How to plan a food hall that actually trades

1. Where does the footfall actually come from?

Before any layout is drawn, establish the real numbers: who passes the entrance, at what hours, and whether they arrived intending to eat or intending to shop. A food hall on the top floor of a department store is fed by shoppers travelling upward; a food hall in a transport interchange is fed by commuters with eight minutes. Those two sites support completely different offers, price points and service speeds.

Cross-check catchment data against comparable operations and against the capacity the site can physically serve. It is cheaper to test an optimistic sales assumption at feasibility stage than after the fit-out is paid for.

2. How long will people stay — and how many can you serve at peak?

Dwell time drives everything. If the average visit is 20 minutes, the seats turn three times an hour and the kitchen must be engineered for that rate. If the average visit is 75 minutes, the same floor serves a fraction of the covers and needs a higher average spend to work.

Run the peak-hour maths explicitly: covers per hour multiplied by seats, checked against kitchen throughput, servery throughput and payment throughput. The binding constraint is usually the last one, not the number of seats.

• Count seats against realistic dwell time, not floor area
• Test kitchen throughput at peak, not at average
• Model queuing at each point of payment
• Plan the delivery and waste routes before finalising the layout

3. What is the concession and own-offer mix?

Most food halls blend operator concessions with own-run food and retail. The mix should be decided commercially, not by who is available:

• Concessions bring specialist credibility and reduce capital exposure, but dilute margin and control of standards.
• Own-run units protect margin and consistency, but demand recruitment, training and management capacity.
• Retail and grocery usually has the strongest basket economics and the weakest experience — it works when it is merchandised as part of the journey rather than left at the exit.

For Tsvetnoy Central Market the food and beverage offer spanned three floors: a food market, café, two restaurants and a night club. Mixing daypart and format that widely only works when each unit has a clear reason to exist at that hour.

4. Can the back of house deliver it?

Every food hall has a throughput ceiling set by its kitchen and logistics, and it is invisible on the floor plan. Decide early whether production is centralised in an on-site kitchen facility or distributed per unit, and how goods arrive, are stored and are removed.

For the Eden Project we designed a full on-site central production kitchen capable of freshly producing all food for the site daily, serving up to 15,000 transactions a day across snack pods, counter service and table service. That only held together because production capacity was designed alongside the customer-facing formats rather than after them.

Common mistakes we are asked to correct

1. Seating counted from floor area rather than dwell time, producing a floor that looks full and serves too few covers.
2. Kitchens sized on average demand, failing on the first busy Saturday.
3. A retail section placed where footfall has already peaked, so it never converts.
4. No waste, delivery or staff route, so service corridors become storage.
5. Concession mix chosen by rent rather than by whether the offer completes the daypart.