1. Why reach and accessibility need separate lines
A defensible site score has to separate two related questions that often get blended together. Reach asks who can realistically get to a location through the road network during the hours the business runs. Accessibility asks what happens after a reachable customer gets near the parcel: can they see the entrance, make the turn, park or queue, reach the door, complete the visit, and leave without friction?
Those questions are related, but they are not substitutes. A large drive-time catchment can create opportunity, but it does not prove the driveway works. A compact catchment can still outperform when the site sits on the right side of the commute, has clean turns, easy parking, and a low-friction visit.
This article uses that separation as the model. Draw reach with time-aware catchments, then score accessibility as site-level friction and confidence. The two core questions are:
- Who can realistically reach this location through the road network, during the hours the business actually runs?
- Once those people are in range, can they use the site without the trip falling apart?
The first is reach, and it is a time problem. The second is accessibility, and it is a friction problem. Keeping them apart is the entire job of a defensible score, because a strong catchment and a usable site are both necessary, and they are not the same evidence.
2. Why reach is a time problem
A three-mile ring is cheap, familiar, and easy to compare across markets. It also assumes customers travel equally in every direction, which they never do. People move along roads, around rivers, through signals, and into whatever traffic happens to be there when they leave. A radius knows none of that.
The sharpest version of the problem is time of day, because the reachable population around a fixed point can change completely between breakfast and midnight.
Consider downtown Washington, DC. Its resident population is roughly 584,000. On a weekday, Census commuter-adjusted estimates put its daytime population above one million as workers flow in. Same coordinates, same road network, a customer base nearly twice as large at lunch as at midnight. A lunch concept that sized its market on residential Census data would undercount its real audience by hundreds of thousands of people. A late-night concept making the opposite assumption would overcount by just as much.
Congestion bends the catchment the other way. INRIX's 2018 Global Traffic Scorecard estimated the average American driver lost 97 hours a year to congestion, measured across peak, off-peak, and free-flow conditions. Rush hour does more than slow a trip. It shrinks the ground a customer can cover in the ten minutes they have, so a breakfast catchment and a dinner catchment around one store are different shapes.
This is why the routing engine matters. Mapbox, which powers Geod's trade areas, builds isochrones as equal-travel-time contours that follow the actual road graph, and its driving-traffic profile with a depart_at parameter lets the catchment reflect conditions at a chosen hour. A 7:30 a.m. isochrone and a 6:00 p.m. isochrone for the same site describe two different markets, and the tool can show both.
Barriers make it concrete. In Poughkeepsie, New York, the Hudson River separates the city from Highland on the far bank. On a map the two look adjacent. On the road, a customer in Highland has to route to the Mid-Hudson Bridge to cross, which turns a short straight-line distance into a real detour. A radius counts those households as reachable. A drive-time catchment does not. Rivers, rail lines, limited-access highways, and one-way grids all open the same gap between nearby and reachable.
None of this is exotic, and short trips make it matter more. One widely cited consumer study found that 93% of consumers travel twenty minutes or less for everyday purchases. The shorter the acceptable trip, the more the exact shape of that trip decides who shows up, and the shape is set by time and network, not by distance.
All of that work belongs in Reach. It is the craft of drawing an honest catchment, and it says nothing yet about whether the site itself is any good to use.
3. Accessibility is the friction after reach
Once a customer is inside the reachable market, a second question starts, and a radius cannot touch it: can they actually finish the visit?
Accessibility in site selection is the practical usability of a site for the trip the concept depends on. It lives at the edge of the parcel and runs through the visit: turning in, finding the entrance, parking or queuing, walking from the car or the bus stop, completing the errand, and getting back out without deciding never to return.
