A skills matrix lists the skills or tools a team needs and rates each person’s proficiency, which helps with staffing and training. A competency matrix sets out what good looks like at each level of a role, in observable behaviours, so you can use it to assess performance and decide promotions. Tech scale-ups usually need the competency matrix for reviews, and may keep a skills matrix for training.

This page gives you ready role-and-level templates for Product Owner / Product Manager, Product Designer and Sales Account Executive, and a practical guide to using them in reviews and promotions. The templates are starting points written for this site, not a validated or industry-standard framework. There is no software-engineering matrix here, because that should come from your own framework (or one you have permission to use), not from an invented one. Use this with the talent bar by seniority, the employee performance rating scales, and the guide to development plans and career conversations. The Excel and CSV files are free with your work email. Disclosure: Ash Taherian is Product Sales Executive at Revolut People, which sells performance management software. Treat product mentions as coming from an interested party.

Skills matrix vs competency matrix

What is the difference between a skills matrix and a competency matrix? A skills matrix lists the skills or tools a team needs and rates each person’s proficiency, which helps with staffing and training. A competency matrix sets out what good looks like at each level of a role, in observable behaviours, so you can use it to assess performance and decide promotions. Tech scale-ups usually need the competency matrix for reviews, and may keep a skills matrix for training.

Both are grids, and both use words like “skill” and “level”. The difference is what you do with the grid.

Skills matrix vs competency matrix (playbook distinction)
QuestionSkills matrixCompetency matrix
What sits on the rows?Tools, technologies, licences or training modules (for example “SQL”, “on-call certified”).Competencies for a role (for example “customer and problem discovery”, “negotiation and closing”).
What sits on the columns?People, or a simple proficiency scale applied to each person.Levels of the role, with a behavioural description in each cell.
What is it for?Staffing, coverage, onboarding and training gaps.Performance reviews, promotion decisions and career conversations.
How do you score someone?Can they do this thing, and at what proficiency?Do they meet the behaviours for their current level, with evidence from the period?

If you only keep a skills matrix, reviews tend to become a list of tools someone has used. If you only keep a competency matrix, you may still need a separate view of coverage for training and staffing. The rest of this page is about the competency matrix, which is the version you can tie to the talent bar and to promotions. The training-and-coverage version is covered at the end, under when you need a skills matrix instead.

Download the matrices

Three editable templates are included: Product Owner / Product Manager (7 competencies), Product Designer (7) and Sales Account Executive (8), each across four levels, with a column for evidence. Preview two Product Manager competencies below. The full matrices, an Excel workbook of all three and a CSV for each role are free with your work email. They are starting templates written for this site, so rewrite the descriptors in your own language before a live cycle. Levels map roughly to the talent bar by seniority.

Preview: Product Manager matrix (2 of 7 competencies)

Product Owner / Product Manager (preview): 2 competencies × 4 levels
CompetencyLevel 1: Developing
Roughly Junior on the talent bar
Level 2: Proficient
Roughly Mid
Level 3: Advanced
Roughly Senior
Level 4: Expert
Roughly Lead / Principal
Customer and problem discoveryInterviews, research, support tickets, usage dataJoins customer interviews and research sessions; summarises what was said accurately; writes problem statements with help.Runs discovery independently; combines interviews with usage data; writes clear problem statements that the squad agrees with before solutions are discussed.Sets the discovery plan for a product area; spots unmet needs others missed; kills weak ideas early with evidence rather than opinion.Builds a continuous discovery habit across several squads; coaches PMs on research quality; reframes the strategy when evidence shows the market has moved.
Data, metrics and experimentationProduct analytics, A/B tests, dashboardsReads the squad dashboard and explains movements with help from the analyst; defines simple success metrics for a feature.Defines the primary and guardrail metrics for every initiative; designs experiments with the analyst and reads results correctly, including inconclusive ones.Chooses the metric that best represents customer value for an area; links feature metrics to the company goal; challenges vanity metrics.Designs the metric tree for a product line; sets experimentation standards; makes high-stakes calls under uncertainty and explains the trade-off.

The full matrices (all competencies for Product Manager, Product Designer and Sales AE, 4 levels each) and the Excel and CSV files are free with your work email.

A software-engineering matrix is not included. Do not invent one from this page. If engineering is your first job family, use the steps in build your own for another role and write it with the function leader.

Get the full matrices

How to read a matrix: levels and observable behaviours

Each template uses four levels: Developing, Proficient, Advanced and Expert. Those map roughly to Junior, Mid, Senior and Lead / Principal on the talent bar by seniority. The levels themselves — the families, roles and published levels the matrix hangs off — are set out in the job architecture and levelling framework. The useful part is not the level name. It is the behaviour written in the cell, which should be something a manager can see or check.

