How do you compare performance across multiple locations without the comparison being unfair?

Configuration drift ruins comparison before math does

The first thing to check in any multi-location rollup is whether the underlying systems agree on what a job type is, when a job is complete, which lead sources exist, and what counts as a booked call. In practice they rarely do, especially where locations were acquired and kept their own habits.

This is unglamorous and it is the whole battle. A rollup that averages inconsistent definitions produces confident numbers about nothing, and everyone in the room senses it, which is why nobody acts on the report.

Normalize for what the manager cannot control

• Scale. Compare revenue per truck-day, not total revenue. A larger branch should produce more; that is not performance.
• Demand density. Calls per capita or per household in the service area separates market from execution.
• Season timing. Two markets do not hit peak in the same week. Compare against each location's own seasonal baseline, not the calendar.
• Crew tenure. A branch that hired four technicians last quarter will show lower productivity for reasons that are healthy.
• Lead mix. A location fed by service agreements and repeat customers converts differently than one fed by paid search.

Rank on controllables

Once normalized, the useful comparisons narrow to a short list of things a location genuinely owns: booking rate on bookable calls, callback rate, days to first available, unsold estimate follow-up completion, and average ticket within job type. Those are behaviors. Total revenue is an outcome shaped by market conditions nobody at that branch chose.

Ranking on outcomes creates defensiveness. Ranking on controllable behaviors creates a conversation about process, which is the point of having the report at all.

Build the rollup so it can be argued with

Every number in a multi-location report should be traceable to its source records in one click, because the first question from a branch manager is always whether the number is right. If the answer takes a week, the report loses.

That drill-through requirement is a design constraint on the whole reporting stack, and it is why we build multi-location views on top of a proper integration layer rather than as spreadsheets assembled by hand. Delivering the same view at three altitudes, one for the branch, one for the region and one for ownership, is standard in brief design.