| Factor | The question it answers |
|---|---|
| Ingress | Can customers enter safely, without hunting for the driveway? |
| Egress | Can they leave without a dangerous turn or a long wait? |
| Turn movements | Do medians, left-turn restrictions, or right-in/right-out limits suppress visits? |
| Side of road | Is the site on the convenient side for the direction of the trip? |
| Parking | Can customers find, enter, use, and leave parking at peak? |
| Queue or stack | Can drive-thru or service vehicles line up without blocking the lot? |
| Pedestrian path | Can nearby walkers and transit riders reach the door safely? |
| Visibility and wayfinding | Can a driver recognize the site in time to make the turn? |
| Delivery and loading | Can vendors and couriers operate without fighting customer traffic? |
| Confidence | Was each of these measured, inferred from imagery, field-verified, or unknown? |
A large ten-minute catchment does not prove the driveway works. High traffic does not prove drivers can turn in. Dense surrounding households do not prove the parking lot is big enough. A transit stop across the street does not prove there is a safe, obvious path to the door.
Traffic engineers have a formal discipline for part of this, called access management. The Federal Highway Administration defines it as the deliberate management of vehicular access points to parcels along a road, using driveway spacing, signal spacing, turn lanes, median treatment, and right-of-way design to keep both the parcel and the roadway working. That is accessibility described in the language of the people who build the roads.
4. Operators already separate reach from accessibility
The cleanest evidence that these are two different things comes from operators writing down what they want, with money on the line.
Scooter's Coffee, in its real estate criteria, asks for a ten-minute drive-time population of 10,000 or more. That is a reach requirement. Separately, it asks for sites on morning commuter routes, with strong visibility, full access from all directions, road speeds at or below 45 mph, full turning movements, a minimum ten-car drive-thru stack, employee parking, delivery and trash access, and room for a bypass lane around the stack. Every item after the population threshold is accessibility or feasibility. None of it is about how many people live nearby.
ALDI's published property requirements do the same thing. The company lists a dense population within three miles, which is reach. Then it lists a store of roughly 22,000 square feet on a 2.5-acre pad, about 95 dedicated parking spaces, 103 feet of frontage, a preference for a signalized full-access intersection, and traffic counts above 20,000 vehicles a day. The population line describes the market. Everything else describes whether the site can turn that market into completed trips.
Traffic counts sit in between, and they deserve a caution. A high AADT measures exposure, not usability. It can lift a fuel, coffee, or convenience site that lives off pass-by trips, and it can also raise turn friction and egress delay when the access geometry is poor. A busy road with a hard, unsignalized left turn out of the lot can be a worse site than a quieter road with a signal.
5. The clean model: Reach, Accessibility, Feasibility
Three questions get blended together in most committee conversations, and pulling them apart is what keeps a score honest.
| Layer | Core question | What belongs here |
|---|---|---|
| Reach | Who can realistically reach the site, and when? | Drive, walk, and transit catchments; time of day; daytime population; barriers; customer origins; distance decay |
| Accessibility | Can the people in range actually use the site? | Ingress, egress, turn friction, side of road, parking, stacking, pedestrian path, delivery access, wayfinding |
| Feasibility | Can this specific deal and parcel work at all? | Legal access, zoning, entitlement, utilities, site-plan fit, code, ADA compliance, cost, lease terms |
The rule that prevents double-counting is short. Population, households, and reachable demand live in Reach. Income and category spend live in Demand. Competitors live in Competition. Friction at and inside the parcel lives in Accessibility. Deal and code questions live in Feasibility.
A few items touch more than one layer, and the model should expose that rather than hide it. Parking is the clearest case. Whether customers can find and use a space at peak is accessibility. Whether the lot meets the required ratio for entitlement is feasibility. Those are different facts about the same asphalt, and a committee deserves both.
6. A brief history of measuring access
The instinct to treat distance as friction is almost a century old. The thread worth following is how "access" went from a single distance term to a network-and-time problem with a curb-level component.
1931, Reilly's Law of Retail Gravitation. William Reilly borrowed Newton's gravity law for retail, with trade drawn in proportion to population and inverse proportion to distance squared. Distance entered the model as friction for the first time.
1964, the Huff model. David Huff made trade areas probabilistic. A consumer's chance of choosing a store rose with the store's attractiveness and fell with travel time raised to a calibrated exponent. Friction stopped being a fixed assumption and became a parameter you tune to a category.
