KPI Examples by Department and Role: Formulas, Denominators, and Operating Rules

Choose a KPI by the decision it supports, not by copying a metric associated with a job title. A KPI is an important indicator; a target is the desired level; an actual result is the measured value. Whether an indicator leads or follows an outcome depends on the outcome being assessed. The candidates below are not benchmarks or ready-made targets.

Define each candidate before using it

Record the formula, eligible population, period, source, owner, and update date. Identical names can conceal different denominators and success conditions. Use the marketing glossary and metric distinctions when terminology is unclear, then select the definition that fits your decision.

Sales, marketing, and support examples

For the win-rate definition above, exclude opportunities that remain open. In a fictional month, 30 wins from 100 closed eligible opportunities give a 30% win rate. At an average contract value of JPY 200,000, won contract value is JPY 6,000,000. This is neither automatically recognized revenue for the month nor proof of an annual target being met.

For session conversion rate, several outcome events within one session do not turn into several successful sessions. Google Ads’ average CPA concerns defined conversions. A contact action’s CPA differs from CAC based on new paying customers and the relevant sales and marketing acquisition costs.

Distinguish satisfaction from recommendation

Define CSAT using the selected question, response scale, positive-response rule, and valid responses. NPS uses a 0–10 recommendation question: the percentage scoring 9–10 minus the percentage scoring 0–6. Its −100 to 100 score is not a percentage of satisfied customers or a count of successful referrals. Check response rates and respondent mix before comparing surveys.

Keep hiring and employee populations separate

In a fictional hiring group, 20 valid offers, 16 acceptances, and 14 actual starters produce 80% acceptance and an 87.5% join rate among acceptances. Record whether the other two are awaiting their start dates or have withdrawn. Acceptance and joining are separate outcomes.

The leavers/average-headcount approach appears in CIPD’s 2020 report. Use the formula here, not its historical survey results as current benchmarks. In another fictional year, 12 leavers and average headcount of 210 give approximately 5.7% turnover. If 190 of the 200 opening employees remain, opening-cohort retention is 95%. Ten opening employees left; the other two leavers were later hires. Different populations and denominators mean retention is not 100% minus 5.7%.

Pair delivery speed with quality

If using DORA change lead time, measure commit to production in the application, service, or team context and read it alongside stability. Avoid individual speed rankings. Code coverage measures execution of lines, branches, or other selected structures during tests, not the percentage of code proven correct. Google’s coverage guidance explains why pursuing 100% alone can create false confidence.

Back-office candidates include receipt-to-completion time and returned items divided by eligible processed items. Align deadlines, working calendars, and case complexity. Faster processing should not reward skipped approvals or safety checks.

Check suitability for individual goals

For a fictional content role, publication count should be read with qualified traffic and inquiry quality. Select the smallest useful set rather than imposing a universal metric count. Test the calculation first; investigate misses or sudden changes using the KPI diagnosis and review workflow before attributing them to performance.