KPI standardisation

HR and performance

KPI standardisation

510 KRA names for 198 people, reduced to 57 KPIs

  • 1,783 → 57KRA rows into standard KPIs
  • 99.3%rows mapped to a reference unit
  • ~88%estimated auto-trackable if source data exists
KPI standardisationscreenshots and video coming

The problem

1,783 KRA rows for 198 employees across 37 roles. 510 distinct KRA names and 608 distinct KPI texts. The only clean field was the weightage. Reviews were run on what people presented, not on a scorecard.

What I did

  1. A rule-based classifier collapsed 510 names into 15 KRA heads and 57 standard KPIs, each with a formula, unit, direction, frequency, source system and tracking method.
  2. Every KPI reduced to one of six measurement archetypes: achievement against target, zero-incident count, turnaround time, compliance percentage, volume, manager-rated evidence.
  3. Every one of the 1,783 rows kept and mapped to the reference business unit’s KPI. 99.3 percent coverage.
  4. A composite-role model for people who hold more than one function: percentage-allocated KPI blocks, with self-development held once at ten percent.
  5. An honest estimate: about 88 percent of rows could be system-fed if the source data existed, with ten named data-quality blockers. This is “fix data first” expressed in numbers.

What it changed

It became the backbone of KRAstra and the clearest statement the company had of what it could measure automatically and what it could not.