The Fifth Outlet Problem: Why Kitchens Drift When You Repeat Them

A kitchen that works once is not the same thing as a kitchen that works five times.

The first outlet gets the attention. The chef is involved, the contractor is on site daily, and whatever does not quite work gets fixed by someone standing there. By the fourth or fifth outlet, neither of those conditions holds. Two variables have entered that the first kitchen never had to survive: the building, and the person on shift.

Most rollout problems are one of those two wearing a disguise. And the usual response — more training, a tighter SOP, a stricter opening checklist — treats the symptom. Training does not change gas pressure. A checklist does not make a piece of equipment easier to clean at 11pm.

What follows are four problems we were asked to solve for multi-site operators in the UAE. None of them were fixed by training. All four were fixed upstream of the kitchen, in the specification or in the equipment itself. The brands are not named.

When the site is the variable: a wok range that failed differently at every outlet

The wok range is the centre of a Thai kitchen. Get it wrong and nothing downstream recovers.

A Thai casual-dining brand had wok range trouble at outlet after outlet — in a different way each time. That detail is what made it hard. A fault that appears identically everywhere gets diagnosed quickly. A fault that presents differently at every site looks like bad luck, or like a series of unrelated problems, and tends to get absorbed rather than solved.

Two root causes sat underneath it. In the UAE these ranges are custom-fabricated locally, and Dubai gas pressure varies from site to site. At low pressure the burners underperform — so the same range, built to the same drawing, cooks differently depending on the building it lands in. And at the outlets where output was fine, cleaning and maintenance undid it. The range that ran well on Monday did not necessarily run well by Friday, depending on who had cleaned it.

One cause was the site. The other was the operator. The same two variables, in one piece of equipment.

We treated it as a specification problem rather than a supply one. Before anyone went looking for a supplier, we defined what the burner had to hold and what the cleaning had to do without operator judgment. Then we found a manufacturer whose burner sustains output at low gas pressure, and whose range carries a built-in water-based self-cleaning system. We proved it at one outlet, then specified it across the estate.

Both failure modes went. The burner no longer depends on the gas pressure at the site. The cleaning no longer depends on who is on shift. Four outlets carry it and a fifth is in progress — and it is still sourced through us on outlets built by contractors who are not us.

That last part is the real outcome. What you author outlives the contract it was authored under.

When the equipment doesn't exist: a grill mark without gas

An international grill QSR brand could not open in airports or petrol stations. Gas cooking is not permitted in either, and the brand had only ever grilled on gas. Those site types were simply closed to them.

The obvious answer was electric, and it nearly worked. The electric alternative held heat well enough. What it could not do was produce the grill marks that define the product — not a cosmetic detail for a brand whose entire proposition is flame-grilled chicken. The griddle architecture was wrong for it.

The manufacturer could not solve it.

We worked backwards from a chicken cooked on gas — starting with the result and asking what surface geometry and heat distribution would reproduce it — and through several iterations developed a new grill top, built to sit on the electric unit's existing heating elements in place of the standard top. Not a new machine. One component, replacing one component.

The brand and its franchisors passed it on taste and feel in 2023, and it opened site types the brand could not previously take. In 2026 their London team hit the same wall, and we shipped them a prototype from Dubai.

A component solved once in Dubai, shipped to another market three years later, the moment the same problem appeared there.

When the person is the variable: taking a task off the operator

An Indian food brand ran a porotta production line fed by three oil tanks, all drawing from one main tank. Operators watched the levels and refilled each tank by hand.

It cost time. It spilled. And a tank running low could interrupt a production run — which is the expensive version of the problem, because the cost is not the oil, it is the stopped line.

We designed and built the replenishment onto their existing line: level sensors in each tank, tied to the transfer pump already on site. When a tank drops to its set level, the pump feeds that tank. Manual topping up is largely gone, and the spills with it. Production no longer pauses for a level check.

It is worth being precise about the scope here, because this kind of work is easy to overstate. We did not build their bread line. That was theirs. We took one task off the person running it.

That is a modest intervention, and it is deliberately described as one. But it is the right shape for the problem: the task did not need a better-trained operator, it needed to stop being a task.

Writing it down: eight kitchens, one specification

Cloud kitchens carry the same multi-brand, high-throughput pressure as a dine-in kitchen, without the space or the staffing depth to absorb inconsistency between locations. When something is wrong in a cloud kitchen there is less room, in every sense, to work around it.

A regional delivery platform was opening eight of them. Eight sites solving the same brief eight ways is not a rollout — it is eight different kitchens that happen to share a logo.

We authored one kitchen standard and rolled it out across all eight locations. Layout, workflow and specification held constant regardless of the site, with only the geometry adapting to each unit. That distinction is the whole discipline: the shape of the room is allowed to change, the way the kitchen works is not.

Eight kitchens running to one standard rather than eight interpretations of the same brief, delivered over a two-year rollout.

What the four have in common

None of these were fixed by asking people to try harder.

In each case the failure had been read as an operational problem — inconsistent cooking, a messy station, a line that stops, sites that drift apart — and in each case it was actually a specification problem that had been pushed downstream onto whoever was standing in the kitchen. The wok range was a procurement decision made before anyone knew what the burner had to hold. The grill marks were a constraint nobody had taken back to the equipment. The oil top-up was a task that existed only because no one had designed it away.

This is the argument for involving whoever builds your kitchens earlier than the build. By the time a kitchen is being fitted out, most of what will make it consistent or inconsistent across twenty sites has already been decided — in the equipment schedule, in the specification, in what was assumed about the building.

It is also the argument for treating a recurring fault as a diagnosis job. A problem that shows up differently at every outlet is the hardest kind to see and the most valuable kind to solve, because nobody has solved it yet.

A note on what is not claimed here

Two of these were developed with the client's own teams and signed off by them. One was a single feature added to a production line we did not build. We have described the scope of our work in each and no more than that, and the brands are unnamed deliberately.

If you are planning a rollout and something keeps going wrong in a way you cannot pin down, that is usually worth a conversation before the next site is specified rather than after it.