills they have never been formally tested on, even when they can plainly do the job. Manager ratings carry their own baggage: recency, likeability, and the simple fact that a manager cannot personally observe most of what their team does.

The result is a grid full of confident-looking 3s that means very little. That was survivable when the matrix was a filing exercise. It is much less survivable when you are deciding how to spend levy money that, from August, will start arriving with a twelve-month shelf life and a 25% co-investment bill once it runs dry.

The fix is not to abandon the matrix. It is to be deliberate about where each number comes from:

• Where a skill can be objectively assessed, assess it. Digital skills, English, maths and AI readiness can all be measured directly against a standard (the Essential Digital Skills framework is the obvious reference point for the digital column), and a measured level is something you can defend in a budget meeting. An opinion is not.
• Where a skill cannot be tested directly, use structured evidence: a work sample, a recent piece of delivery, a competency-based review against your published level definitions. Slower, but far better than a number pulled from the air.
• Record which method you used, in that source column. A matrix where half the cells are measured and half are estimated is perfectly workable, as long as everyone reading it knows which half is which.

You do not need to assess everything. You need to assess the skills that are critical and expensive to get wrong, and be transparent about the rest.

Common Mistakes

Too many columns. Forty skills is a matrix nobody will complete honestly. Cut it to the ten to fifteen that carry real weight.

No required level. Without it you have a picture of your team with no way to tell whether the picture is good news.

An undefined rating scale. If "3" is not written down somewhere, it means nothing.

Publishing individual ratings carelessly. A wall-mounted grid showing every person's weaknesses is the fastest route there is to a matrix everyone quietly games. Ratings are personal data about identifiable employees, and they should be handled like it. There is more on that in the FAQs below.

Building it once and never touching it. A skills matrix has a half-life. Roles change, tools change, people learn. Set a review trigger before you set a review date, and the grid stays alive.

Confusing the grid with the goal. The matrix is not the deliverable. The decision it enables is the deliverable.

Excel or Software?

Be honest about scale.

A spreadsheet is genuinely fine for a single team of up to roughly twenty or thirty people with a stable skill set. It is free, everyone can open it, and conditional formatting will colour your gaps red without any help from a vendor. Plenty of public-sector bodies publish free spreadsheet templates for exactly this purpose. If that is your situation, do not let anyone sell you a platform, including us.

Spreadsheets break at three specific points. The first is scale: cross-team roll-ups in Excel become a maintenance job in their own right. The second is refresh: re-measuring a hundred people by hand, twice, is where good intentions die. The third, and most important, is the honesty problem above. A spreadsheet will faithfully store a number; it cannot tell you whether the number is true.

That last point is the real dividing line. If your matrix is informing a genuine spending decision, the question is not "where do I keep the grid?" but "where do the ratings come from?", which is a question about your assessment method, not your software. If you are weighing up tools on that basis, our comparison of initial assessment tools sets out how the main UK options differ.

How Digital Skills Assessment Fits

Everything above is method, and you can run it with a spreadsheet and no vendor at all. But the one part of a skills matrix you cannot solve with better formatting is the honesty problem, and that is the specific thing we built Digital Skills Assessment to fix.

The idea is narrow: for the skills that can be tested, put a measured number in the cell instead of a remembered one. A learner takes an adaptive assessment (roughly twenty minutes) and comes out with a level derived from how they actually performed, not from how confident they felt. The "current level" column in your matrix then holds evidence rather than sentiment, which matters enormously the moment someone senior asks you to justify the training spend.

For a team-wide matrix, the Business plans do the aggregation. Business Basic (£69.99/month, 20 credits) includes the Insights Dashboard, the Next Steps Engine and cross-team analytics, which is, in effect, your matrix assembled for you, with a view on where to act first. Business Ultimate (£199.99/month, 100 credits) adds cheaper top-ups at £2.00 per credit and a dedicated account manager. Credits are universal, so one plan covers the digital, English, maths and AI readiness columns of your grid rather than making you buy four tools.

