Before you open a second shop, model the first one

Owners usually know their first site works. Far fewer know exactly why. The rent was low when they signed, the founder bakes at four in the morning, and the regulars have been coming for twenty years. A second site copies the parts you can see and loses the rest.

That’s why so many second sites disappoint. The shop looks the same, the product is the same, and the numbers come out worse for reasons nobody wrote down.

Model one site from the bottom up

Before we talk about locations, we build a model of the site that already works: rent, opening hours, staff by the hour, sales by the hour, footfall, waste and what each product earns. It usually takes two weeks and a lot of time behind the counter.

At Haldane’s Bakehouse in Glasgow the model showed something the family half knew. Every shop baked its own bread, and the ovens, the early staff and the waste were eating the margin of each new shop before it opened its doors.

The cheapest place to make a mistake is a spreadsheet. The second cheapest is a pilot.

Test on paper, then in a pilot

Once the model matches the real shop within a few percent, we use it to compare ways of growing. For Haldane’s we compared more full bakeries, franchising and one central bakehouse feeding smaller shops. On paper the central bakehouse won by a distance.

A model can still be wrong, so we tested it. For six weeks one existing shop took its bread from the central bakehouse instead of its own ovens. Sales held, waste fell, and the morning shift got two hours shorter.

Only then did anyone sign a lease. Haldane’s opened its new shops near stations, smaller than the originals and without ovens.

They went from four shops to eleven in two years, and every shop they opened after the plan paid back in under eight months.