Site selection analysis is the process of deciding whether a specific location fits an intended use, using data on who is nearby, how they behave, what competes for them, and how the site trades — instead of relying on a broker’s instinct and a windshield tour. For a small commercial real estate firm, it is the analytical spine underneath a listing pitch, a client’s expansion, or an acquisition thesis. The work has not changed in purpose. What has changed is how fast a lean team can produce a defensible answer, because the data that used to take an analyst three days to assemble now feeds directly into an AI-assisted workflow that a principal can run before a meeting.
What Site Selection Analysis Actually Is
At its core, site selection analysis answers one question: will this use perform at this location, given what surrounds it? A retailer asks whether a storefront will draw enough of the right customers. A developer asks whether a parcel supports the density and rents the pro forma assumes. An acquisitions team asks whether the tenant in place, or the next one, can be sustained by the local market. Same discipline, different vantage points.
The traditional version relied on experience and a few reference numbers: a traffic count, a rough population figure, a sense of the competition from driving the corridor. That still has value, but it is easy to argue with, because it rests on judgment a client cannot inspect. Data-driven site selection replaces the hand-wave with evidence a skeptical partner or lender can follow.
The center of the method is the trade area — the geography where most of a location’s customers, tenants, or visitors actually come from. For decades this was drawn as a ring: a one-, three-, and five-mile radius around the pin. Mobility data has since shown that people do not live in neat circles around the places they visit, so the current approach measures the real catchment empirically, from where visitors are observed to originate. Once you know the true trade area, every other data layer — income, spending, competition — is measured against the population that matters rather than an arbitrary circle.
The Data Behind a Location Decision
The phrase “the data behind location decisions” is doing real work. A defensible site study pulls from several distinct layers, and knowing where each comes from is half the battle for a firm without an analyst.
| Data layer | What it tells you | Typical source |
|---|---|---|
| Demographics | Population, income, age, education, employment in the trade area | Census-derived platforms; Esri ArcGIS Business Analyst; RPR (for NAR members) |
| Consumer segmentation | The lifestyle and spending profile of nearby households | Esri ArcGIS Tapestry (60 US segments), similar geodemographic systems |
| Foot traffic / mobility | How many people visit a spot, how often, and where they come from | Placer.ai, SafeGraph, Unacast and other mobility panels |
| Competition and co-tenancy | What already competes for the same customer, and what draws traffic nearby | Listing and market platforms (CoStar, Crexi), mobility panels, field survey |
| Access and visibility | Traffic counts, ingress/egress, parking, corner quality | DOT traffic-count data, site visit, aerials |
| Zoning and land use | What the site legally permits and what entitlements it needs | Municipal records, planning documents, listing package |
| Comparable rents and sales | What space in the submarket actually commands | CoStar, Crexi, broker comps, your own deal history |
No single platform delivers all seven cleanly, which is why the honest answer to “what tool do I need” is usually “it depends on your asset mix.” A firm working consumer-facing retail leans hard on the first three rows. A firm underwriting industrial leans on the last three. That distinction matters more than any brand name, and it is the reason a tool that is essential for one firm is a waste of a subscription for another. Choosing among those platforms is its own decision, which we cover in a companion look at the AI site-selection tools worth buying for a small CRE firm.
Why Asset Class Decides the Method
The single most useful thing to understand about site selection analysis is that the method is not universal. It bends heavily toward consumer-facing property, and that bias determines which data actually predicts performance.
Retail, restaurant, and consumer services. This is where trade-area and foot-traffic analysis earns its keep. Value tracks people: how many pass the door, who they are, what they spend, and what competes for them. Demographics, mobility panels, consumer segmentation, and co-tenancy do most of the predictive work. A well-built retail site study can genuinely change a leasing recommendation, and it is the use case every location-intelligence vendor optimizes for.
Office. Location value runs on labor access, transit, amenity density, and the specific building’s specifications more than on residential demographics. Foot-traffic panels tell you little about whether a tenant will sign a ten-year lease. The relevant “data behind the decision” is comparable rents, tenant credit, commute sheds, and submarket vacancy.
Industrial and logistics. Here the drivers are highway and port access, drive-time to population centers, labor availability, power, and clear-height and dock specifications. A consumer-profile report is nearly irrelevant. The analysis is closer to a logistics problem than a demographic one.
