The right-hand side of the X-matrix gets less attention than the priorities and more arguments later. A KPI without a clear definition, baseline and source will be disputed in the first review where it matters. Write each one properly at planning.
In this guide
Key Takeaways
- →Ten fields per KPI, written before the matrix locks. Half a page each.
- →Activity disguised as a KPI makes every correlation look like it is working.
- →Write the denominator down. A ratio nobody defined is a dispute waiting for July.
- →Pair slow lagging KPIs with leading ones, so the review has something to discuss each month.
- →One exclusion line, settled at planning, is one sentence. Discovered in July, it is a meeting plus doubt over every month reported.
- →A KPI card only helps if it lives where the number lives, not in a separate glossary.
Why KPIs cause trouble mid-year
In planning, KPIs are usually written as short labels: “services share”, “on-time delivery”, “scope 3 CO2”. Everyone in the room nods, because everyone has a version in mind. The versions differ. One site calculates on-time delivery against the promised date, another against the requested date. Finance counts services revenue one way, the services division another.
None of this matters until the KPI misses its target. Then the first twenty minutes of the review are spent arguing about what the number means, and the correlation the matrix was built on can’t be tested at all.
A KPI card for every KPI
| Field | What to write |
|---|---|
| Name | A short label, used consistently everywhere |
| Definition | What it measures, in a sentence a newcomer would understand |
| Formula | Numerator and denominator, with what is included and excluded |
| Unit and direction | Percentage, count, currency; whether up or down is good |
| Baseline | The current value, with the date and period it covers |
| Target | The value and date, as agreed in catchball |
| Frequency | How often it is updated; monthly for anything on the matrix |
| Source | The system or report the number comes from |
| Owner | One named person accountable for the result |
| Type | Lagging (outcome) or leading (early signal) |
Half a page per KPI. Write it before the matrix is locked.
Five common faults
1. Activity disguised as a KPI
“Service centers opened” is a priority’s progress, not a result. The KPI is what the centers should change: services share, contract attach rate. Activity KPIs make every correlation look like it’s working.
2. No baseline
A target without a baseline can’t be assessed for ambition, and progress can’t be measured. If the baseline isn’t available, measuring it becomes the first task in January.
3. A ratio without a defined denominator
“Percentage of on-time deliveries” depends entirely on what counts as a delivery and what counts as on time. Write both down.
4. Local variants
If sites calculate the KPI differently, the group roll-up is meaningless. Standardize the formula at group level, even if sites keep local variants for their own use.
5. Too slow for a monthly review
Some results only move annually or quarterly. A KPI that can’t change month to month gives the review nothing to discuss for most of the year.
Pair lagging KPIs with leading ones
The fix for fault 5, and a strength for any matrix, is to pair each slow lagging KPI with a faster leading one: a measure that should move first if the priority is working.
| Lagging KPI | Leading KPI | Why it helps |
|---|---|---|
| Services share of sales | Contract attach rate on new equipment | Moves within weeks of a priority working |
| Scope 3 emissions | Share of spend with converted suppliers | Shows progress before emissions data arrives |
| Sites at maturity level 2 | Audit criteria met per site | Shows the path to the next level |
Illustrative pairs.
The leading KPI is usually the one named as the “leading signal” in the correlation’s hypothesis. It’s what lets the review spot outcome lag before the lagging KPI confirms it.
If a KPI can only tell you at year end whether the plan worked, it isn’t helping you run the plan.
A worked example: one KPI, before and after
Before: “Services share: 34%”. After, the full card:
| Field | Entry |
|---|---|
| Name | Services share of sales |
| Definition | Revenue from service contracts, parts and repairs as a share of total revenue |
| Formula | Services revenue ÷ total revenue, both net of intercompany; excludes extended warranties sold with new equipment |
| Unit and direction | Percentage; higher is better |
| Baseline | 31.2%, twelve months to September 2026 |
| Target | 34% for calendar year 2027, agreed in catchball |
| Frequency | Monthly, rolling twelve months |
| Source | Group finance monthly management accounts |
| Owner | Head of service |
| Type | Lagging; paired with contract attach rate on new equipment |
Illustrative.
The exclusion of extended warranties is the line that would otherwise have caused a dispute. The service division counted them; finance didn’t. Settled at planning, it takes one sentence. Discovered in a July review, it would have taken a meeting and cast doubt on every month already reported.
The KPI gate
- Card complete. Every field filled, including baseline and source.
- A result, not an activity.
- One formula across sites.
- Moves monthly, or is paired with a leading KPI that does.
- One owner, who agreed the target through catchball.
In Profit.co, each KPI on the X-matrix is a record with its owner, current value and progress against target, and its correlations to priorities are graded on the matrix. Because KPIs are tracked through check-ins, the baseline and every monthly value stay in its history. More on the planning software page.
