A margin turnaround fails when it starts with a cut. Cutting a shift or a supplier before you know which one is actually the leak just moves the pain around — sometimes onto service, which then costs you the volume you needed to recover. The sequence that works runs in the opposite order: measure first, decide second, cut last and narrowly.
Weeks 1–2 — measure
Cost your top twenty sellers properly — real ingredient prices, real portions, a real fiche technique for each. Twenty items is usually 70–80% of volume on most menus, so this is where the real numbers live, not a full-menu audit that takes a month and delays every other step. By the end of week two you know your real food cost, not the one from the last time someone checked.
Weeks 3–6 — fix pricing and portions
With real costs in hand, two moves: reprice anything that's been selling below a healthy margin for reasons nobody remembers, and standardize portions on anything that's drifted — the dish that's costed at 30% and served at 38% because the portion crept. This is the fastest-moving phase because it changes nothing about how the kitchen operates, only what leaves the pass and what it's sold for.
Weeks 7–10 — renegotiate purchases
Now that six weeks of consigned invoices exist, renegotiate the top ten lines with real numbers instead of a general sense that things cost more than they used to. This phase moves slower — suppliers need a real conversation, not an email — but it's where the gains compound with everything fixed in weeks 3–6.
Weeks 11–13 — lock the routine
A turnaround that isn't habitual reverts within a season. Lock in the cadence that got you here: prices checked against invoices weekly, portions spot-checked, the top-twenty costing refreshed every quarter. The 3-number diagnosis — food cost, labor, rent — becomes the monthly check that catches the next drift before it needs another 90 days.
Expect a few points of food cost recovered, not a miracle — a well-run 90 days typically closes a meaningful chunk of the gap between theoretical and actual cost, not all of it. That's still, on a typical 5–10% net margin, a real share of the profit that was quietly leaving the building every month.