A sensible way to test the claim rather than take my word for it: run a free demo, or buy a couple of one-off assessments at £7.00 each (no account needed) for two people whose current matrix ratings you would happily bet on. Then compare. If the measured levels match your grid, your grid is in better shape than most and you should carry on. If they do not, you have just learned something worth considerably more than £14. Current pricing is on the pricing page.

And to be straight about the limits: for skills that cannot be tested (commercial judgement, stakeholder management, coaching), no assessment tool should pretend to have an answer, and ours does not. Measure what is measurable, evidence the rest with structured judgement, and label clearly which is which. That mixed matrix is not a compromise; it is just an honest one.

The Bottom Line

A skills matrix is a simple grid that does two jobs at once: it shows you where your organisation is exposed, and it shows each person where to grow next. Keep it small, define the levels, set a required level for every skill, and read it down the columns as well as across the rows.

But the format was never the hard part. A matrix is only as good as the numbers inside it, and most matrices are filled with guesses. From August, with new levy funds expiring in half the time and co-investment rising from 5% to 25%, those guesses acquire a price. Measure what you can, evidence what you cannot, record which is which, and the grid stops being a spreadsheet nobody opens and starts being the thing you plan the year around.

Frequently Asked Questions

What is a skills matrix?

A skills matrix is a grid that maps people against skills, with a proficiency rating in each cell. Reading down a column shows how many people hold a given skill and at what level, which reveals organisational risk and single points of failure. Reading across a row shows one person's profile and where they should develop next. It is the standard way of displaying the results of a skills audit.

How do you create a skills matrix?

Set the scope to one team and one decision; choose the ten to fifteen skills where a gap would actually change what you do, each with a one-line definition and a required level per role; publish a rating scale with every level clearly defined; measure people's current levels, using objective assessment wherever a skill can be tested; fill the grid and read it both down the columns and across the rows; then assign actions to the gaps worth closing and re-measure afterwards using the same method.

What should be included in a skills matrix template?

At minimum: people (rows), clearly defined skills (columns), a current proficiency rating in each cell, a required level for each skill per role, a criticality rating, and the calculated gap. The columns most people forget but later wish they had are a source-and-date column recording how each rating was arrived at, plus planned action, owner and review date for the gaps you intend to close.

What is the difference between a skills matrix and a competency matrix?

They are structurally the same grid. A skills matrix rates discrete, demonstrable skills: building a pivot table, writing a query, using a particular system. A competency matrix rates broader behavioural capabilities such as commercial judgement or coaching. Skills are far easier to measure objectively; competencies usually require structured evidence and human judgement. Many organisations combine both in one grid, which works well provided everyone knows which cells were measured and which were assessed by judgement.

Can you build a skills matrix in Excel?

Yes: for a single, stable team of a couple of dozen people, a spreadsheet is genuinely the right tool, and you should not let anyone talk you out of it. The three points at which spreadsheets start to hurt are rolling up across multiple teams, re-measuring everyone at refresh time, and confirming that the ratings are accurate in the first place. The third is the one that matters, and it is a question about your assessment method rather than your software.

How often should a skills matrix be updated?

Treat it as a living document with three triggers rather than a fixed calendar item: refresh a person's row when they complete training or change role, refresh a column when the tools or standards behind that skill move, and run a full re-measure at least annually so the whole grid is comparable. The trap is the column that quietly goes stale: a rating can stay accurate about the person while becoming wrong about the requirement, because the bar moved and the matrix did not.

Are skills matrix ratings personal data?

Individual skill ratings relate to identifiable employees, so they should be handled with the same care as other HR data: be clear about why you are collecting them, collect only what you need, tell people how the information will be used, and think carefully about who can see individual-level ratings. Aggregated, team-level views are usually sufficient for a wider audience. This is general guidance rather than legal advice: if you are unsure, check your approach with whoever handles data protection in your organisation.