A radius, a drive time, and a walk time drawn at the same address measure three different things, so a shortlist that mixes them ranks methods instead of sites. Pick one definition per store format, write down the settings behind it, and hold both constant for every site you compare.
Most teams already know a circle is a rough tool. The problem in this piece is different, and it sticks around even after you fix that. Everyone agrees to switch to drive times, and the definition still keeps sliding around between sites without anyone noticing. By the time the shortlist reaches the committee, it isn't a fair fight anymore.
What each method actually measures
Each method is answering a different question. A radius asks who's nearby as the crow flies. A drive time asks who can get to the site along actual roads within a set number of minutes. A walk time asks that same question of the sidewalk network instead. You end up with three different populations, not three estimates of the same one.
We cover the full rundown of the five delineation methods in common use, gravity models and customer-derived boundaries included, in our trade area analysis guide. This piece picks up after you've already chosen one.
What the three methods return at one real address
We ran all three at one downtown Pittsburgh address in GrowthFactor in September 2026, using the platform's default demographic preset. The point isn't that the walk-time zone comes back smaller. It's that the walk-time zone is a different place altogether.
| Measure | 3-mile ring | 10-minute drive | 10-minute walk |
|---|---|---|---|
| Population | 161,724 | 132,259 | 5,170 |
| Median household income | $64,777 | $66,047 | $109,641 |
| Graduate or professional degree | 22.7% | 22.7% | 43.6% |
| Median home value | $241,321 | $245,857 | $521,516 |
| Average household size | 1.93 | 1.92 | 1.41 |
The ring and the drive time sit close on every demographic line, and the ring counts about 22% more people. Look at just those two columns and you'd conclude the method barely matters here.
The walk-time column is a different market entirely. Median household income runs 69% higher than the ring, the share of residents with a graduate degree is roughly double, and average household size drops by half a person. That's the residential core of a downtown, and it looks nothing like the metro around it. Underwrite a concept against the ring, then open it against the walk-up trade, and you're aiming at the wrong customer. No amount of precision inside the wrong boundary catches that.
One address is an illustration, not a study. What it illustrates is the mechanism: the boundary decides which population you're describing, and swapping boundaries can swap the customer.
The gap is widest exactly where it matters most
Divergence between methods shows up worst at small geography. The City of Hanford, California publishes a demographic comparison for its own commercial sites that makes this easy to see, because it prints both methods at several sizes side by side.
At the tight end, a 3-mile radius counts 55,555 people while a 5-minute drive time counts 21,343, a gap of about 2.6 times. At the wide end, the same document shows a 10-mile radius at 104,619 and a 15-minute drive time at 103,158, a difference of under 1.5% (City of Hanford, municipal economic development document, June 2023).
That pattern is worth sitting with. The convenience, quick-service, and urban-infill deals, the ones where a few blocks decide the outcome, are exactly the deals where the two methods disagree most. The regional big-box deals, where teams tend to fuss over methodology the most, are where the methods already nearly agree. Method discipline pays off in inverse proportion to how much attention it usually gets.
Circles overstate reach, and the research says by how much
The academic work here comes mostly out of public health, not retail, because researchers studying walkability ran into the same measurement problem and had reason to quantify it.
A 2022 study using Multi-Ethnic Study of Atherosclerosis data across six US sites compared circular buffers against network buffers of the same nominal size. At a quarter-kilometre scale, the circle captured a median of 2.8 times more walkable destinations than the route-based boundary. At one kilometre the ratio dropped to 1.8, and at five kilometres it fell to 1.5. The measurement choice also biased the measured relationship between the built environment and actual walking behaviour downward, by roughly 20% to 80% across the six sites (Li et al., International Journal of Health Geographics, September 2022).
An earlier study in the same journal found the geometry changes the composition, not just the size. Comparing circular and network buffers around the same Vancouver-area respondents, residential land made up 51.25% of the circular buffer and 64.94% of the network buffer, while park and recreational land fell from 11.81% to 4.47%. One land-use association that looked statistically insignificant under the circle turned significant under the network boundary (Oliver, Schuurman and Hall, International Journal of Health Geographics, September 2007).