Land and self-storage. Land decisions hinge on entitlement risk, zoning, and future absorption; self-storage is driven by household density, unit supply within a short drive, and turnover. Each has its own data spine.
The practical takeaway: before you buy any tool or run any study, name the asset class. If your book is not consumer-facing, most of what the site-selection industry sells does not apply to you, and your analytical dollars belong in deal screening and underwriting workflows instead.
How the Analysis Is Done, Step by Step
A competent site study, stripped to its bones, follows the same sequence regardless of asset class. Only the data layers change.
- Define the use and the question. A drive-thru quick-service restaurant, a 120,000-square-foot distribution box, a value-add office repositioning — each sets different success criteria. Write down what “good” looks like before you pull a single number.
- Establish the trade area. For consumer uses, derive it from mobility data or a defensible drive-time. For office and industrial, define the relevant labor shed, commute band, or logistics radius.
- Pull the matching data layers. Only the ones the asset class needs. Resist the urge to bury the study in demographics that do not drive the decision.
- Assess competition and supply. Map what already competes for the same customer or tenant, and where new supply is coming.
- Reconcile against economics. Overlay comparable rents and sales so the location read connects to a number the deal can carry.
- Write the recommendation. Turn the layers into a clear position a client, partner, or lender can act on, with the reasoning visible.
Step six is where most studies quietly fall apart. A firm gathers strong data and then hands the client a stack of screenshots instead of a recommendation. The analysis is only as good as the narrative that closes it, and that closing step is exactly where AI has become useful.
What AI Changes, and What It Does Not
AI has not replaced site selection analysis. It has compressed the parts that used to require an analyst and left the parts that require judgment where they belong.
What AI genuinely changes:
- The last mile. The hardest, most time-consuming step for a lean firm is turning raw layers into a defensible written recommendation. A general AI assistant — ChatGPT, Claude, Gemini, or Microsoft Copilot — will take the demographic, comp, and mobility data you already have and draft a structured, client-ready location narrative in minutes, with the caveats and reasoning spelled out. That is where most of the recoverable time sits.
- Reading unstructured inputs. Site work is buried in PDFs: broker packages, zoning ordinances, traffic studies, lease abstracts. Current models read those documents and pull the facts you need into a usable form, which used to be manual transcription.
- Speed and consistency. A firm can run a first-pass read on ten candidate sites in the time it once took to properly study one, then reserve the deep work for the survivors. That volume shift is the real story of AI in how market research across CRE is being done.
What AI does not change:
- The data source. A general AI assistant cannot conjure proprietary foot-traffic counts or licensed rent comps. It works on the data you feed it. The panel or platform subscription is a separate decision, and if the underlying data is thin, no model fixes that.
- Ground truth and judgment. Whether a corner “feels right,” how a landlord will actually behave, whether an entitlement will clear the local council — these still come from experience and a site visit. Treat model output as a strong first draft, not a verdict.
- Accountability. When you put a recommendation in front of a client, you own it. That is a reason to keep a human reviewing every number, and it separates genuine analysis from the kind of statistical guesswork explored in what predictive analytics in real estate actually delivers.
The firms getting the most out of this are not the ones with the biggest data budgets. They are the ones whose people know how to prompt a model against the data they already license — which is a learnable skill, not a software purchase.
Where Site Selection Fits in the Deal Workflow
Site selection analysis rarely stands alone. For a brokerage it feeds a pitch; for an acquisitions shop it is one input into screening and underwriting a deal. The location read tells you whether demand is there; the underwriting tells you whether the numbers work at that demand. Together, they are how a lean team decides which opportunities deserve attention and which to pass on quickly. The full sequence — from first-look screening through underwriting a location-backed thesis — is laid out in our playbook on screening and underwriting more deals with a lean team.
That efficiency is the whole point for a firm of four to twenty people. You are not trying to out-spend a national brokerage on data. You are trying to reach a confident location decision faster and defend it as well as a shop three times your size. That is the operating advantage a small firm can actually build, and it runs through the broader case for how small CRE firms out-operate institutional giants.
Frequently Asked Questions
What is site selection analysis in commercial real estate?