A useful descriptor starts with a verb. “Defines success metrics for each initiative” can be observed. “Has a strong grasp of metrics” cannot. Between neighbouring levels, change one thing: scope (own work, a team, several teams), independence (with help, independently, coaching others) or impact (a feature, an area, a product line). If two levels differ only by an adjective such as “good” and “very good”, rewrite them.

Avoid frequency words on their own. “Often” and “sometimes” mean different things to different managers. Tie them to evidence instead: “in every initiative this cycle”. Keep each cell to one or two sentences. Long cells get skimmed, and skimmed cells get scored on impression.

Here is a rewrite, written for this page as an illustration, not taken from a real review. For a Sales Account Executive at Level 2, “Good at discovery” becomes “Runs discovery conversations that confirm the buyer’s problem, timeline and decision process before a proposal is sent.”

Once the cells are written that way, a manager can turn a row into a yes/no check: “Defines primary and guardrail metrics for every initiative: yes or no”. That is the scorecard approach in the performance management framework, and it is how you keep the matrix usable in a review rather than as a poster on the wall.

Using the matrix in reviews and promotions

How do you use a competency matrix in performance reviews? Score each person against the behaviours for their current level, using evidence from the period, and treat consistent evidence at the next level as the case for promotion.

Assess against the bar for the person’s level, not against teammates. A Level 2 product manager who fully meets Level 2 is on track, not “average”. Score each competency separately, with one line of evidence each: a deliverable, a decision, or a piece of feedback. Rating competencies one at a time limits the halo effect (see reducing manager rating bias).

The overall performance grade still comes from your published rating scale, not from averaging every cell into a number. Use the matrix for the skills lens of the review, next to deliverables and culture, as the framework sets out. For wording, start with the performance review examples (labelled fictional) and the review template. If you collect peer or upward input, bring themes from the 360 feedback template and check them against the same behaviours, rather than inventing a second bar.

Promotion is a higher bar than “had one good quarter”. Treat a promotion case as sustained evidence at the next level, not a single example in each cell. Gaps on the current level belong in a development plan. Low scores across a team that is also missing targets is often the skill-gap pattern, a training need rather than a hiring need. If you only open the matrix at review time, skills gaps stay invisible until it is too late; it is more useful in 1:1s, where people can see what the next level looks like.

Fictional worked example

The example below is fictional. The name, company and evidence were written for this page. They are not real employee data.

Maya Chen is a Level 2 Product Manager at Northwind Pay, a fictional company. In this cycle her manager finds evidence that she meets every Level 2 descriptor for discovery and for metrics. She also has two pieces of evidence that look like Level 3 discovery: she set the discovery plan for the onboarding area, and she killed a weak idea after interview evidence showed the problem was already solved in another flow. That is a development plan, and a reason to give her a Level 3 stretch in the next period. It is not yet a promotion case, because the Level 3 pattern has not been sustained across competencies or across cycles.

A published customer story on this site, How Cleo rebuilt its skills frameworks in 3 weeks, describes the same move from vague skills to a shared picture of excellence. As that page states it, Cleo defined and rolled out new skills frameworks across the organisation in 3 weeks, after department heads adapted simple templates from the Revolut People skills library. The story quotes Chief People Officer Matthew Maher: “The competency frameworks embedded in Revolut People are one of the main ways it helps provide clarity. We were able to define every single role in the organisation, from top to bottom, in 3 weeks.” Quote those results only as the story states them.

In Revolut People, public documentation checked on 13 November 2026 describes a competency matrix on each role, mapping skills to seniorities. Skills sit in a shared Skill Bank and can be reused on performance-review scorecards. Growth plans can cover promotion (a skills-gap view against the next seniority), development in the current role, or a change of role, and can be pulled into 1:1s. Skills analytics uses review ratings to show strengths and gaps. See the docs on skills, roles, specialisations and seniorities. For a walkthrough of how this would sit in your cycle, book a 15-minute review.

Rolling a competency matrix into reviews and promotions? Bring a draft to a 15-minute review and we will check whether the levels are observable, whether promotion evidence is clear, and how it sits next to your rating scale.

Book a 15-minute review15 minutes with me. No preparation needed. Or email me my review-cost number

Build your own for another role

How do you build a competency matrix for a new role? Start from the outcomes the role exists to deliver, pick a small set of competencies that drive those outcomes, write the mid level first, and test the wording on two real people before the cycle opens. The steps below are playbook recommendations: practical defaults to start from, not research findings.

