Multi-location operations
How to compare performance across restaurant locations
Comparing locations is not a metrics problem, it is a definitions problem. Get the definitions wrong and your league table ranks your accounting decisions rather than your operators — and you will send your area manager to the wrong store.
Agree the definitions before you agree the metrics
Four decisions have to be identical across every location. Each is defensible either way; what is indefensible is different stores making different choices silently.
| Decision | Why it matters | Typical damage |
|---|---|---|
| Comps and discounts | Deducted from sales, or booked as a cost? | 2–4 points of prime cost |
| Salaried managers | In labour, or in operating expenses? | 3–6 points of labour |
| Delivery commission | Cost of sale, or marketing? | 2–5 points, worst at high-delivery sites |
| Period boundaries | Calendar month, or four/five-week periods? | An entire extra Saturday |
Comparing a month with five Saturdays to one with four, then holding a manager accountable for the difference, is the most common reporting error in multi-unit restaurants. Use like-for-like periods or normalise by trading day.
The four measures worth ranking on
1 · Prime cost %
COGS plus labour over net sales. The headline. Everything else explains it.
2 · Sales per labour hour
Net sales ÷ hours worked. Underrated, because labour percentage moves when your prices move but sales per labour hour tells you whether the shift was staffed correctly for the volume that actually walked in. Track both — they disagree usefully.
3 · Labour cost coverage
The share of worked hours carrying a cost. Not optional. Without it you cannot tell whether a store is efficient or simply under-reporting, and a group with uneven coverage produces a ranking that is partly a map of its data problem.
4 · Average order value
Net sales ÷ orders. Stops you judging a high-volume, low-check site by the same labour percentage as a low-volume, high-check one.
What a real comparison looks like
A twelve-store group, August 2026, every location on identical definitions:
| Store | Net sales | Prime % | Sales / labour hr | Avg order |
|---|---|---|---|---|
| Parkview | $94,078 | 23.60% | $37.69 | $16.75 |
| Eastway | $115,567 | 29.51% | $267.70 | $18.04 |
| Hillcrest | $79,863 | 47.25% | $47.52 | $20.75 |
| Cedarfield | $76,512 | 61.29% | $60.27 | $20.85 |
| Riverbend | $79,324 | 72.22% | $33.28 | $22.16 |
| Millbrook | $50,586 | 75.84% | $31.92 | $19.49 |
Notice Eastway: a sales-per-labour-hour of $267.70 against a group norm in the $30–60 range. That is not an extraordinary store — it is a store whose labour hours are largely unmapped, so the denominator is tiny. An outlier that good is a data signal, not a performance signal.
Rank on the gap, not the average
The group above averages 50.2% — respectable, and useless as a management target. The spread runs 23.60% to 75.84% on the same brand and menu. The gap is the more credible improvement target, because you are not asking a store to hit a consultant’s benchmark; you are asking it to hit what the store eleven miles away already achieves.
Work out what your gap is worth — it usually surprises people.
Give each level only its own numbers
A comparison is only actionable if the person who can act on it can see it. That usually means three views: the GM sees their store, the area manager sees their region, the executive sees the group — and the GM does not see the group’s margins.
One chart of accounts and one set of definitions across every location, with a comparison view that ranks every store on the same measures and reports labour coverage alongside, so you know which rows to trust. Access is scoped per store at sign-in, so a GM gets their own operating figures without the group’s financials.