1960s to 1970s, transportation planning. Planners formalized accessibility as the ease of reaching opportunities through a real network under real impedance. Cumulative-opportunity measures counted what you could reach within a threshold; gravity-based measures weighted destinations by friction. Both treated time, not distance, as the cost that mattered.
Access management as a discipline. Traffic engineers, later codified by the Federal Highway Administration and AASHTO, built a body of practice around the curb itself: driveway spacing, median treatment, turn lanes, signal spacing, and the safety record of left turns. This is the part of accessibility that has nothing to do with the catchment and everything to do with the parcel.
2003 and 2009, floating catchment methods. Luo and Wang introduced the two-step floating catchment area method, and Luo and Qi added distance-decay weighting in the enhanced version. Access became a supply-to-demand ratio inside a travel-time catchment, which is the right tool for capacity-constrained formats like urgent care.
2010s, observed access. Smartphone location data gave the field a ground truth for where customers actually come from, and routing APIs made time-aware isochrones cheap. Modeled access could finally be checked against real behavior.
2020s, integrated platforms and shifting dayparts. Remote work reshaped daytime population and flattened the classic two-peak commute, which made static catchments built before 2020 less reliable. The current generation of tools combines time-aware isochrones, demographic aggregation, and competition into one workflow, without a GIS team.
7. How the access factors are actually derived
A score is only as honest as the data under each factor, so it helps to know where each one comes from and how solid it is.
Catchment shape and time of day come from a traffic-aware routing engine. Mapbox's isochrones follow the road graph, and the depart_at parameter pulls historical or current traffic for a chosen hour. This is the most automatable factor and the most defensible, as long as the brief states the hour it used.
Daytime and commuter population come from Census LEHD/LODES workplace data and ACS commuter-adjusted estimates. Together they show how the reachable base shifts between residents and workers, which is what separates a lunch site from a dinner site.
Side of road and directional fit come from directional traffic counts where a state DOT publishes them, or from LODES home-to-work flow direction as an approximation. This factor carries weight for commute-driven concepts and can be ignored for destinations.
Turn friction, medians, and one-way movement come from the road network. OpenStreetMap tags for one-way streets, turn restrictions, and road class give a usable first pass. The finer distinctions, such as a raised non-traversable median versus a two-way left-turn lane, often need imagery or a site plan, so this factor should carry lower confidence until it is checked.
AADT comes from state DOT count programs and the FHWA Highway Performance Monitoring System. It is public and useful, and also coarse and infrequently measured, which is worth disclosing rather than presenting as precise.
Walk and transit access come from the OpenStreetMap pedestrian network and agency GTFS feeds. Where sidewalks or schedules are poorly mapped, the result is unreliable, and the right response is to widen the confidence band rather than report a clean-looking number with nothing behind it.
Ingress, egress, parking friction, stacking, sight lines, and signage are the factors desktop data sees worst. Imagery gives a hint. A site plan gives more. A field visit gives the truth. These are the items most likely to need verification, and the model should say so on every brief.
8. Distance decay by category
Distance decay is the rate at which visits fall off as travel cost rises, and it differs sharply by category. The exponent that captures it, often written as beta, is calibrated rather than assumed.
Esri's ArcGIS Pro documentation notes the exponent usually runs between 1.5 and 2, with grocery taking a high exponent because people travel only a short way for it, and furniture taking a low one because people will travel farther for a planned purchase. A supermarket calibration in Esri's own Huff white paper put the grocery exponent near 2.66 on road distance. A later study using mobile-visit data across the ten largest US cities found grocery decay consistently steeper than department-store decay in every city, with mean home-to-store distances around 7.8 km for grocery and 10.3 km for department stores.
The exact figures differ because the studies use different distance units and calibration methods, so a decay exponent is a method-specific result, not a universal constant. The pattern under them is stable. Convenience categories such as coffee, grocery fill-in, and fuel have tight catchments and steep decay. Big-ticket and destination categories such as furniture and specialty retail have wide catchments and gentle decay.
Restaurants are the instructive exception. A validation study against real card-transaction data found gravity models fit grocery, fuel, and clothing well, and fit restaurants poorly, because dining is more destination-driven than distance-driven. For restaurant concepts, daypart traffic and access friction carry more weight than a clean distance-decay curve, which is a useful caution before leaning on a single drive-time threshold for a dining brand.
