How to Choose a Restaurant Location: A Site-Selection Framework

Choosing a restaurant location comes down to one question: can this address carry the sales your concept needs, on the worst plausible month, with the lease you are about to sign? Most advice starts with footfall counts and rent ratios. A better method starts with what can kill the site outright, then looks at the street by the hour, and only then does the arithmetic.

A site is a bet on sales you cannot yet see. The framework below is built to make that bet smaller.

Why do most location checklists fail?

They rank sites before they have removed the ones that cannot work. A scorecard that gives a charming corner unit 4 out of 5 for "visibility" will happily rank it first, even if the building cannot take a kitchen extract or the use class forbids a restaurant.

Order matters. Use three passes:

1. Gates. Pass or fail, no scoring.
2. Street audit. Observe the place at the hours you will trade.
3. Lease stress test. Do the rent maths against a bad month, not a plan.

Rank only what survives all three. This keeps the ranking clean, because a weighted score can hide a fatal flaw behind three good numbers.

Which gates should remove a site immediately?

Any site that cannot legally or physically host your kitchen should be dropped before you visit it a second time. Four gates do most of the work:

• Permitted use. Is the unit already approved for food service, or would you need a change of use? Rules differ widely. In England, for example, the Planning Portal notes that a shop and a cafe share a use class, while a pub or takeaway sits in a class of its own and needs permission. Elsewhere the categories and the process will be different.
• Extraction and utilities. The same guidance notes that a commercial flue for cooking can itself need planning permission. Ask where the duct can run, whether the landlord and neighbours will accept it, and what power, gas and drainage the unit already has. Our commercial kitchen requirements checklist covers the questions to put to engineers.
• Licences. If alcohol, late hours or outdoor seating are central to the model, check whether the street and the building allow them. Pavement tables on a public highway usually need separate permission.
• Access and servicing. Where does a delivery van stop, where does waste go, and can a trolley reach the kitchen without crossing the dining room?

Write the answers down. A gate you "assumed" is a gate you have not tested. This is general information, not legal or tax advice. Check current local rules with the planning authority and a qualified adviser before you commit to any lease.

Before the gates, fix the concept's needs on paper: hours, average spend, seat count, kitchen demands and the guest you want. A site cannot be judged against a concept that is still vague. If you are unsure of those, work through concept development first, because every later number in this framework depends on it.

How do you audit a street by daypart?

Visit at the hours you plan to sell, because the same street can be two different markets across a day. A tourist artery that is full at 2pm may be empty at 8pm on a Tuesday. An office street that feeds a lunch trade can be silent after seven.

Footfall analytics firm MyTraffic makes the point well in its site selection guide: volume alone says little about fit, and a street that gains 81% on weekends is a poor match for a weekday lunch concept. Treat that as a reason to look at the shape of the week, not just its total. (It is a firm that sells footfall data, so read its examples as illustrations of method, not as benchmarks.)

A simple audit sheet, filled in on site, beats a dashboard you cannot check:

Do it three times, on different days, ideally including a wet one. Count what you see for fifteen minutes at each hour. If you cannot stand on the pavement for that long, you are not ready to commit to the lease for ten years.

Is a busy street always better?

No. Busy streets cost more, and the crowd is not always your crowd. Competition cuts both ways: a cluster of good restaurants can create a dining destination, while a street with your exact offer already serving your exact guest is crowded in the way that hurts.

Ask two separate questions about nearby competitors. First, does anyone already serve this guest with this offer at this price? Second, do the neighbours bring people to the street who would not come otherwise? A coffee roaster next door may feed your breakfast trade. Another place doing the same thing in the same room probably will not.

The quieter option can win on the numbers. A destination concept with a loyal following, a reservation book and a clear reason to travel often does better on a cheaper side street than a walk-in concept does on the high street. The test is whether guests will find you without help from passing trade.

How should you test the rent?