  1. List the outcomes the role exists to deliver. Three to five, agreed with the function leader.
  2. Pick the competencies that drive those outcomes. Start from the closest template on this page and remove anything that does not apply, rather than adding more. The recommendation is 5 to 8 role-specific competencies, with company values assessed separately. Below about 5, each competency becomes so broad that it hides where someone is actually weak. Above about 8 to 10, managers start skimming and the scores cluster in the middle.
  3. Write Level 2 first. It is the level most people are at, so it anchors the scale. Then write the levels above and below by changing scope, independence or impact.
  4. Test it on two real people. Ask two managers to score the same person independently. Where their scores differ, the descriptor is probably ambiguous.
  5. Publish it before the cycle. People should be able to read the matrix before they are assessed against it. Review each role’s matrix about once a year, and do not change it mid-cycle, so everyone in a cycle is assessed against the same version.

Split ownership three ways. The function leader owns the content of each role’s skills (the CPO for product and design, the CRO or Head of Sales for sales, the CTO for engineering). The People team owns the format: the number of levels, the rating language and the calendar, so a “Level 3” means the same in sales as in product. Line managers own the assessment, with evidence each cycle. The function leader should draft updates with one or two strong senior practitioners in the role. Start with two critical job families, as the framework suggests: write their matrices, run one cycle, fix what did not work, then extend.

Related: scorecards for designers, juniors and interns · team performance diagnostic · rating scales explained · why the 9-box grid is not enough on its own.

When you need a training or skills matrix instead

A competency matrix is the wrong tool when the question is coverage, not performance. Use a skills or training matrix when you need to know who can do a specific task this week: who is licensed, who is trained on a system, who can cover a shift, who has completed a compliance module. Those grids are useful for staffing, onboarding and audit. They are a poor promotion scorecard, because “has completed the training” is not the same as “meets the behaviours for this level”.

If your immediate problem is a team missing its targets, start with the team performance diagnostic and the note on skills gaps that stay invisible, rather than writing a new matrix first. You can keep both artefacts: a competency matrix for reviews and promotions, and a shorter skills matrix for training and coverage. Do not merge them into one grid and then try to score a review from it.

FAQ

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

A skills matrix lists the skills or tools a team needs and rates each person’s proficiency, which helps with staffing and training. A competency matrix sets out what good looks like at each level of a role, in observable behaviours, so you can use it to assess performance and decide promotions. Tech scale-ups usually need the competency matrix for reviews, and may keep a skills matrix for training.

How do you use a competency matrix in performance reviews?

Score each person against the behaviours for their current level, using evidence from the period, and treat consistent evidence at the next level as the case for promotion. The overall grade still comes from your published rating scale, not from averaging the cells.

What should a competency matrix include?

A small set of role competencies (the playbook recommendation is 5 to 8), four or so levels that map to your seniority bar, a short observable behaviour in each cell, and a place for evidence. Keep company values separate. The Excel and CSV files include an evidence column.

How do you build one for a new role?

Agree the outcomes of the role with the function leader, pick the competencies that drive those outcomes, write the mid level first, test the wording with two managers on the same person, and publish the matrix before the cycle opens. The steps are in build your own for another role.

Can I use these templates for free?

Yes. Preview them on this page, and get the full matrices plus the Excel and CSV files free with your work email. Copy, edit and use them inside your company. They are starting points written for this site, not a validated or industry-standard framework.

How many levels should a matrix have?

These templates use four, which map roughly to Junior, Mid, Senior and Lead on the talent bar. Fewer levels blur real differences; many more make it hard for managers to tell neighbouring levels apart.

Should company values be in the matrix?

Keep them separate. Role skills differ by job; values apply to everyone. The framework scores deliverables, skills and culture as three separate lenses, so a strong performer with poor behaviours stays visible rather than being averaged out.

Is there a software-engineering competency matrix?

Not on this page. An engineering matrix should come from your own framework, or from one you have permission to use. Do not invent level definitions for engineering from this article. Use the steps in build your own for another role, with the CTO or engineering leader owning the content.

When should I use a skills or training matrix instead?

When the question is coverage, licensing or training completion, not performance or promotion. See when you need a training or skills matrix instead.

Related guides

Job architecture and levelling framework · Talent bar by seniority · employee performance rating scales explained · measuring performance in subjective roles · employee development plan and career conversations · skills gaps that stay invisible until too late · How Cleo rebuilt its skills frameworks in 3 weeks · 360 feedback template · performance review examples by rating level · performance review template · why the 9-box grid is dead · free tools · book a 15-minute review · People Performance Playbook home

Follow us on Google

Want a second look at a draft competency matrix before the next cycle?

Bring a role, the level names you use, and a few descriptors to a 15-minute review. We will check whether the behaviours are observable, whether promotion evidence is clear, and how the matrix sits next to your rating scale. Soft fit-check only.

The PDF asks for a work email. The 15-minute review is a fit-check, not a sales pitch.