Quarterly · anonymised · operator-submitted

The GGB GCC Restaurant Margin Index.

Every GGB diagnostic collects real operating figures. With each operator's explicit permission, those figures are pooled — no names, no venues — into the margin picture nobody else in the region has: what food cost, labour, delivery mix and prime cost actually look like, by market.

Edition status

First edition publishes when the sample threshold is met.

The honesty rules are fixed before the first number is published: only figures with the operator's explicit "include mine" consent enter the pool; no cell publishes with fewer than 8 contributing operators; no edition publishes below the quarterly sample floor. No interpolation, no industry-copy dressed up as GGB data — a thin sample reads "insufficient sample", never a synthetic number.

Contribute your numbers by running any free diagnostic — the consent box is yours to tick, and groups under 8 operators are never published.

Run the Profit Leak Audit All free diagnostics →

What each edition reads

Five numbers, by market and outlet count.

Food cost %

Where the GCC operator base actually runs — not where suppliers say it should.

Labour vs sales

Productivity spread by market and outlet count, from operator-submitted figures.

Aggregator mix

How much of revenue rides delivery platforms — and what commission load comes with it.

Prime cost

The discipline number: food + labour together, distribution and quartiles.

Rent load

Occupancy cost as a share of revenue, as captures carrying rent figures accumulate.

Fixed before the first number

How the Index is governed.

These rules are enforced in the aggregation machinery itself, not in editorial judgement — and they were fixed before any edition existed, so no number can ever bend them.

Consent basis: Only figures with the operator’s explicit, unticked-by-default “include mine” consent enter the pool. Figures captured before the consent box existed are excluded by construction — backfill is structurally impossible.
Edition floor: No edition publishes until a quarter holds at least 30 consented submissions with figures. Below that, the honest state is “no edition” — not a thinner report.
Small-cell suppression: No cell (market × outlet-band × metric) publishes with fewer than 8 contributing operators. Suppressed cells read “insufficient sample” — they are never interpolated, filled from industry copy, or merged to sneak past the floor.
One operator, one voice: One contributor counts once per quarter: where a contact record exists, the latest submission supersedes earlier ones. Consented figures without a contact record cannot be identity-checked and are admitted as-is — a stated limit of the method.
Impossible values rejected: Figures outside plausible operating ranges (food cost 5–90%, labour 3–80%, aggregator share 0–100%, rent 0–80%) are rejected at admission — never clamped or corrected into range.
Outlier policy: Bounded admission plus percentile reporting: editions publish p25 / median / p75, never means — so a single extreme venue cannot drag a market’s number.
Synthetic & test exclusion: Monitoring canaries and synthetic test rows are excluded from every count, whatever flags they carry. Real operators only.
Corrections & versioning: Editions are versioned by quarter. If an error is found after publication, the edition republishes with a dated correction note — numbers are corrected in the open, never silently.

What the Index does not prove: it is a picture of the consenting GGB diagnostic base, not a census of the region’s restaurants — markets and formats where few operators contribute are under-represented, and a distribution describes the pool, never any single venue. It is context for your numbers, not a target, a benchmark promise, or a substitute for reading your own P&L.

Reserve the first edition

Be first to read it.

Leave an email and the first edition lands in your inbox the day it publishes — as a PDF, free. That single delivery is all this form signs you up for.