Site selection analytics is the practice of scoring and comparing candidate locations with data instead of opinion. It combines demographics, foot traffic, competitor positions, and your own store performance into a forecast for each site, so a real estate team can rank a long list, defend the shortlist to a committee, and show which inputs drove the answer.
Why It Matters More Than It Used To
The market gives you less room to be wrong than it used to. US retail vacancy sat at 6.0% in Q2 2026, three basis points above the prior quarter and still well under the 7.4% historical average, per Cushman & Wakefield's Q2 2026 US Retail MarketBeat. Tight availability means fewer second chances on a corner you passed on, and a longer wait before the next one comes up.
I'm Clyde Christian Anderson, Founder and CEO of GrowthFactor.ai. Between my family's retail business and a stint in investment banking, I've watched both sides of this decision get made badly, and I've seen what the right location intelligence platform changes.
What Is Site Selection Analytics?
Three jobs sit under the term, and vendors rarely separate them. Screening ranks a long list of markets or addresses so you know where to spend your travel budget. Forecasting estimates what a specific site would do in its first full year, with a range rather than a single number. Portfolio work asks what a new store does to the ones you already have, which is where cannibalization and market saturation get answered.
A team that only needs screening should not pay for a forecasting engine, and a team signing 20 leases a year needs all three. Most disappointing software evaluations come from buying one job and expecting the other two.
Evolution of Site Selection Analytics
Site selection used to mean driving the trade area, counting cars at the light, and drawing a census ring around the pin. The ring was the weak part. People do not shop in circles, and census counts describe where somebody sleeps rather than where they buy lunch.
Foot traffic data fixed that, showing where customers actually travel rather than where a circle says they sleep. Machine learning found patterns across thousands of stores that no analyst has time to check by hand, and cloud GIS dropped the price of entry far enough that a 20-store brand can run analysis that once needed a research department. See our guide to data-driven site selection.
Site Selection Analytics in Different Industries
The method holds across industries; what changes is which input carries the weight.
- Retail and restaurants: Foot traffic, visibility, and access, with drive-thru and parking often deciding a quick-service site outright.
- Healthcare: Drive-time coverage of underserved populations, plus the referral network already in the area.
- Real estate investment and CRE: Long-run market trends across a portfolio rather than the merits of one address.
- Logistics and EV charging: Corridors, labor supply, and utility capacity, where the customer is a route rather than a neighborhood.
Key Factors & Data Inputs Every Analyst Needs
Successful site selection analytics depends on having the right data. The goal is to match your customer profile to the neighborhood's DNA using a variety of inputs.
- Demographics: The foundation, including population density, age, income, and education levels.
- Foot traffic: Quantifies the flow of people, including daily volume, peak hours, and dwell times.
- Consumer spending: Reveals local economic activity, including average purchase amounts and category spending.
- Points of Interest (POI): Maps the surrounding commercial ecosystem, from competitors to complementary businesses.
- Mobility patterns: Shows where visitors come from and the routes they take, revealing a location's true reach.
- Psychographics: Dives into the lifestyle and values of potential customers to ensure a cultural fit.
- Competition density: Analyzes market saturation and identifies untapped opportunities.
- Labor supply: Wage levels, unemployment, commuting distance, and how many other employers are hiring the same shift workers you need.
- Zoning and permitting: Use classification, signage rules, parking minimums, drive-thru restrictions, and how long the local authority actually takes.
Those last two get skipped in most guides and they end more deals than demographics do. A site can score well on every customer measure and still be unstaffable, and a trade area full of your ideal buyer is worth nothing if the municipality will not permit a drive-thru on that parcel. Check both before the site visit, not after the letter of intent. For foundational guidance on setting criteria, see this resource on how to determine criteria for site selection.
Trade Area & Catchment Best Practices
The trade area you draw decides every number that follows it. Get the shape wrong and the population count, the competitor list, and the sales forecast are all wrong together, in the same direction, with no obvious sign that anything broke.
Many businesses still define that shape with radius rings, a 1- or 3-mile circle around the site. Circles are fast and they are wrong in a specific way: they count people on the far side of a river, a rail line, or a divided highway who will never make the trip. Drive-time isochrones fix the geometry by routing the road network, but they still describe who could reach you rather than who does. The observed trade area, built from where real visits and loyalty records originate, is usually lopsided, and the lopsidedness is the finding. Half of the ring you would have counted was never your market.
