Demographic site analysis in site selection

Demographic site analysis studies the people near a candidate location and asks whether they match the customer a business depends on. The raw material is familiar: population counts, household income, age distribution, and household composition, plus the spending and segment behavior tied to a specific category.

The useful output is a count of likely buyers among the people a site can reach. That number runs far below total residents nearby, and it is the one that tracks sales.

A frequent error in site selection is reading a population count as a demand figure. The site sits beside 80,000 residents, a slide labels the demographics strong, and the deal advances on that basis. Population describes a pool of people, and most of that pool may never walk through the door.

Demand is what survives a few filters: category fit, who can actually reach the door, the daypart the business runs on, and whether residents resemble buyers you already convert. A population count without those filters reports a headcount and leaves its link to sales to chance.

A neighborhood holds different populations depending on the hour. Residential population counts who sleeps there. Daytime population counts who is present during working hours, including commuters, office workers, and visitors who live elsewhere.

Which one matters depends on the business. A lunch-driven quick-service restaurant or a coffee bar runs on daytime and commuter traffic. A dinner house or a weekend grocery leans on residents. A dense business district can look thin on residential counts and still post enormous weekday volume, while a commuter suburb can look strong on residents and empty out by nine in the morning. Match the population to the daypart, or a site that reads well on paper underperforms once it opens.

Two trade areas can hold identical headcounts and still differ on demand. Household income separates them, and category spend separates them more precisely. What you want to know is how much the people who can reach the site spend on the goods you sell, measured against the category rather than a generic income band.

A premium concept needs disposable income concentrated within reach of the site, since a strong regional average can hide thin spending block by block. A value concept often does better where category spend holds steady across a broad middle. Once the dollars attach to specific products, income carries real predictive weight and earns its place beside the population count.

Few sites sell to the average resident, because the average is a blend that no real household matches. The question worth answering is how much of the reachable population resembles the customers you already convert. Households with young children, early-career renters, retirees, and long-commute professionals shop on different schedules and budgets, so a tight match on one segment can outweigh a modest headline population.

Segment fit is the step where demographics start to describe a customer model with predictive value. If your strongest stores over-index on one segment, the useful measure for a new site is the count of that segment inside the trade area, set against what your existing units already show.

Every demographic number inherits the boundary you used to collect it. A three-mile radius is a circle drawn with no regard for roads, rivers, highways, or rush hour. It counts households stranded across a freeway who will never cross it, and it skips households just past the ring who reach the door in five minutes.

A drive-time or walk-time trade area follows the actual road and pedestrian network, so the population, income, and segment counts inside it describe people who can genuinely get to the site. The same address can report very different demographics under a circle versus a drive-time polygon. Reach decides who becomes a customer, which makes the drive-time figures the ones that track demand.

Demographics drift over time. Neighborhoods add households, grow older, and shift in income, so a figure that was accurate three years ago can mislead a committee today. Public sources like the Census and the American Community Survey carry authority but arrive on a lag, while modeled current-year and forward estimates rest on assumptions a reviewer should be able to inspect.

The habit that keeps this honest is recording the vintage. Every population, income, and segment figure should carry its source and the date it represents, so a reviewer can tell whether a number is current, modeled, or out of date. A figure with no vintage attached asks the committee to take it on faith.

Geod folds demographics into an explainable site score as one weighted input among several. It draws a drive-time or walk-time trade area around the candidate, reads population, income, and segment data inside that boundary, and filters toward the customer and category you specify. The output is a demand component you can inspect and adjust, sitting alongside competition and cannibalization against your own units.

Because the score is explainable, you can see how much the demand component contributed and raise or lower its weight when your read of the market differs. Every figure also exports into a site brief with its source and vintage attached, so the demographics a committee reviews arrive already filtered to your customer and stamped with a date.