What the KPI card needs from the software
A card written in a planning document solves the problem for about a quarter. The dispute in the July review doesn’t happen because nobody ever wrote the definition; it happens because the definition is in a file somebody made in November and the number on screen has no definition attached to it. The card has to live where the number lives.
That is the test worth applying to the ten fields. Click the KPI in the review and the formula, the exclusions, the source and the owner should be there, next to the value that is missing its target.
| Card field | What the tool has to do with it | How to test it in a demo |
|---|---|---|
| Formula and exclusions | Hold them on the KPI itself, visible in the review | Click a KPI mid-review and show me its formula |
| Baseline | Keep it with its date and period, not overwritten by the current value | Show me what this KPI was when the plan was set |
| Frequency | Monthly values as history, not a single current number | Show me every month of this KPI since January |
| Source | A named system, ideally an integration rather than a retype | Where does this number come from, and who enters it? |
| Owner | One person, who checks in on it | Show me the last check-in and who made it |
| Type and pairing | A leading KPI linked to the lagging one it precedes | Show me the leading signal for this outcome |
| One formula across sites | A group definition that site-level roll-ups inherit | Show me this KPI at group and at two sites |
Row 1 is the one that prevents the twenty-minute argument this article opens with.
Row 7 deserves attention in any group with sites or divisions, because fault 4 is a structural problem rather than a discipline problem. If each site maintains its own version of the KPI in its own file, the group roll-up is an addition of different things, and no amount of writing it down at planning prevents that. Deploying hoshin kanri across multiple sites covers what else changes when the plan has more than one level.
What a disputed KPI costs
The worked example puts a price on one line of a card. Extended warranties, counted by one side and not the other, would have cost a meeting in July and doubt over every month already reported. Doubt is the expensive part: once a number is contested, the correlation it was meant to test cannot be judged, and the priority attached to it coasts to December without a verdict. That is one bet lost from the reflection, every year, per disputed KPI.
Against that, pricing models differ mostly in how many people can see the number. OnStrategy lists $200 per month for unlimited users. i-nexus publishes a free Starter tier for up to five users and $28 per user per month at its Standard tier. Row 7 of the table above is the relevant test: a KPI with site-level roll-ups needs the site owners inside the system, not emailing values to a central spreadsheet.
Profit.co publishes two plans, Essentials and Enterprise, and quotes against modules and scale rather than a public seat rate. KPIs are tracked through check-ins, so the baseline and every monthly value stay in the KPI’s own history, which is rows 2 and 3 of the table. Hoshin Kanri sits in Essentials with OKRs and Goals, Balanced Scorecard and Strategy Roadmap, so a KPI on the matrix and the key results the delivery teams update through the quarter are the same kind of object rather than two numbers kept in two systems. What drives the cost of hoshin kanri software covers the rest, including the integrations that feed row 4.
The 30-day free trial is long enough to write cards for the five KPIs on your current matrix and find which of them two departments define differently.
KPIs on the matrix in Profit.co
Taken field by field, the card has somewhere to sit. Definition and formula travel with the KPI rather than a glossary, so the review clicks through to them instead of arguing. The baseline stays with its date while check-ins build the monthly series the frequency field promises. One owner is named on the record, and that is the person whose check-in the review reads.
The pairing in fault 5 works the same way: the leading KPI is a record on the matrix like any other, correlated to the same priorities, so the review can watch the attach rate move in March while the services share is still flat. That is the gap outcome lag describes, and it is only visible if both measures are on the matrix rather than one on the matrix and one in a team’s own report.
Free Product Demo
Check the KPIs on your current matrix
A 30-minute walkthrough, tailored to your team.
Book a free demo
What happens in the call
How your teams set goals, track progress and review results.
Strategy, performance, projects and engagement on one platform, with AI.
A practical path to align teams, execute faster and drive results.
Questions this article answers
A clear definition and formula, a baseline and target with dates, a monthly update frequency, a named source system and one accountable owner.
A lagging KPI measures the outcome, such as services share of sales. A leading KPI moves earlier if the priority is working, such as contract attach rate, so the review can spot problems sooner.
Because measures like “centers opened” track the work, not its effect. They make every correlation look successful even when the result isn’t moving.
On the KPI record itself, so the review can click through from the number to its formula and exclusions. A definition in a separate planning document stops being consulted within a quarter.
Set one formula at group level that site roll-ups inherit, rather than each site keeping its own version in its own file. Sites can still hold local variants for their own use.
Each KPI is a record with its owner, current value and progress against target, correlated to priorities on the matrix. Check-ins keep the baseline and every monthly value in its history.
