How to Set Team KPIs That People Can Actually Move

A team KPI is a number that tells you whether a team is succeeding at the thing the company needs from it. That sounds obvious, and yet most team KPIs fail one of two tests: the team can't actually influence the number, or the number can go up while the team's real contribution goes down.

This is a method for setting ones that survive both.

How many KPIs should a team have?

Three to five. Below three you're measuring one dimension of a job that has several. Above five, attention divides and none of them drives behaviour.

If your team has eleven KPIs, you do not have eleven priorities — you have a dashboard, and the team is quietly choosing which three to care about without telling you which three.

The four tests a team KPI has to pass

Run every candidate through these. Most fail at least one, and the failure is usually informative.

1. The influence test

Can this team materially move this number through their own decisions?

A support team measured on churn fails this. Churn is influenced by product quality, pricing, onboarding, account management and the economy — support is one input among many. Measure them on something they own: first response time, resolution rate, the proportion of tickets that needed a second contact.

Failing the influence test is demoralising in a specific way: the team can do excellent work and watch the number go the wrong way for reasons nobody in the room controls.

2. The gaming test

Could someone hit this number while making things worse?

Almost always yes, and the exercise is to name how. Then decide whether the gaming is plausible enough to matter.

• Tickets closed per day → close them fast without solving anything
• Story points delivered → inflate estimates
• Meetings booked → book unqualified meetings
• Lines of code → write worse code

The standard fix is a paired measure: tickets closed and reopen rate. Velocity and defect escape rate. Meetings booked and meetings that progressed to a next stage. One number is an incentive; two numbers that constrain each other is a target.

3. The visibility test

Can the team see this number without asking anyone?

A KPI that lives in a monthly leadership deck cannot change behaviour on a Tuesday. If it requires an analyst to compute, it will be computed monthly at best, and by then the team has already spent the month.

The practical bar: the team can check it themselves, in under a minute.

4. The ladder test

Does this connect to a company priority?

Every team KPI should be traceable upward. If you cannot draw the line from a team KPI to something the company said mattered this year, one of two things is true — the KPI is measuring maintenance rather than progress, or the company priority list is incomplete.

Both are worth knowing. See company, department and individual KPIs for how the levels should relate.

Lead and lag: have both

Most teams set only lagging measures — the outcome, visible after the fact. Lagging measures tell you whether you succeeded. They do not tell you in time to do anything about it.

Rule of thumb: one lagging measure per team, two or three leading ones. The lagging measure is what you're judged on; the leading ones are what you actually manage week to week.

Setting them: a 45-minute working session

Per team, with the team present — not for them.

1. State the company priority this team serves (5 min). If nobody can, stop; that is the real problem.
2. Ask what "a good quarter" looks like (10 min). In words, not numbers, first. Numbers proposed before the shared picture exists tend to be the numbers that are easiest to pull.
3. Convert to candidate measures (15 min). Aim for six or seven candidates.
4. Run the four tests (10 min). Most candidates die here. That is the session working.
5. Cut to three to five, set targets, name who checks them (5 min).

Doing this with the team rather than to them is not a nicety. A team that has argued its way to its own measures defends them; a team handed measures optimises them.

The mistakes worth avoiding

Cascading the company number unchanged. If the company measures revenue, giving every team "revenue" as a KPI tells nobody what to do differently on Monday. Each level should translate, not repeat.

Measuring activity as though it were outcome. Calls made, tickets touched, documents written. These are useful leading indicators and terrible primary KPIs.

Targets set without a baseline. "Improve response time by 20%" requires knowing the current response time. Measure for a month first if you don't.

Never retiring one. KPIs accumulate. Review them annually and remove the ones that stopped driving decisions — an unused KPI still costs attention in every review meeting it appears in.

Individual KPIs that are just the team KPI. If everyone on a five-person team has the same number, nobody has a personal one. Individual measures need to reflect the specific contribution.

Function-specific examples

The general method above applies everywhere, but the useful examples are function-specific. Start with the KPI examples by department hub, or go straight to sales, engineering or design.

If setting these reveals that the company priorities above them don't exist yet, that is the more urgent problem — and the one our free planning workshop is built to solve.