9. How to score accessibility
Accessibility works best as a concept-specific friction index, not a second catchment score. A practical version uses five sub-scores.
Approach and entry. How easily a user gets from the road into the site: signalized access, curb cuts, driveway placement, median type, turn restrictions, corner versus mid-block position, road speed, and side-of-road fit. Side of road carries real weight for commute-driven coffee, breakfast, and fuel, where the site competes for a stop inside a trip already underway. It barely registers for a weekend destination.
Exit and return. Most models overweight entry and forget the trip out, which is the part customers actually remember. Score left-turn delay, the route back to a major corridor, exit sight lines, and conflicts with internal circulation.
On-site circulation. What happens after arrival, which varies sharply by format. Grocery means parking supply, cart path, and weekend flow. Drive-thru means stack depth, bypass lane, and conflict with parked cars. Auto service means bay access, vehicle storage, and tow access. Healthcare means drop-off, parking, and a clear path to the entrance.
Multimodal access. Whether the site works beyond the car, drawing on the GTFS, LODES, and pedestrian-network sources above. This sub-score earns its weight for urban clinics, grocery, banks, campuses, and any concept serving workers or transit riders, and it can sit near zero for a suburban drive-thru.
Confidence. Every sub-score should disclose how it was derived: field-verified, modeled from credible data, inferred from imagery, or unavailable. A low-confidence 82 should never outrank a verified 74, and the brief should make that impossible to miss.
10. What to weight, and what to gate
Some access factors move a score by degree. Others can end a deal on their own, and burying one of those inside a weighted average is how a fatal flaw gets approved.
| Factor | Weight it | Gate it |
|---|---|---|
| Right-turn convenience | Yes | Sometimes, for commute-driven coffee or fuel |
| Signalized access | Yes | Sometimes |
| Left-turn friction | Yes | Sometimes |
| Parking convenience | Yes | Yes, if the concept cannot run without it |
| Drive-thru stack | Yes | Yes, if drive-thru is central to the format |
| Pedestrian and transit path | Yes | Sometimes, for healthcare and urban concepts |
| Delivery and loading | Yes | Yes, for delivery- or service-heavy operators |
| Legal access | No | Yes |
| Zoning and entitlement | No | Feasibility gate |
| ADA-accessible route and parking | No | Compliance and feasibility gate |
| Site-plan fit | No | Feasibility gate |
ADA deserves its own line. General spatial accessibility and disability-access compliance are related questions, and they are different questions. ADA.gov's guidance on accessible parking and routes - accessible spaces on the shortest route to an accessible entrance, routes free of curbs and stairs, minimum widths, stable surfaces - is a compliance matter, and it belongs in Feasibility. A brief can flag customer-access friction under Accessibility while keeping the compliance review where it belongs, and it should never let a good drive-time number stand in for an accessible entrance.
11. Accessibility is category-specific
The same parcel can be excellent for one concept and weak for another, because the trip is different.
| Concept | The access test | Where the friction concentrates |
|---|---|---|
| Coffee drive-thru | Can the morning commuter stop without leaving the route? | Going-to-work side, right-turn ingress, stack, bypass lane, signage |
| QSR drive-thru | Can vehicles enter, queue, pay, pick up, and exit without blocking the lot? | Stack depth, curb cuts, signal, pickup lane, egress |
| Fast casual | Can lunch and dinner guests complete a short trip quickly? | Parking turnover, walk path, pickup access, daytime approach |
| Grocery | Can a household run a stock-up trip without parking or cart-path pain? | Signalized access, parking supply, weekend flow, loading separation |
| Urgent care | Can a patient arrive, park, find the door, and enter under stress? | Drop-off, parking, ADA route, transit path, evening access |
| Fitness | Can a member repeat the trip several times a week at peak hours? | Evening parking, home and work approach, entrance clarity |
| Auto service | Can a vehicle enter, queue, reach a bay, and exit safely? | Bay circulation, storage, tow access, curb cuts |
| Convenience and fuel | Can a driver use the site as part of a trip already happening? | Corner access, canopy circulation, side of road, fast egress |
| Specialty retail | Can a destination customer find the site and finish a planned trip? | Regional approach, wayfinding, parking, co-tenancy circulation |
A universal accessibility score disappoints operators for exactly this reason. The score has to know the trip before it can judge the site.
