The retail site selection process: from trade area to approval brief

Expansion mistakes usually trace back to an inconsistent process. One deal gets scored on drive-time and demographics; the next rides on a broker pitch and a gut read. When the criteria shift from deal to deal, a strong location and a confident champion start to look the same on paper, and the stores that underperform teach you nothing.

A repeatable process closes that gap. Every candidate runs through identical criteria and a single scoring model, so sites become comparable and the reasoning behind each call gets recorded. Judgment still drives the decision. The process makes it explicit, which lets the committee argue about the weights and the evidence instead of rebuilding the analysis each time.

The eight steps below move a location from a name on a map to a decision record. You can run the early steps across dozens of markets and reserve the deep work for the short list, but the sequence holds every time.

Before any address gets scored, agree on what a good site means for this concept. Start with strategy: who the customer is, and how a new unit is supposed to make money. Criteria then split into two kinds, and keeping them apart is what keeps the rest of the process honest.

• Gates. Pass-or-fail rules that disqualify a site regardless of how it scores elsewhere: a minimum trade-area population, a rent ceiling, required visibility, a zoning requirement. They stop a high weighted total from papering over a dealbreaker.
• Weights. The relative importance of whatever factors remain, from daytime population and household income to competitor density and access. Set them once, in the open, and resist the urge to re-tune them around a deal you already like.

Settling gates and weights up front keeps the later score grounded in the criteria rather than in who presents the deal. In Geod, the criteria and weights live in the model itself, so every later step inherits the same definition of a good site.

With criteria set, widen the funnel before narrowing it. Market screening scans metros and submarkets for the places where customer density and open competitive space justify a closer look. Treat it as a ranking pass rather than a verdict. The point is to drop the obviously weak areas before anyone spends field time on them.

A good screen ends with a short list and a reason attached to each entry, whether that is strong demographics or an underserved pocket the competition has missed. Everything downstream costs less when the list reaching Step 3 already holds only the places worth deeper analysis.

A site is only as good as the people who can actually reach it, which makes the trade area the foundation under every later number. A radius or a ZIP ignores how customers travel. A river, or a highway with no nearby exit, can leave a household close on the map and far away on the road.

Build the boundary from the actual road and pedestrian network. A drive-time area captures who can reach the site within a realistic commute; a walk-time area does the same job in dense, pedestrian-first locations. Match the method to the format, so a destination concept leans on drive-time and an urban grab-and-go leans on walk-time. Whichever you pick, that boundary defines who counts in the demand and competition figures that follow.

With a real boundary in place, fill it with evidence. Two questions decide whether the location can carry a unit: how much demand sits inside the trade area, and how much of that demand is already spoken for.

• Demand. Pull demographics scoped to the trade area instead of the surrounding county: resident and daytime population, household income, age mix, plus any concept-specific drivers. Because the numbers stop at the travel-time boundary, they describe the people who can genuinely reach the site.
• Competition. Map direct and indirect competitors inside the same boundary. Their count and proximity show how contested the area already is, and whether enough unserved demand remains to carry another unit.

Scoping both to the same trade area keeps demand and competition on equal footing. Geod attaches the source and vintage to each figure, so a reviewer can see where a number came from and how current it is.

For a multi-unit operator, a strong site is only half the question. The other half is whether it adds business or just relocates it. A candidate that looks excellent on its own can quietly pull sales from the store two miles away, leaving the real network gain well below what the site score implies.

Cannibalization analysis lays the new trade area over your existing units and estimates how much of its demand would come at their expense. What remains is net-new demand, the genuinely additive share once the overlap with your own network is removed. Geod models this with network-gravity cannibalization, so the overlap surfaces while you can still walk away from the lease, rather than months later in comparable-store sales.

Now the evidence collapses into one comparable number. The score applies the Step 1 weights to the demand and competition from Step 4 and the net-new demand from Step 5, then checks the gates. A site that trips a gate gets flagged no matter how high its weighted total runs.

The score has to be explainable. A number nobody can take apart hides the reasoning a committee needs to challenge, so it earns little trust around the table. An explainable score traces back to its named components, so a reviewer can see exactly how much each one contributed, from access to demographics to competition, and why one site sits above another. Geod produces a glass-box score of this kind, and because the model is shared, any two sites get graded the same way.

The brief is where the analysis turns into a decision. It gathers the whole workflow into one document a real estate or investment committee can read in a single sitting: the trade-area map, the demand and competition figures with their sources and vintages, the cannibalization estimate, and the explainable score broken out by component.

A consistent format lets the committee compare deals on the same terms rather than on slide quality, and it keeps the discussion on the weights and assumptions behind each score. Geod exports the PDF brief from the same model that produced the score, so the document on the table matches the analysis behind it.

An unrecorded decision teaches you nothing later. Whenever the committee approves, defers, or passes, store that outcome next to the brief and the score it rested on, along with the evidence and the assumptions that drove it. Months on, when someone asks why a site got approved, that record is what lets you answer.

The loop closes after the doors open. Compare actual performance against the demand and the score the model predicted, then feed the gap back into the Step 1 gates and weights. That is how the criteria improve over time, shifting from educated guesses toward what your own stores have proven about which factors predict success.

The payoff comes from running the same steps on every deal, over shared data, until a site that looks good and a site that scores well are the same site. Run the early steps wide and the later steps deep. Keep the gates and weights honest, and write down the score and the decision both. Done consistently, that turns site selection from a stack of one-off arguments into a process you can defend and improve on.