Running a restaurant group: the same figures, read site by site

Before you compare two sites, they need to speak the same language. A dish has the same name and the same code everywhere, and points to the same recipe card, with the same ingredients, the same portion weights and the same cost per portion. An ingredient is bought, stored and counted in the same unit, or with a conversion written down once and for all: a case of six bottles can’t count as ‘6’ at one site and ‘1’ at another.

Definitions follow the same rule. A cover is a customer served: decide once whether a children’s menu, a drink on its own or a takeaway order counts, and the rule applies everywhere. Revenue is read excluding VAT, after discounts. The calendar is shared: weeks run from Monday to Sunday, months close on the same date, stock is counted on the same day.

Then there is what moves between sites. A crate of tomatoes lent by one site to another is recorded, with the date, on both sides: out at one, in at the other. A forgotten transfer inflates the food cost of the site that gives and flatters that of the site that receives.

A head office dashboard fits into seven indicators, provided each has a written definition, and only one. Revenue excluding VAT and the number of covers describe activity; average spend shows what a customer spends.

Food cost measures what was actually consumed, transfers included; divided by revenue, it becomes the food cost percentage. Labour cost covers gross wages, employer contributions and casual staff; divided by revenue, it becomes the labour cost percentage. Gross margin shows what is left to pay the team, the rent and everything else.

Finally, the stock variance compares what was counted with what should have been there. Theoretical stock is calculated from sales and recipe cards.

A negative variance points to missing goods. It is read in euros and ingredient by ingredient, never as a total alone.

Totals favour large sites, and the group average hides the ones that drift. Read each site before the average, brought back to a common base: per cover for food, staff and margin; per opening hour for activity, because a station site open from 07:00 to 22:00 can’t be compared with a restaurant open for two services.

Format matters too: each sales mix has its own normal food cost percentage. Compare each site with its own theoretical percentage, derived from its sales and the recipe cards.

In totals, Bastille consumes the most food (€52,080) and Part-Dieu makes the smallest gross margin (€70,560): you would go and visit the wrong site.

Per cover, the picture changes. At Confluence, a cover uses €6.20 of food against €5.60 expected: €0.60 × 6,200 covers = €3,720, which is its 3-point variance. Part-Dieu shows 30%, but that is its normal percentage. Across the group, the variance falls to 0.9 points (€3,720 out of €410,800 revenue excl. VAT): the average dilutes the problem.

Same trap with staff: Part-Dieu has the highest labour cost percentage (34%), but the lowest cost per cover (€6.12). It is its lower average spend that inflates the percentage, not its rota.

Example: figures invented for the calculation.

A group is run to a fixed schedule. Every Monday, head office and the site managers review the week just closed, site by site, with the same table: covers, average spend, labour cost percentage, flagged variances. The meeting stays short if the figures are ready beforehand, not during it.

Every morning, each site checks its own alerts: a stock-out, a till discrepancy, a rota that doesn’t cover the footfall expected by the sales forecast. Food cost and gross margin are read at the end of the month, after the stocktake. A food cost percentage calculated each week from purchases alone follows delivery days, not consumption.

Each review ends with one action per site, with an owner and a date; the next one starts by checking they were done. Head office compares in order to help: a site that drifts gets a question, not a reprimand. A variance is not a fault; it is a lead to check: portions, waste, purchase prices, stock count or an unrecorded transfer.

It is not about centralising everything or delegating everything, but about writing down who decides what. One rule works well: head office sets the framework, the site decides within it.

For purchasing, head office approves suppliers and negotiates prices; the site places its own orders, because it knows its stock and its week. For scheduling, head office sets the rules: positions, required skills, target percentage; the manager builds their team’s rota. For prices and the menu, in a company-owned group, the decision lies with head office: it is the group’s identity. A site can propose, not change things on its own.

Harmonising recipe cards and purchasing remains head office’s job: if each site adjusts its portion weights or picks its own products, the theoretical percentage no longer means anything and comparisons fall apart. In return, everything the site does not decide must be explained to it, and everything it does decide must be measured with the same definitions as the others.

In a company-owned group, all sites belong to the same business: managers are employees, and head office can set the menu, prices and scheduling rules. In a franchise, in quick service groups as elsewhere, each site is a separate business and the franchisee an independent trader: they run the business at their own risk and for their own profit, and remain free to set their prices.

The franchisor can require the franchisee to meet the network’s standards (ingredients, production processes, hygiene), to buy from approved suppliers, to accept visits and audits, and to send regular reports on performance and sales. It cannot impose, even indirectly, a minimum resale price: Article L442-6 of the French Commercial Code punishes this with a €15,000 fine.

The franchisee must also keep room for manoeuvre: a contract that left them only illusory autonomy could be reclassified as an employment contract. The franchisor’s dashboard is therefore for comparing and advising; operating decisions remain the franchisee’s.

Finally, Article L330-3 of the French Commercial Code, which stems from the Loi Doubin, requires the prospective franchisee to be given a truthful pre-contractual disclosure document and the draft contract at least twenty days before signing or, where applicable, before any required payment.

Managing on totals. The largest site always has the biggest food cost. Express every figure per cover or as a share of revenue.

Comparing raw percentages across formats. A brasserie and a takeaway counter don’t have the same normal food cost percentage: compare their variances against theoretical.

Letting each site define its own figures. If one site counts drinks on their own as covers and the other does not, their average spends can no longer be compared.

Looking only at the group average. It can stay stable while one site drifts, offset by another.

Piling on indicators. Beyond one page, the dashboard can no longer be read. An indicator that triggers no decision can go.

Turning the review into a tribunal. A site that feels judged rounds off its stock counts; a site that is supported counts accurately.

How many indicators should head office track? Few. The seven in this guide cover activity, food, staff and margin. Add one only if it triggers a specific decision.

How many sites before you need a shared dashboard? From the second. That is when definitions start to diverge without anyone noticing, and when they are easiest to fix.

How do you compare a small site with a large one? Per cover, per opening hour, and against its own theoretical percentage. Never in totals.

Can a franchisor impose its prices? Not a minimum price: Article L442-6 of the French Commercial Code prohibits it, and the franchisee remains free to set their prices. The franchisor may, however, impose prices on an occasional basis, for example for the duration of a short low-price campaign.

Who sees which figures? Head office sees every site, with the same definitions; each manager sees their own. In a franchise, what the franchisee passes on to the franchisor is set out in the contract.