Trade areas also move. A new interchange, a closed anchor, or a shift in commuter patterns redraws the shape, so a study from three years ago is a historical document rather than a current input. Our guide on Retail Store Site Selection explores these concepts in more detail.
Avoiding Cannibalization
Cannibalization is what happens when a new store takes its sales from your own existing locations rather than from the market. The chain nets far less than the opening forecast promised, and the damage shows up on a different store's P&L than the one being celebrated.
Overlap analysis is the first step. By mapping the trade areas of existing and potential stores, you can identify high-risk overlaps. Sales transfer studies go deeper, predicting how much revenue might shift between locations. This helps you understand the net impact on your entire portfolio.
Some cannibalization is worth accepting, usually to block a competitor from a corner or to cut wait times at an overloaded store. The point is to price that trade rather than discover it in the quarterly numbers. See our sales forecasting tips for retail site selection.
Traditional vs. Analytics-Driven Site Selection
| Feature | Traditional Site Selection Methods | Analytics-Driven Site Selection Methods |
|---|---|---|
| Methods | Gut instinct, anecdotal evidence, simple radius rings, manual traffic counts, "windshield surveys" | AI, Machine Learning (ML), GIS, predictive modeling, spatial interaction models, dynamic trade area analysis, scenario testing |
| Data Breadth | Limited, often outdated census data, basic demographic reports | Vast, real-time, integrated datasets (demographics, foot traffic, consumer spending, POI, mobility, social sentiment, internal sales data) |
| Accuracy | Subjective, prone to human bias, no way to check the forecast afterwards | Back-tested against your own store portfolio, so the error rate is a number you can quote and argue with |
| Speed | Slow, weeks to months to evaluate a few locations | Fast, minutes to hours to analyze hundreds or thousands of potential sites, enabling rapid expansion |
Modern Techniques & Tools for Powerful Site Selection Analytics
Modern site selection analytics runs on a handful of components that most platforms assemble in some combination. Knowing which piece does which job makes vendor demos much easier to sit through.
- Model-based forecasting: Machine learning systems trained on your existing stores project sales for a site that does not exist yet, and report the range they are confident in.
- GIS mapping: The layer that puts demographics, foot traffic, and competitor positions on the same map so spatial relationships are visible rather than inferred.
- Spatial data: The underlying geographic datasets, which is mostly what you are paying a vendor for.
- APIs: The connection between the platform and your own systems, which decides whether the analysis reaches your pipeline or stops at a PDF.
A GIS pass often surfaces sites nobody proposed. See how a GIS analysis identifies a site's suitability.
Integrating Internal & External Data Sources
External market data tells you what a place is like. Your own POS, CRM, and loyalty records tell you which places have worked for you, and that second half is the one teams most often withhold during a pilot. Without it, a model can say a trade area looks good in general. With it, the model learns what your good store looks like, which is a different and far more useful question. See our guide to AI in real estate for how those models get built.
Predictive Modeling & Scenario Testing
Predictive modeling acts as a crystal ball for site selection analytics, allowing you to forecast potential outcomes before investing. These models use historical data and algorithms to project sales, traffic, and cannibalization risk.
Techniques range from gravity models, which estimate customer attraction from store size and distance, to machine learning models like regression trees and neural networks that find relationships nobody thought to look for. Interactive what-if dashboards let you test scenarios, such as a competitor opening across the street or a planned housing development filling up.
The part that decides whether anyone trusts the output is whether the score opens. A number on its own starts an argument in committee. A number with its inputs attached, showing that demographics scored badly and market potential scored well and why, is something a team can actually discuss. GrowthFactor breaks every site score into five lenses with a written justification for each, so the disagreement happens over the input rather than over the number.
Automating the qualification steps is what makes volume possible: GrowthFactor customers evaluate roughly five times more potential sites than they did by hand.
