Site selection criteria: a checklist for multi-unit operators

A site selection process has to answer two separate questions, and teams often run them together. One is whether a site is eligible at all. The other is which of the eligible sites ranks highest. Gates settle the first. A weighted score settles the second.

Keeping the two apart protects the budget. Score everything on a single number and a fatal flaw, say an impossible left-turn into the lot, can get outvoted by a strong population figure. The site looks fine on the scorecard right up until the deal collapses in due diligence. Run the gates first, drop the failures, and spend analysis time only on candidates that can actually be built and operated.

What follows is the checklist itself. Copy the gate list, copy the scored-criteria list, set your weights one time, and apply both to every candidate that reaches your desk.

• Confirm the use is permitted by right rather than dependent on a variance, conditional-use permit, or rezoning. Anything that needs a discretionary approval carries risk, and a use that is flatly not allowed ends the conversation.
• Customers have to physically get in and out. Check curb cuts, median breaks, left-turn access, and whether the parcel is reachable from the direction your traffic comes from. No workable access is a hard no.
• The site has to meet the parking ratio your format and the local code require. Shared-lot and reciprocal-easement arrangements count toward the total, but a site that cannot hit the minimum without a variance fails the gate for most formats.
• Your prototype has to fit the pad, including drive-thru stacking, loading, and setbacks. A pad that forces you off-prototype rewrites the unit economics, so a poor fit should knock the site out rather than just trim its score.
• For franchised or licensed networks, check whether the site sits inside another operator's protected radius before anything else. Encroachment and exclusivity terms decide this. A site you are not allowed to open is not a candidate.
• Availability, an asking rent inside your range, and an opening date inside your window all have to line up. A site you cannot afford or cannot deliver on schedule fails the gate however good the geography reads.

Once a site clears the gates, the scored criteria take over. Reach and demand cover who can get to the site and whether enough of them want what you sell.

• Reach: population and households inside a realistic trade area. Build the trade area from drive-time or walk-time on the road network instead of a flat radius, since roads, rivers, and highways decide who can actually reach the door.
• Demand: whether the people inside that trade area match your customer. Look at category spend, target-household share, daytime versus residential population, and the demographics that map to your concept, well past a raw headcount.
• Generators and anchors: nearby traffic drivers such as a grocery anchor, big-box co-tenancy, employment density, or a commuter corridor that pulls demand toward the site through the day.

Demand measures the size of the opportunity. What matters next is how much of that opportunity is already claimed, by rivals and by your own existing units.

• Competition: the count, proximity, and strength of competing locations inside the same trade area. More competitors usually pull the score down, though in some categories a cluster signals a market that has already proven itself.
• Cannibalization: how much of this trade area overlaps with stores you already run. A strong-looking candidate that mostly transfers sales from a unit two miles away adds little net growth, so the overlap should lower its score. Severe overlap is worth promoting to a gate.
• Net-new demand: what remains after you subtract demand already served by your own network and by rivals. This figure tracks incremental sales most closely, which is why competition and cannibalization sit on the scorecard instead of in a footnote.

Two sites with identical trade areas can perform very differently because of what happens in the last hundred feet. These criteria stay close to the parcel itself.

• Accessibility: how easy entry and exit are in practice, measured by number of access points, signalized intersections, turn lanes, and traffic flow at the hours that matter for your daypart.
• Visibility: whether drivers can see and read the site from the road, which depends on frontage, set-back, sign rights, and how much reaction time approaching drivers have to turn in.
• Co-tenancy and adjacency: complementary co-tenants lift a site, while a vacant or declining center weighs on it even when the demographics look strong.
• Physical and operational fit: drive-thru stacking, delivery and pickup staging, loading access, and anything else that shapes how the unit runs once it opens.

The two-bucket structure holds across formats, but what fills each bucket shifts with the concept. Treat these as starting points and tune them to your own.

• Retail: co-tenancy, anchor strength, and visibility carry heavy weight, and the trade area often runs a longer drive-time. Parking and frontage show up as common gates.
• Restaurant and QSR: daypart demand, drive-thru stacking, and ingress dominate, and the trade area stays tight. Drive-thru feasibility and stacking depth are frequently hard gates rather than scored factors.
• Healthcare and services: insured-population and age-band demand, referral patterns, and parking convenience matter most. Licensing, accessibility compliance, and proximity rules can act as gates that retail never encounters.

A weighted scorecard earns its keep through consistency. Decide once how much reach, demand, competition, and accessibility each matter to your concept, write the weights down, then run the identical model on every site. Weights that live in someone's head and drift from deal to deal turn the scorecard from evidence into an argument.

Calibrate the weights against units you already understand. Run a handful of your best and worst existing locations through the criteria and adjust until the model agrees with reality on the easy cases. Once it does, freeze the weights and let every new candidate inherit them.

• Write the gates as a fixed pass-or-fail list and run them before any scoring.
• Assign a weight to each scored criterion so the weights add up to a clean total.
• Score every surviving candidate on the same criteria and the same scale.
• Keep the component scores visible so a reviewer can trace why a site ranked where it did.
• Revisit the weights on a set schedule rather than deal by deal, so the model stays comparable over time.

Geod is built to run exactly this way. Gates filter out the ineligible sites, the survivors get an explainable score across Reach, Demand, Competition, and Accessibility with weights you control, and each candidate lands in an exportable PDF brief that shows the component contributions behind the number.