Both findings point the same way. A circle drawn at a walkable scale isn't a conservative approximation of a walkshed. It's a systematically larger area with a different mix inside it, and the smaller the boundary, the worse the overstatement gets.
Your isochrone tool has a traffic default you probably have not checked
Most teams never audit this part. "Drive time" isn't one setting. Whether traffic gets modelled at all, and which traffic, is a per-vendor default, and the major providers don't agree with each other.
- Esri's Generate Drive Time Trade Areas tool treats Time of Day as optional, described only as the time and date used when calculating distance. The page never says what the tool falls back to when you leave it blank, so you cannot tell from the documentation alone whether traffic reached your polygon (Esri ArcGIS Pro documentation).
- Google's Isochrones API, which entered public preview in July 2026, defaults to a traffic-unaware routing preference that Google describes as producing "a deterministic shape that does not fluctuate based on the time of day" (Google Maps Platform documentation).
- Mapbox's Isochrone API hides the decision in the profile string. Ask for
mapbox/drivingand you get average conditions; ask formapbox/driving-trafficand you get current and historic traffic. One word in the request separates them (Mapbox documentation). - HERE's Isoline Routing API goes the other direction. Send no departure time and it uses the current time, which pulls traffic in; you have to pass
departureTime=anyto switch it off (HERE developer documentation). - TravelTime exposes a traffic model parameter set to pessimistic, balanced, or optimistic, so the congestion assumption is something you dial rather than something you inherit (TravelTime documentation).
Two analysts on two platforms, both leaving the defaults alone, end up with different polygons for the same address and the same stated method. Neither one did anything wrong. Neither one can even tell from the output that it happened.
We looked for a credible published figure on how much a drive-time polygon shrinks at rush hour and couldn't find one that traces back to real measurement. The number floating around the trade press comes from a vendor page that says a peak contour "can shrink by close to a third," phrased as an illustration with no dataset behind it, which a later article then restated as a precise measured percentage. We've stopped citing the precise version. We'd rather show the mechanism through the vendor defaults above than quote a statistic that falls apart the moment you chase it.
A trade area number is four decisions, not one
Method is the decision teams argue about. It's only one of four, and the other three move without anyone noticing.
The threshold sets the size. The traffic assumption sets the shape. The data vintage sets which year's population you're counting. Change any one of these between two sites and the comparison quietly stops being a comparison, even though both rows on the shortlist still read "drive time" and look like the same measurement.
This is the same failure that makes cannibalization estimates drift between analyses. Overlap between two trade areas is entirely a function of how both were drawn, so a cannibalization number inherits every one of those four decisions twice over.
When walk time is the right instrument
Walk time is the right call when customers arrive on foot, and the wrong call everywhere else. Downtown cores, transit-adjacent sites, dense mixed-use districts, and small-format urban stores earn it. Suburban and highway retail don't, and forcing a walkshed onto them hands you a catchment too small to underwrite.
Two things are worth knowing before you use one. The pedestrian network is sparser than the road network, so a walkshed ends up a much smaller fraction of its nominal circle than a drive-time polygon is of its own, which is exactly what the buffer research above measures. And thresholds aren't standardized: the five-minute quarter-mile and ten-minute half-mile conventions in wide use trace back to neighbourhood planning, not to retail, and Walk Score's methodology awards full credit inside a quarter mile and decays to zero around a thirty-minute walk.
Pick a threshold that matches your format, then leave it alone. A walk time that changes between an urban site and its comparison set does the same damage as switching methods outright.
How to lock a definition your portfolio can live with
There's no industry standard to point to here. We looked for published guidance from the major retail real estate bodies requiring that trade area methodology be disclosed or held constant in comparative analysis, and didn't find one. Appraisal and feasibility work gets closest, since a lender or a reviewer can always ask you to defend the method behind a conclusion, but no published standard we found names trade area delineation specifically. The discipline is yours to impose.
