Retail site selection as a process: from city screen to committee pack
Retail site selection as a repeatable process: screen markets, rank trade areas, score sites, check the network, then decide. How to run each stage.
Part of the guide AI employee for site selection
For a chain opening several stores a year, retail site selection is a process, not a single analysis: screen the cities worth entering, rank trade areas inside each, score the sites the property team finds, check what each new store does to the stores you already have, and put the survivors in front of an investment committee with the same figures every time. Teams that run it as a funnel with clear gates open fewer weak stores, because every site is judged on the same evidence and the reasons for a no are written down.
This is for expansion managers, real estate leads and analysts at retail and food chains. It describes each stage, what passes a site to the next, a made-up quarter for a convenience chain in Bangkok, the habits that undermine the process, and how an AI employee keeps the evidence flowing so analysts spend their time on judgement.
The five stages
- Market screen. Which cities or regions are worth entering this year? Count your competitors, the anchors your customers use, and residents where population data exists. A city where rivals are thin and the anchors are plentiful moves up.
- Trade area ranking. Inside a chosen city, rank districts or drive-time areas on the same measures, plus how many of your target shoppers they hold. The output is a short list of areas the property team should search.
- Site scoring. For each unit or plot the agents offer, measure the area customers really reach, the competition in it, the anchors near the door and how the roads move at the busy hours. Score it against your best existing stores.
- Network check. How much of the new store's area overlaps an existing store's? A strong site that mostly takes sales from your own shop down the road adds little.
- Committee decision. The same pack for every site: the map, the figures, the comparison with similar stores, the risks, and the analyst's recommendation.
Each stage has a gate. Write the rule down: for example, no site goes to scoring unless it sits in a ranked trade area, and none goes to committee if more than a third of its drive-time area overlaps an existing store. The exact thresholds are yours; what matters is that they are agreed before the sites arrive.
What each stage needs from data, and what it does not
- Places and competitors come from a places database: counts, chains and independents, ratings and when places opened.
- Residents come from a population grid in the countries that have one (France, Canada, India, Thailand, Egypt and Australia among them; not the United Kingdom or the United States).
- Reach is a travel area along the roads, by car, on foot or by bike.
- Roads are live speeds and congestion measured at the hours you choose; they are not counts of cars.
- Busyness of an area can be read from footfall, counted as people-hours of presence and arriving about ten days late, in the cities it covers.
- What data cannot give you: rent, the landlord, the condition of the unit, planning rules, and how a neighbour's renovation will change the street. Those come from the property team and a visit.
A made-up quarter in Bangkok
This example is made up. A convenience chain plans twelve openings in Bangkok over a year. In the market screen the city passes easily, so the team goes straight to trade areas: it ranks districts on residents within a 10-minute walk of each main road, the number of rival convenience stores, and offices, schools and transit stops as anchors. Nine districts make the list.
Over the quarter, agents send 41 units in those districts. Site scoring drops 23 on weak anchors or saturated competition. The network check drops 6 more whose walking area overlaps an existing store by more than a third. Twelve reach committee, each with the same two-page pack; eight are approved, two are sent back for a rent renegotiation and two are declined because a visit found the frontage hidden by a footbridge. The declined reasons go into the team's notes so the same mistake is caught earlier next time.
Habits that undermine the process
- Skipping the gates for a favourite site. The site a senior person likes is the one that most needs the same scrutiny.
- Changing the measures site by site. If one site is judged on a 10-minute walk and the next on a 2 km circle, the committee is comparing different things.
- Scoring without a benchmark. A score means little unless you know what your best and weakest stores would score on the same rules.
- Forgetting the network. Overlap with your own stores is the most common reason a good-looking site disappoints.
- Letting the evidence go stale. Competitors open and close every month. A pack built in spring may be wrong by autumn.
Keeping the evidence flowing with an AI employee
In xMap Flows, AI employees can do the counting at every stage so the team's time goes on judgement and negotiation. The overview of the jobs is in the site selection guide; a typical set-up for site scoring looks like this.
- New rows: the property team adds each offered unit to a sheet in Google Sheets or a table in My data.
- Travel area: the reach your format uses, such as a 10-minute walk.
- Demographics, Nearby places and Traffic Flow: residents, rivals, anchors and road speed for each unit.
- Find nearest against your store list: how close each unit is to your existing stores, for the network check.
- xMap LLM: score each unit against your rules and say which gate it fails, if any.
- Approval, then Make a report: the analyst approves the units going to committee and the pack is built.
Ready-made templates cover other stages: expansion committee pack, every month, where my rival is and I am not for trade area screening, and stores competing with each other for the network check.
Questions
What is the difference between a trade area and a site?
A trade area is the district or drive-time zone a store would draw customers from; a site is the specific unit or plot inside it. Rank trade areas first, then score sites within the good ones.
How do I set the thresholds for each gate?
Score your existing stores with the same rules and see where the good and weak ones fall. Set each gate near the line that separates them, and revisit it once a year.
How often should site scores be refreshed?
Before every committee meeting at least. Competitor openings and closures change scores, so a monthly refresh of live candidates is sensible.
Can the same process work for a franchise network?
Yes, with one more step: checking the territory the franchisee will hold and that it does not cut into a neighbour's.
Templates that do this
Read next
More on site selection and expansion · Every article · Templates