Kill the tablet wall

It is 8:40 on a Friday and the pass is glowing. Four tablets in a row, propped against the wall over the line: Uber Eats, Wolt, Bolt Food, Deliveroo, each chirping its own alarm on its own schedule. One buzzes, then a second, then the first again because nobody has tapped "accept" yet. A runner leans in, reads an order off the Wolt screen, and re-types it into the POS by hand so the kitchen printer will fire. Halfway through, the Bolt Food tablet lights up and he loses his place.

By 9:15 the kitchen is behind. Not because the cooks are slow — because the front of the pass has become a switchboard, and the person working it is watching four apps instead of expediting food. This is the tablet wall. Almost every busy takeaway kitchen has one, and almost everyone treats it as normal.

The wall is not a hardware problem

It is tempting to blame the tablets. Buy a bigger shelf, get a louder ringer, put one person on "tablet duty." None of that fixes anything, because the problem is not the four screens. The problem is the four separate systems behind them.

Each marketplace runs its own app, its own queue, its own timer, its own back office. Nothing talks to anything else. So a human being has to stand there and be the integration: watch every screen, accept every order, re-key each one into the POS, and reconcile it all against what the kitchen actually made. That work does not show up on any invoice, but it is real, and it fails at exactly the moment you can least afford it — the Friday rush.

Where the tickets get missed

Watch the wall long enough and you see the same three failures, over and over.

Missed tickets. A notification comes in on the Wolt tablet at the exact second the runner is re-typing a Bolt Food order. He does not hear it for ninety seconds. That delay in noticing one tablet pushes out the estimated delivery time on that order — the courier is already assigned, the customer is already watching the clock, and now the food starts late.

Duplicated orders. In the scramble, the same ticket gets keyed into the POS twice, and the kitchen makes it twice. One of the two walks out the door as waste.

Manual entry errors. A note on the Yemeksepeti order — "no onions" — gets skipped in the re-typing, because the runner's eyes are on the next alarm. The wrong item is made. It comes back. Now the line is remaking a plate during the busiest hour of the week.

None of these are exotic. They are the ordinary, predictable output of asking a person to hand-copy orders between systems under time pressure. The wall does not fail on a quiet Tuesday. It fails on Friday at 9pm, when a missed ticket costs you a table, a courier, and a review.

One queue, fed by every channel

The fix is not a bigger wall. It is no wall. Every channel — dine-in, takeaway, own web ordering, and every marketplace — feeds one order queue on one screen. An order does not get re-typed; it arrives, already parsed, and routes straight to the right station. The runner stops being a switchboard and goes back to running food. The cook does not care which tablet an order came from, and now he does not have to — he just reads the ticket in front of him.

And the count only goes one way. The average restaurant is already on about 2.5 delivery platforms, and climbing. Add a fifth marketplace to the wall and you have added a fifth alarm, a fifth queue, a fifth thing to re-key at 9pm. Add a fifth channel to one queue and you have added one more line to a screen that was already handling the load.

You do not have to leave the marketplaces to kill the wall — those apps are demand you want. You just stop asking a person to be the glue between them. When order management puts every channel in one queue and routes each ticket to the right station on its own, Friday at 9pm stops being a switchboard shift and goes back to being service.