Visualization & Stakeholder Communication
The analysis has to survive the committee meeting, and that is where most good site work dies. What travels is a map with the reasoning attached, a dashboard that answers the follow-up question in the room, and one document per site sized for the packet. The test: can someone who was not in the analysis explain, from your materials, why this site beat the other four?
Finding Whitespace & Market Opportunities
Whitespace is a market where demand for your concept exists and you are not there yet. Finding it is the difference between an expansion plan and a list of sites brokers happened to send.
Whitespace analysis scans every market against your performance model rather than against a hunch, which is how a brand ends up finding demand in a metro nobody on the team had proposed.
Void reports score every candidate market against the profile of your stores that already work. Co-tenant analysis shows which neighboring businesses actually share your customer, based on where visitors go before and after, rather than on which brands feel adjacent. Saturation scores answer whether a market has room left. For a complete walkthrough, see our guide on the retail site selection process.
Competitive Analysis & Benchmarking
Competitor data answers a narrower question than most teams ask of it: not whether the competition is good, but whether this trade area still has unserved demand in it.
Counting storefronts is the shallow version. What matters is how each competitor performs and how directly it sits between you and your customer:
- Adjacent competitors: Located in the same shopping center, they have the highest impact.
- Impacting competitors: Within a short drive, they serve the same customer base.
- Intercepting competitors: Situated along common customer routes, they can divert traffic.
A cluster of busy competitors is ambiguous on its own. It can mean the market is proven or that it is full, and the tiebreaker is usually whether their visit volumes are still growing.
Performance Forecasting & KPI Setting
The ultimate goal of site selection analytics is to translate data into reliable performance predictions. This gives you realistic expectations for a new location.
Sales projections are the foundation, and the useful ones come as a range with a stated error rate rather than a single confident number. Ask any vendor what their forecast error is on your format, and how they measured it. GrowthFactor customers report forecast error roughly half the industry norm (source: GrowthFactor customer survey, January 2026).
The breakeven timeline matters as much as the revenue number, and the forecast doubles as the benchmark you track the store against after it opens.
Implementing a Data-Driven Site Selection Analytics Process
Rolling this out is an organizational change more than a software install. Four things decide whether it sticks.
- Governance: Agree on the criteria and the weights before the first site runs through, not while arguing about a score.
- The team: Real estate, finance, and operations in the same review, so the store that scores well is also one you can staff and fund.
- Data quality: A stale demographic file or a store list with bad coordinates produces a confident wrong answer. Fix the inputs before blaming the model.
- Change management: The first time the model disagrees with a senior person's instinct decides whether anyone uses it again. Have that conversation early, over a site nobody is emotional about.
Step-by-Step Quick-Start Checklist
Five steps, in order.
- Define the goal. Revenue, market entry, or blocking a competitor. Each one weights the criteria differently.
- Gather both halves. External market data, plus your own sales history. The internal half is what makes the model yours.
- Model it. Real trade areas, competitive pressure, and cannibalization against your existing stores.
- Go see it. Visibility, access, and parking do not appear in a database.
- Review the decision. Present the score with its inputs, not just the number.
For a deeper dive, see our guide on real estate site selection.
Post-Opening Monitoring & Optimization
The loop closes after the store opens. Track each location against the forecast that justified it, and the gaps tell you where the model is weak: consistently over on urban infill sites, consistently under on highway pads, and so on.
That comparison is the only real accuracy check anyone has, and it is also what makes the next forecast better. A brand that never records what it predicted has no way to improve, and no answer when the committee asks how often the model has been right.
Who Builds Site Selection Analytics Tools
The vendor landscape moved in 2026, and two of the names in every RFP now answer to something different.
Buxton relaunched as Audiense on July 8, 2026, folding its customer-analytics and location products into one brand after acquiring Audiense in 2025 and Elevar in 2024 (announcement via PR Newswire). Buxton's location tools continue as product names inside the Audiense portfolio, so a proposal that still says Buxton is not necessarily stale. Ask who you would be contracting with.
SiteZeus launched Atlas on May 12, 2026, a rebuild of its Locate product with conversational AI, live for-sale and for-lease inventory on the map, and one-click forecasting from a listing (announcement via PR Newswire). The company keeps its name; Atlas is the platform.