12. A worked accessibility scorecard
Here is how the model reads on a single site. The example is a composite, built to show the format rather than to report a real location, and the numbers are illustrative.
A regional coffee chain is evaluating a suburban pad on a six-lane arterial. The trade area looks healthy: a weekday 7:30 a.m. ten-minute catchment holds well above the chain's 10,000-person threshold, and the daytime worker population is strong. That is reach, and it passes. The accessibility read is where the site gets interesting.
| Sub-score | Illustrative score | Confidence | Note |
|---|---|---|---|
| Approach and entry | 58 | Medium | Site sits on the going-home side, not the going-to-work side, for the morning commute |
| Exit and return | 44 | Low | Raised median forces a right-out and a downstream U-turn to head back toward downtown |
| On-site circulation | 71 | Medium | Stack depth fits, bypass lane is tight against three parking rows |
| Multimodal access | 30 | Medium | Car-dominant location, low transit relevance for this concept |
| Confidence | n/a | Mixed | Egress and median verified from imagery only, flagged for field check |
The composite accessibility score lands in the low 50s, well below what the reach number alone would suggest. The decisive issue is not in the table as a weight. The morning-commute side-of-road problem combined with the median-forced exit is a gate for a coffee drive-thru, because the concept lives on the morning trip and this site fights it twice. A model that averaged everything would have buried that under the strong stack and strong reach. A model that gates it surfaces the problem before the lease, not after the first quarter of soft morning sales.
The same parcel might score well for a dinner-oriented fast-casual brand, where the going-home side is the right side and the median matters less. The site did not change. The trip did.
13. Where gravity, Huff, and floating-catchment methods fit
The academic toolkit matters, and it belongs in a specific layer. Gravity and Huff models estimate the probability that a customer chooses a store from its attractiveness and the travel friction to reach it, relative to the alternatives. Liang and colleagues extended this into a time-aware dynamic Huff model calibrated on mobile-location data, showing how predicted share shifts by hour and store type. Suhara and colleagues validated gravity-based market-share models against real transaction data.
Those methods make Reach, Demand, and Competition more realistic. They estimate who is reachable, how far people will travel for a category, and how a nearby competitor erodes capture. Healthcare planning adds the two-step floating catchment family, which measures access as the ratio of provider supply to population within a travel-time catchment, useful for urgent care, primary care, and other capacity-constrained formats.
The discipline is to keep all of that on the Reach and Competition side of the model. The moment a gravity calculation becomes the accessibility score, accessibility has turned back into catchment math, and the double-count returns through the back door.
14. Data you can build versus data you have to buy
A useful accessibility model mixes open data, licensed data, and field verification, and a buyer should know which factors come cheap and which carry a cost.
Open and public data covers most of the catchment and the network. Traffic-aware isochrones run on low-cost routing APIs. Resident demographics come from ACS. Daytime population and home-to-work flow come from Census LODES and LEHD. Coarse AADT comes from state DOTs and the FHWA HPMS. Road attributes for one-way streets, turn restrictions, and road class come from OpenStreetMap. Walk and transit proxies come from OpenStreetMap plus agency GTFS feeds.
Licensed or hard-to-get data sharpens the curb-level and behavioral factors. Dayparted and directional vehicle counts in fine time bins come from vendors such as INRIX and StreetLight. Observed trade areas and foot-traffic ground truth come from mobile-data vendors such as Placer.ai and Unacast. Polished walk, transit, and parcel scores come from Walk Score and Local Logic. The finest access geometry, such as raised median versus two-way left-turn lane, driveway permits, and sight lines, often needs a manual review or a DOT permit record. A decay curve calibrated to a specific brand needs that brand's own customer or sales data.
