People framework
A competency matrix wired directly to pay
Sixty criteria, a five-point scale, and a formula where the average score multiplies the base salary. Useful, transparent — and it had an arithmetic flaw nobody had noticed.
- Grouped, defined, and scored on a five-level mastery scale
- 60+ criteriaGrouped, defined, and scored on a five-level mastery scale
- Every row assessed twice, with a comment and a deadline for the gap
- Current and targetEvery row assessed twice, with a comment and a deadline for the gap
- The average score multiplies base pay — a formula, not a conversation
- Score × baseThe average score multiplies base pay — a formula, not a conversation
The client
The analysis function of a software services company, where analyst grades and salary reviews were decided through a documented assessment rather than through a conversation at review time.
The engagement
A methodology review: the matrix, the assessment procedure and its link to compensation, read as a system rather than as a form.
The problem
Most competency matrices are decoration: they exist, they are filled in once, and pay is still decided in a conversation. This one was wired directly to money, which makes it honest and also makes its arithmetic consequential — and the arithmetic had not been examined since it was written.
What I did
I read it as a system rather than as a form. The two-column structure — current level and target level, with a comment on what is missing and a deadline — is the strongest part, because it turns an assessment into an agreed path rather than a verdict, and the definitions are written well enough to lift straight into a job description, a statement of work or the team section of a proposal. The flaw is in the aggregation: because the outcome is a plain average across sixty-plus rows, an individual skill barely moves the result, but any row left unscored shifts the average for everyone assessed inconsistently — and in the completed sheets, rows were routinely left blank. That is not a philosophical objection; it means two people with identical skills get different numbers depending on how thoroughly their assessor filled the form.
What was built
A competency framework of sixty-plus criteria with definitions and a five-level mastery scale, assessed in two columns — where the person is now and where they need to be — with a comment and a deadline per gap, feeding both an individual development plan and a compensation formula.
On the table at the end
- Competency framework with definitions and a five-level scale
- Assessment sheet linking current and target levels with deadlines
- Documented grade-to-pay formula
- Finding: unscored rows distort the average
What it changed
Documented a rare thing — an explicit formula from competency score to salary — made it reusable for assessing internal, partner and candidate analysts, and identified an arithmetic flaw that quietly distorted individual results.
How it ran
- 01
Read the scale definitions
Five levels from 'does not see why it matters' to 'teaches it to others', with half-points used in practice.
- 02
Note the two-column design
Current and target assessed together with a comment and a deadline, turning assessment into an agreed development path.
- 03
Trace the money
The average of the current column multiplies base pay — an explicit formula rather than a discussion at review time.
- 04
Test the arithmetic
A plain average across sixty rows makes any single skill nearly weightless — and makes blank rows silently decisive.
- 05
Generalise the asset
The definitions are reusable for candidate screening, partner analyst assessment and the team section of commercial proposals.
Other work
All case studies →- Delivery staffing
Rejecting the entire shortlist
A regulated milestone depended on one rare engineering skill. Everyone available was a comfortable yes, so I said no to all of them and re-sourced against a single weighted axis.
- Professional services
A project office that lives in a git repository
Project knowledge lived in people's heads and chat threads, and AI tools were producing confident documents with no provenance. I made the repository the single source of truth and gave every project management role a written protocol.
- Enterprise software
Data and people existed; the process and the cockpit did not
A delivery assurance function owned everything and controlled nothing. I diagnosed nineteen gaps, designed the target operating model, and put stop conditions in the roadmap so it could not become another initiative nobody uses.
Something similar on your plate?
Thirty minutes, no deck. I will tell you whether it is worth doing at all.