KPI Examples by Department
KPI examples for sales, engineering, design, marketing, support, finance and HR — with the paired measure each one needs to stop it being gamed.
Every KPI on this page comes with a paired measure — the second number that stops the first one being gamed. That pairing is the whole point. A single number is an incentive; two numbers that constrain each other are a target.
Start with the function you're setting KPIs for. If you want the method rather than the examples, how to set team KPIs covers the four tests every candidate should pass.
Sales
| KPI | Paired with | Why the pair |
|---|---|---|
| Closed revenue | Win rate | Revenue alone rewards volume of poor-fit deals |
| Qualified opportunities created | Opportunity-to-close rate | Otherwise "qualified" inflates |
| Average deal size | Sales cycle length | Big deals that never close aren't wins |
| Pipeline coverage | Pipeline age | Stale pipeline flatters coverage |
Leading indicators worth tracking: discovery calls held, multithreaded accounts (more than one contact engaged), proposals sent.
→ Full sales team KPI examples
Engineering and development
| KPI | Paired with | Why the pair |
|---|---|---|
| Deploy frequency | Change failure rate | Shipping fast and breaking things isn't shipping |
| Lead time for changes | Review turnaround | Reveals whether the bottleneck is coding or waiting |
| Uptime / availability | Time to restore | Uptime hides how bad the bad days are |
| Story points or throughput | Defect escape rate | Otherwise estimates inflate |
Avoid: lines of code, commits, individual velocity. All three actively reward the wrong behaviour.
→ Full engineering team KPI examples
Design
| KPI | Paired with | Why the pair |
|---|---|---|
| Designs shipped | Rework rate after handoff | Fast handoff that engineering can't build isn't done |
| Task success rate in usability testing | Time on task | Users succeeding slowly is a finding |
| Design system adoption | Number of one-off components | Adoption claims need a counter-check |
| Research studies run | Decisions changed by research | Research nobody acts on is theatre |
→ Full design team KPI examples
Marketing
| KPI | Paired with | Why the pair |
|---|---|---|
| Qualified pipeline generated | Pipeline-to-close rate | MQL volume is easy to inflate |
| Cost per qualified opportunity | Blended CAC payback | Cheap leads that don't convert aren't cheap |
| Organic sessions | Conversion rate by source | Traffic without intent is vanity |
| Content published | Content that generated pipeline | Volume is not the outcome |
The classic failure: measuring MQLs without measuring what fraction become real opportunities. Marketing hits target, sales complains about lead quality, and both are correct.
Customer support
| KPI | Paired with | Why the pair |
|---|---|---|
| First response time | Resolution rate | Fast acknowledgements aren't help |
| Tickets resolved | Reopen rate | Closing isn't solving |
| CSAT | Response volume on CSAT | High scores from 4% of customers mean little |
| Time to resolution | Escalation rate | Speed via escalation just moves the cost |
Finance
| KPI | Paired with | Why the pair |
|---|---|---|
| Days to close the month | Post-close adjustments | Fast closes that get restated aren't closed |
| Forecast accuracy | Forecast frequency | Accuracy is easy if you forecast rarely |
| Cash runway | Burn multiple | Runway extended by not investing is a different story |
| DSO | Bad debt rate | Aggressive collection has a cost |
People and HR
| KPI | Paired with | Why the pair |
|---|---|---|
| Time to hire | 12-month retention of new hires | Fast hiring that doesn't stick is expensive |
| Offer acceptance rate | Diversity of the pipeline | Acceptance can be optimised by narrowing |
| Voluntary attrition | Regretted attrition specifically | Not all attrition is bad |
| Engagement score | Response rate | A high score from half the company is unread |
Operations
| KPI | Paired with | Why the pair |
|---|---|---|
| On-time delivery | Quality or defect rate | On-time-and-wrong is not on-time |
| Utilisation | Rework hours | Busy is not productive |
| Cost per unit | Service level | Cheap and unreliable is a choice, not a win |
Three rules that apply to every function
Three to five per team, not eleven. More than five and the team is silently choosing which to care about without telling you which.
Every KPI needs a named owner and a visible source. If checking it requires an analyst, it will be checked monthly at best.
Every KPI should ladder to a company priority. If you can't draw the line upward, the KPI is measuring maintenance rather than progress — worth knowing, and different from progress. See how the levels connect.
Before you copy any of these
These are starting points, not answers. A KPI copied from a list measures a generic version of your function; the useful ones are argued into existence by the team that will be measured on them.
And if setting department KPIs surfaces that the company priorities above them aren't agreed, that is the more urgent problem. It is what our free planning workshop is built to fix — you leave with the company layer and the department layer that hangs off it.
Run this session with us instead
Everything above is the version you run yourself. If you'd rather not facilitate your own planning session, we'll run it with you free — you leave with an agreed North Star, a KPI structure from company down to individual, and a plan your team can actually execute.
Book a free planning workshopKeep reading
How to Set Team KPIs That People Can Actually Move
A practical method for setting team KPIs — how many to have, how to tell a real one from a vanity metric, and the ownership test that catches most bad ones.
How Company, Department and Individual KPIs Connect
How KPIs should relate from company level down to an individual — what changes at each level, and why cascading the same number everywhere doesn't work.
Sales Team KPIs and the Measures to Pair Them With
The sales KPIs worth tracking at team level, the paired measure each one needs, and the four that quietly reward the wrong behaviour.