The practical consequence is that a desktop screen can get the reach and network factors close to right, and the highest-value curb-level factors stay partly manual. A model that hides this behind a single number is selling false precision. A model that shows it builds trust.
15. What to automate versus field-verify
Not every factor deserves the same confidence, and the workflow should automate the screen, flag the uncertainty, and send a human to the variables that move the decision.
| Factor | Automated screening | Field verification |
|---|---|---|
| Drive-time isochrones | Yes | Usually not |
| Walk and transit access | Yes, where GTFS and sidewalks are mapped | Sometimes |
| Daytime and commuter population | Yes | Not needed |
| Traffic counts | Yes, where sourced | Sometimes |
| Barriers | Yes | Sometimes |
| Side of road | Partly | Often |
| Turn restrictions and median type | Partly | Often |
| Ingress and egress | Partly | Yes |
| Parking supply | Partly | Yes |
| Parking friction | Limited | Yes |
| Drive-thru stacking | Partly | Yes |
| Visibility and signage | Limited | Yes |
| ADA-accessible path | Limited | Yes |
| Delivery and loading | Partly | Yes |
16. How accessibility should appear in a site brief
A brief should show the three layers separately, so a reader can see which question each fact answers. Reach carries the travel mode, time threshold, daypart, routing source, traffic assumption, reachable population, barriers, and any customer-origin validation. Accessibility carries the ingress and egress read, turn friction, side-of-road fit, parking supply and friction, circulation, stack and loading constraints, pedestrian and transit path, wayfinding, delivery access, and a confidence level for each. Feasibility carries legal access, zoning, entitlement, utilities, code, ADA compliance, prototype fit, cost, and lease terms.
Geod's current habit of marking accessibility "unavailable" rather than inventing a number is the right instinct, and it should be stated more plainly, not softened:
Accessibility not yet scored. Geod does not have verified inputs for ingress, egress, turn movements, side of road, parking, frontage, signage, pedestrian or transit path, or delivery access at this site. The overall score is partial. Verify these in the field before approval.
That disclosure builds trust because it tells the committee exactly what the system measured, what it inferred, and what no one has checked yet. A missing input left honestly missing is worth more than a confident number with nothing behind it.
17. Best practices
- Score reach and accessibility on separate lines, and never let catchment size leak into the accessibility number.
- Pick the daypart that matches the concept's peak before you draw a single isochrone. A breakfast brand cares about the 7 to 9 a.m. catchment; a dinner brand cares about 6 to 8 p.m.
- Treat fatal access conditions as gates, not weights, so a strong reach number cannot rescue a site the concept cannot operate on.
- Attach a confidence level to every sub-score, and present a low-confidence number so it can never outrank a verified one.
- Field-verify the factors desktop data sees worst: sight lines, parking friction, stack behavior, and true turn difficulty.
- Calibrate decay and weights against your own store performance once you have it. A brand's real customer origins beat a generic category exponent.
- Re-check a site after the road around it changes. A new signal, a new median, or an interchange reconfiguration can move an accessibility score more than a year of demographic drift.
18. The future of accessibility scoring
Accessibility scoring should move from broad catchment proxies toward evidence about the actual visit. The next version of a mature site score separates five components, each answering a distinct question.
| Component | What it measures |
|---|---|
| Reach | The time-aware, network-aware catchment |
| Demand | Target customers and category potential inside it |
| Competition | Alternative supply competing for that demand |
| Accessibility | The friction between the network and a completed visit |
| Feasibility | Whether the site, deal, and physical plan can work |
That structure gives an expansion team a decision record it can defend, and it gives a product team a roadmap. Reach gets smarter with time-aware isochrones, observed customer origins, distance decay, and commuter flows. Accessibility gets sharper with ingress, egress, turns, parking, side of road, circulation, multimodal access, and honest confidence. Feasibility remains the gate for legal, physical, code, and deal constraints.
The direction is not a single, more confident number. It is a score that knows what it has measured, what it has inferred, and what still needs a site plan or field visit. A location-strategy system earns trust by showing where it is strong, where it is uncertain, and where someone still has to walk the site. The aim is a score a CFO can interrogate line by line, and a brief that survives the meeting it was built for.
