Performance analytics for a 2,000-seat operation
Problem
Team performance was assessed from spreadsheets assembled by hand each month. By the time a report existed it described a situation six weeks old, and two departments computed the same metric differently, so the numbers could not be compared.
Approach
We defined the metrics once, in one place, with the people who would be held to them. The platform reads from the existing operational systems rather than asking anyone to enter data twice, and computes every figure from the same definitions. It is built for 2,000 concurrent users, so aggregation is precomputed and access is scoped per role: a team lead sees their own team, and no one can quietly widen that.
Outcome
Current figures, one definition per metric, and a visible audit trail of how each number was produced.
Stack
- SvelteKit
- TypeScript
- PostgreSQL
- Docker