The rest of the field sorts by what they sell. Placer.ai sells foot traffic depth. Esri sells the GIS substrate half the industry builds on. CoStar sells the property and lease side rather than the customer side. Kalibrate and Sitewise sell forecasting with a consulting relationship attached. GrowthFactor sells the workflow: scoring, trade areas, and deal pipeline in one place, with the inputs behind every score visible. Our guide to retail site selection software covers where each fits.
Software or a Consultant?
Both are legitimate, and the choice is mostly about frequency. A consultant produces a study, the right shape for a question you ask once: a new country, a distribution center, one large capital decision in front of a board. Software produces a habit, the right shape when site decisions arrive monthly and cost per site matters more than the depth of any one report.
The line sits around a handful of openings a year. Below it, a consultant is cheaper and the study is enough. Above it, the per-study fee stops making sense and nothing accumulates: every engagement starts over, while a platform keeps last year's decisions where this year's model can learn from them.
What Site Selection Analytics Cannot Tell You
Every guide to this topic is written from the upside. A team that expects the model to answer everything stops trusting it the first time it is wrong about something it was never given.
Three problems recur. Data quality beats model sophistication: a stale demographic file or a store list with wrong coordinates produces a confident wrong answer with no warning. Small portfolios limit what a model can learn, since a brand with 12 stores has 12 data points and should read a forecast as a range. Changing conditions undo good work quietly, because a trade area analyzed before a highway realignment describes a place that no longer exists.
Then there is the category the data does not cover at all.
None of that is an argument against the analysis. It is an argument for reading the score as what it is: a ranked shortlist with the reasoning attached, which beats a stack of broker emails. The analysis narrows the field and shows its work. The call is yours.
Frequently Asked Questions about Site Selection Analytics
What is the difference between site selection analytics and traditional market research?
Traditional market research relies on surveys, focus groups, and periodic reports that quickly become outdated, while site selection analytics draws on continuously refreshed datasets including foot traffic, mobility patterns, and consumer spending signals. Site selection analytics produces faster, more objective location scores rather than qualitative opinions about market potential.
How can site selection analytics prevent cannibalization?
Overlap analysis maps the real trade areas of your current and proposed locations and reports how much of the new site's catchment an existing store already serves. Sales transfer models then estimate how much revenue would move rather than being created, so you can see whether an opening adds to the portfolio or just reshuffles it. Brands that skip this step often book a strong forecast for the new store and quietly absorb the drop at the old one.
Do I need site selection analytics software or a site selection consultant?
A consultant is the better fit for a one-time question that needs a defensible answer, such as entering an unfamiliar country or siting a distribution center, because the deliverable is a study rather than a habit. Software wins once site decisions become routine, since the cost per site falls with volume and every past decision stays in one place to calibrate the next one. Teams opening more than a handful of locations a year usually end up running software continuously and hiring a consultant for the occasional problem it was never built for.
How do retailers validate site selection analytics models before relying on them for expansion decisions?
Model validation typically involves back-testing: running the model against the brand's existing location portfolio to measure how accurately it would have ranked high-performing versus low-performing stores based on the variables available at the time of opening. Analysts look for strong rank-order correlation between predicted and actual performance before deploying the model for new site decisions.
How does GrowthFactor compare to Placer.ai for site selection analytics?
Placer.ai's device panel is the category benchmark for foot-traffic depth, offering granular visitor data that analytics teams rely on. GrowthFactor integrates foot traffic with demographic, vehicle traffic, and competitive data behind a single scoring and deal management workflow, removing the spreadsheet layer most Placer.ai users build. TNT Fireworks used GrowthFactor's integrated analytics to screen and open 153 locations on budget, with every site decision backed by explainable five-lens scoring.
Where to Start
If you run site decisions out of a spreadsheet today, the first upgrade is not a platform. It is writing down what a good store looks like for you: the trade area, the co-tenants, the access, the rent you can carry. That list is what any model calibrates against, and most teams have never put it on paper.
From there, follow the checklist above. GrowthFactor customers evaluate around five times more sites than they did by hand, which mostly means the obvious rejections stop consuming a week each.
Our all-in-one real estate platform for retail brings scoring, trade areas, and deal pipeline into one place, with every input behind a score visible to whoever asks.