19. FAQ
What is accessibility in site selection? Accessibility is the practical usability of a site for the target trip. It measures whether reachable customers, employees, delivery drivers, or patients can enter, use, and exit the site with acceptable friction. It is separate from how many people live nearby.
How is accessibility different from reach? Reach measures who can realistically get to the location through the road network, including time of day and barriers. Accessibility measures whether those reachable people can complete the visit. Drive-time catchments and reachable population belong in Reach. Ingress, egress, parking, turn friction, and on-site circulation belong in Accessibility.
Is drive time an accessibility metric? Drive time is usually a Reach metric, because it defines the realistic catchment. Accessibility starts after the catchment is drawn: can users in that catchment enter, park, queue, walk, load, and exit without the trip breaking down?
What is ingress and egress in site selection? Ingress is entering the site; egress is leaving it. They matter because a site with strong demographics and high traffic can still underperform when customers face difficult turns, unsafe exits, poor driveway placement, or confusing circulation.
How do traffic counts relate to accessibility? Traffic counts measure exposure and road context, not usability. A high AADT can help a pass-by concept like fuel or coffee, and it can also increase turn friction and egress delay when the access geometry is poor.
What accessibility factors matter most for restaurants? It depends on format. Drive-thru concepts need stack depth, full turning movements, pickup flow, and clean egress. Fast casual needs parking turnover, walkability in urban markets, and pickup access. Casual dining weights evening and weekend access, parking, and wayfinding more heavily. Pure distance-decay models fit dining poorly, so daypart and access matter more here than a single drive-time threshold.
What accessibility factors matter most for healthcare? Patient travel time, parking, drop-off, transit and pedestrian access, an ADA-accessible route, and a clear entrance under stress. Market planning for capacity-constrained care can also use floating catchment methods to compare provider supply with population demand.
What should be automated versus field-verified? Reach can largely be modeled with routing engines, isochrones, customer origins, demographics, and network data. Accessibility is partly automatable, but ingress, egress, sight lines, signage, parking friction, drive-thru stacking, and pedestrian safety usually need imagery review, a site plan, or a field visit.
How should missing accessibility data be handled? Mark it unavailable or low confidence. Never impute it into a precise-looking score. A trustworthy brief shows what was measured, what was inferred, and what still needs verification.
Related reading
- Trade area analysis for retail site selection
- Trade area overlap and cannibalization
- White space analysis
- Market saturation analysis
- Defensible site selection scores
- Site selection scoring models
References
- Geod. "Methodology."
- Geod. "Trade Area Analysis for Retail Site Selection."
- Scooter's Coffee. "Real Estate."
- ALDI US. "Property Requirements."
- Federal Highway Administration. "Access Management."
- Mapbox. "Isochrone API."
- INRIX. "2018 Global Traffic Scorecard."
- U.S. Census Bureau. "Commuter-Adjusted Daytime Population."
- Access Development. "The Impact of Retail Proximity on Consumer Purchases" (2016).
- U.S. Census Bureau. "LEHD Origin-Destination Employment Statistics (LODES)."
- General Transit Feed Specification. "Reference."
- ADA.gov. "Restriping Parking Spaces."
- Huff, D.L. (1964). "Defining and Estimating a Trading Area." Journal of Marketing, 28(3), 34-38.
- Huff, D., and McCallum, B.D. (2008). "Calibrating the Huff Model Using ArcGIS Business Analyst." Esri White Paper.
- Liang, S., et al. (2020). "Calibrating the dynamic Huff model for business analysis using location big data." Transactions in GIS, 24(3).
- Suhara, Y., et al. (2021). "Validating Gravity-Based Market Share Models Using Large-Scale Transactional Data." Big Data, 9(3).
- Luo, W., and Wang, F. (2003). "Measures of spatial accessibility to health care in a GIS environment." Environment and Planning B, 30(6).
- Luo, W., and Qi, Y. (2009). "An enhanced two-step floating catchment area (E2SFCA) method." Health & Place, 15(4).