Four rules cover most of it.
- Set the definition by store format, not by site. One definition for urban small-format, one for suburban inline, one for highway. Sites within the same format are competing against each other, so they need to be measured with the same instrument.
- Record all four settings with every number. Method, threshold, traffic assumption, and data vintage need to travel with the population figure, or that figure won't be reusable next quarter.
- Calibrate against real customers where you have them. For an existing store, actual customer origins tell you which threshold reproduces your real trade area. That calibration is what makes the definition defensible for the sites where you don't have customers yet. In custom modelling work we've done, right-sizing trade areas from actual customer data rather than arbitrary ring studies changed which variables mattered, and several conventional assumptions turned out to run in the opposite direction.
- Re-run the incumbent, do not trust the old number. When a site approved two years ago becomes the benchmark for a new one, re-measure it under today's definition before you compare. It's cheap, and it's the step most often skipped.
None of this makes a boundary correct. It makes a set of boundaries comparable, which is the only property a shortlist actually needs. A trade area that's honestly drawn and consistently drawn beats a cleverer one applied unevenly, the same way foot traffic comparisons only mean something inside a consistent category.
Keeping four settings attached to every site across a growing pipeline is more a record-keeping problem than an analytical one. GrowthFactor's market planning workflow keeps the trade area definition on the site record alongside the score and the deal status, so the settings behind a number are still visible when someone asks six months later where it came from.
Frequently Asked Questions about Trade Area Definitions
What is the difference between a radius and a drive-time trade area?
A radius measures straight-line distance from the site and produces a circle. A drive-time trade area measures travel along the road network and produces an irregular polygon that stops at rivers, freeways without exits, and rail lines. The circle is faster to produce and easier for outside parties to accept. The polygon is closer to how customers actually reach the store, which is why the two can return very different populations at the same address.
Can I compare two sites if they use different trade area methods?
No, not directly. Population, income, and competitor counts all inherit the boundary they were measured inside, so a site measured on a 3-mile ring and a site measured on a 10-minute drive time are answering different questions. If the two numbers sit in the same column of a shortlist, the ranking partly reflects the method rather than the sites. Re-run both under one definition before comparing them.
When should I use a walk-time trade area instead of a drive time?
Use walk time when most customers arrive on foot: downtown cores, transit-adjacent sites, dense mixed-use districts, and small-format urban stores. Walk time is the wrong instrument for suburban and highway retail, where it will return a catchment far too small to underwrite. The test is how your customers actually get there, not how dense the surrounding area looks on a map.
Does the time of day change a drive-time trade area?
It can, and the tool you use decides whether it does at all. Some isochrone services default to traffic-free routing, which returns the same polygon regardless of when you run it. Others use live or historical traffic by default, so the same request returns different shapes at different hours. Two analysts on two platforms, both leaving the defaults alone, can produce different boundaries for the same address.
How does GrowthFactor compare to Kalibrate for trade area analysis?
Kalibrate has deep roots in fuel and convenience retail, with catchment modeling and demand forecasting built around that sector and delivered largely as consulting-led engagements. GrowthFactor is a self-service platform where trade area definition sits inside the same workflow as site scoring, demographics, and deal tracking, so the boundary settings behind a number stay attached to the site record. Teams can re-run a site under a different definition in seconds rather than commissioning new work.
The version that matters
The interesting question about a trade area isn't whether the boundary is right. Every boundary is wrong somewhere. The real question is whether the boundary around site A is the same instrument as the boundary around site B, because that's the assumption the whole shortlist rests on, and almost nobody writes it down.
Pick the definition once, per format. Write down what it was. Re-measure the incumbent when it becomes a benchmark. The comparisons get honest, and the arguments in committee move on to the sites, which is where they belong. Our companion guide to trade area mapping methods covers the definitional groundwork if you're setting a standard from scratch.