Run the rent against a downside case, not against the sales you hope for. Rent is a fixed amount. Sales are an estimate. The ratio between them changes with every month you trade, so a single percentage on a spreadsheet tells you very little.

A widely repeated rule says rent should be 6 to 10% of sales. One analysis of the rule could not trace it to any survey or dataset, and shows why a ratio misleads: the same monthly rent of 9,000 is 15% of sales at 60,000 a month and 6% at 150,000. (This is a commercial adviser's blog, so treat its figures as illustration.)

What published data exist are modest and context-bound. The National Restaurant Association's 2025 operations data, based on more than 900 US operators, put median occupancy costs in 2024 at 5.7% of sales for full-service restaurants and 5.2% for limited-service. In city centres the median was 6.0% for both. These are medians of businesses that survived to report, in one country, under varied lease terms. They are context, not a target.

Use a stress test instead:

1. Estimate monthly sales for the site, using your seats, turns and average spend.
2. Cut that figure by 30%.
3. Add the full occupancy bill: base rent, service charges, rates or property taxes, and insurance.
4. Ask whether the business still covers its costs and pays the people running it.

Take a hypothetical 40-seat bistro, counted in units rather than a currency, with rent and charges of 100 units a month and planned sales of 1,500. Rent is 6.7% of sales, which looks comfortable. At a 30% miss, sales are 1,050 and the same bill is 9.5%. If food and labour already use most of the rest, the margin is thin. The site is not necessarily wrong, but the lease must give you room to survive that month.

Sign the lease you could still pay in your worst month.

Which lease clauses decide whether you survive?

The clauses that matter are the ones that bend when sales fall short. The same Katalyst piece suggests asking, of each term, "if sales come in 30% under plan, which clauses decide whether you survive it?" Points worth raising with your adviser include:

• When rent starts. Tie it to permits being granted, not just to the keys being handed over. Fit-out periods of 90 to 150 days are typical in the US examples it cites.
• Percentage rent and breakpoints. If rent includes a share of sales above a threshold, find out exactly where that threshold sits and how "sales" is defined, including delivery orders, discounts and gift vouchers.
• Personal guarantees. Ask whether a cap, a time limit or a step-down is possible.
• Assignment. If the business fails or sells, can you pass the lease on, and on what conditions?
• Service charges and exclusives. Look for caps, audit rights and protection from a competitor moving next door.

Terms vary by country and landlord, and many are negotiable. Our guide to restaurant lease red flags goes through these in more depth. Have a solicitor or equivalent adviser read the lease before you sign.

What do you do when two sites tie?

Pick the one with the better downside, not the better upside. If two sites survive the gates, the audit and the stress test, the choice usually comes down to a few items:

• Fit-out cost. A site that is already a restaurant often costs far less to open than a shell, even at a higher rent.
• Flexibility. Shorter terms, break clauses and an option to renew reduce the cost of being wrong.
• Concept fit. The room should suit the experience you are selling. Our hospitality design brief guide shows how to describe what the space must do before you start looking.
• Your own presence. A site you can pass every day is easier to run well than one an hour away.
• Neighbours. Talk to the people who already trade nearby. Ask when it is busy, how the landlord behaves, and what changed in the last two years. Ten minutes with a café manager next door often tells you more than a report.

If a scorecard's totals are within a small margin, treat them as equal. Differences of a few tenths on an invented scale are noise. The real information is in the gates and the stress test.

Key takeaways

• Remove sites that fail permitted use, extraction, licensing or access before you score anything.
• Audit the street at the hours you will trade, across several days, and note who passes, not just how many.
• Do not trust a single rent-to-sales ratio. The 6 to 10% rule has no traceable source, and published medians are only context.
• Stress-test every finalist with sales 30% under plan and the full occupancy bill.
• Read the lease for the clauses that bend when sales fall: rent start, percentage rent, guarantees and assignment.
• When sites tie, choose the one with the better worst case.