KPIs that survive contact with the business

A recurring scene in leadership meetings: the sales director says the pipeline is up eighteen per cent, finance says bookings are flat, and the next twenty minutes are spent discovering that one of them counts a verbal commitment and the other counts a signed contract. Both were right. Neither definition was written down.

Choosing which metrics to track is the part everyone enjoys and the part that matters least. Defining them precisely, sourcing them consistently, and refusing to change them casually is the part that determines whether the dashboard is a decision tool or a monthly argument.

Every measure needs a written definition

For each KPI in the pack, record the plain-language meaning, the exact calculation including numerator and denominator, the system of record it comes from, the owner, the reporting frequency, and the treatment of the awkward cases. Utilisation is the classic example. Does the denominator include annual leave? Public holidays? Training days? Is a non-billable internal project counted as available or utilised? Each choice is defensible and each produces a different number from identical timesheets.

Write the answer down once, put it in an appendix to the pack, and the argument does not recur.

One source of record per measure

Where a metric can be produced from two systems, it will be, and the two will disagree. Headcount from the HR system will not match headcount from payroll, because one counts contractors and the other counts leavers to their final pay date. Revenue from the CRM will not match revenue from the ledger. Neither system is wrong; they are answering different questions.

Name one system as the source for each measure, and where a second system is used operationally, publish the reconciliation rather than pretending the difference does not exist. A single page showing why CRM revenue and ledger revenue differ by a known, explicable amount removes more friction than any dashboard redesign.

A short test for whether a metric belongs in the pack

• Someone in the room can change it through a decision they are actually able to make.
• It moves before the financial results do, rather than restating them.
• There is a target or a threshold, so that a value can be judged rather than merely noted.
• It cannot be improved in a way that damages the business. Utilisation without a quality or margin counterweight is the standard cautionary case.
• It has survived unchanged for four quarters, so the trend means something.

Change definitions rarely, and restate when you do

Definitions do need to change occasionally, when the business changes shape or a measure turns out to be driving the wrong behaviour. When that happens, restate the history on the new basis and show both series for a period. A metric whose definition quietly changed in March is worse than no metric at all, because the trend line lies with a straight face and everyone believes it.

Eight to twelve measures, defined precisely, sourced consistently, and held stable for years will out-perform a thirty-metric dashboard rebuilt every other quarter. The value is in the pattern recognition a leadership team builds when the ground stops moving underneath them.