Site selection analysis is the structured evaluation of whether a specific location fits an intended use, based on data rather than instinct. It combines the trade area (where customers or tenants come from) with demographics, consumer behavior, competition, access, zoning, and comparable rents to predict whether a use will perform. For a broker it supports a listing or a client’s expansion; for an acquisitions team it is one input into underwriting a deal. The goal is a recommendation a client, partner, or lender can follow and trust.
What data is used in site selection?
The core layers are demographics (population, income, age, education, employment), consumer segmentation, foot-traffic or mobility data, competition and co-tenancy, traffic counts and access, zoning and land use, and comparable rents and sales. Which layers matter depends on the asset class. Consumer-facing retail leans on demographics and foot traffic; industrial leans on access, labor, and building specifications. No single platform provides all of it cleanly, which is why the data sources you need follow from your asset mix.
What is a trade area, and why does it matter?
A trade area is the geographic zone where most of a location’s customers, tenants, or visitors actually originate. It matters because every other data layer — income, spending, competition — should be measured against the population that truly matters, not an arbitrary circle on a map. Trade areas were once drawn as one-, three-, and five-mile rings, but mobility data shows real catchments are irregular. Measuring the true trade area first makes the rest of the analysis far more accurate.
Does site selection analysis work for office and industrial, or only retail?
It applies to all asset classes, but the method and data change sharply. Foot-traffic and consumer-profile analysis is built for retail, restaurant, and consumer services, where value tracks people. Office value runs on labor access, transit, amenities, and building specifications; industrial runs on highway and port access, drive times, labor, and clearance. If your book is not consumer-facing, most location-intelligence products add little, and your analysis should focus on comps, access, and underwriting instead.
How is AI changing site selection analysis?
AI compresses the slow, analyst-heavy parts of the work. Its biggest contribution is the last mile: turning demographic, comp, and mobility data you already have into a clear, client-ready location recommendation in minutes. It also reads unstructured inputs like zoning documents and broker PDFs, and lets a firm run first-pass reads on many candidate sites quickly. What it does not do is generate proprietary data, replace a site visit, or remove your accountability for the final recommendation.
Can I do site selection analysis with ChatGPT or Claude?
Partly, and it is often the smartest first step. A general AI assistant cannot pull licensed foot-traffic or comp data on its own, but it excels at the last mile — synthesizing the data you already license into a structured recommendation, drafting the narrative, and surfacing caveats. For firms running occasional studies, that analysis step is where most of the recoverable value sits, at a fraction of a data platform’s cost. The skill that unlocks it is knowing how to prompt the model against your own data.
Do I need an expensive platform to run a site study?
Not necessarily. If your work is consumer-facing and frequent, a demographic platform or foot-traffic panel can pay for itself. If you run only a handful of studies a year, an annual subscription often costs more than the value it returns, and a general AI assistant paired with data you already license is the more sensible spend. The honest test is to divide any platform’s annual price by the number of studies you realistically run, then compare that per-study cost to the alternative.
How much do site selection tools cost?
Costs range widely by category. A general AI assistant runs roughly $20 to $30 per user per month. Demographic and GIS platforms are typically annual subscriptions in the mid-hundreds to low-thousands per user. Foot-traffic panels are quoted annually and, by third-party estimates, commonly fall in the low five figures to tens of thousands of dollars per year, with enterprise tiers higher. Match the spend to your asset mix and study volume rather than to feature lists.
How does site selection analysis fit into deal underwriting?
Site selection tells you whether demand exists at a location; underwriting tells you whether the deal’s numbers work given that demand. In a lean firm they run together: a quick location read helps you screen which opportunities deserve full underwriting and which to pass on fast. Treating the two as one workflow is how a small team reviews far more deals without adding headcount, focusing its deep analysis only on the sites that survive the first look.
Where to Start
The first question is not “which data platform,” it is “what does my firm’s pipeline actually need, and where are we losing hours we could get back.” A free AI-readiness assessment gives you that read: a short working session that looks at your asset mix, how many location studies you realistically run, and where your research time goes, then returns an honest recommendation on whether a data subscription, an AI-assisted workflow on data you already license, or focused training on prompting is your best next move. If your team wants that fluency directly, our workshops teach exactly how to prompt current AI assistants against your own market data, comps, and documents. Book a free AI-readiness assessment before you commit to a platform you may only use a few times a year.
Arthur Wandzel
