Cutting a restaurant chain's stock waste by a third

A four-branch restaurant group came to us with a problem they described as "we need a better POS". After a week of watching service, it was clear the till was not the problem. The problem was that nothing the till knew ever reached anybody who made a decision.

What was actually happening

Orders were taken on paper, carried to the kitchen, and rung into a till separately. Three consequences followed, and none of them looked like a software problem from the inside:

• Stock was a weekly guess. Orders were placed from habit. The same quantities

went in every week regardless of what had sold, so popular items ran out and perishables were thrown away.

• Closing took forty minutes a night, per branch. A manager reconciled paper

tickets against the till by hand, producing a number the till already had.

• Orders got lost between floor and kitchen. Remade dishes cost twice: the food,

and the wait.

The owners could see the money leaving. They could not see where.

What we built

A single connected system, delivered in stages so each piece paid for itself before the next began.

Stage one — POS and kitchen display. Orders go from the floor straight to a kitchen screen. No handwriting, no lost tickets, no re-entry. This alone removed the remake problem within two weeks.

Stage two — recipes and live stock. Every menu item is mapped to its ingredients, so a sale deducts stock automatically. The weekly guess became arithmetic: the system knows what left the building.

Stage three — ordering against consumption. Suggested purchase orders generated from actual sales, adjusted by a manager before going to suppliers.

Stage four — one dashboard across branches. Owners see all four branches together: sales, stock, waste and labour, in numbers that agree with each other because they come from one source.

What changed

The waste figure is the one the owners quote, and it came almost entirely from stage three. Ordering to actual consumption instead of to habit is not a sophisticated idea. It was simply impossible before the data existed.

We were ordering the same amount every week regardless of what we sold. The system just showed us the difference.

the group's operations manager

What made it work

We watched a Friday service before writing anything. Every wrong assumption we would have made was corrected in that one evening. No specification document would have surfaced them.

Staged delivery. Each stage went live on its own and earned its keep before the next started. If the project had stopped after stage one, it would still have been worth doing.

One branch first. The busiest one, for three weeks, before the other three. The things that went wrong went wrong once.

What we would do differently

We underestimated recipe mapping. Mapping every menu item to ingredients is tedious, it needs the head chef rather than a developer, and we scheduled it as if it were a data-entry task. It took roughly twice as long as planned and it is the foundation everything else sits on.

We built reporting too early. The first dashboard was designed before anyone had lived with the data, and half of it went unused. The version that stuck was built in month three, from questions people had actually started asking. This is the trap described in your dashboard is lying to you — we walked into it ourselves.

Would this work for you?

The gains here came from a specific situation: multiple branches, perishable stock, and a process where information was captured but never reached a decision. If that describes you, the numbers are plausible for your business too.

If you are a single site with simple stock, a good off-the-shelf POS will get you most of this for a fraction of the cost. That comparison is worked through in ready-made or custom POS.

Common questions

How long did it take?

Fourteen weeks from first visit to all four branches live, across the four stages. Stage one was running in the first branch by week five.

What about the e-invoicing requirements?

Built in at stage one rather than retrofitted, which is considerably cheaper. See connecting your POS to the ETA system.

Did staff resist it?

Less than the owners feared, because the kitchen screen made the cooks' jobs easier on day one. The order matters: automate something that helps the people using it first, and the rest is accepted far more readily.