Site Selection Analysis: The Market Research Methods That Work

Site selection analysis is the process of evaluating potential locations for a business operation using a combination of demographic, competitive, economic, and behavioural data. Done well, it reduces the risk of expensive location decisions by grounding them in evidence rather than instinct or convenience.

The methods range from straightforward trade area mapping to sophisticated multi-variable scoring models. Which approach you need depends on the complexity of your decision, the data available, and how much is riding on getting it right.

Why Site Selection Analysis Is a Market Research Problem First

Most businesses treat site selection as a property decision. They hand it to the real estate team, who evaluate footfall, rent per square foot, and lease terms. Those things matter. But the question underneath all of them is a market research question: is there sufficient demand in this location, and can we capture enough of it to make the economics work?

I’ve sat in enough strategic planning sessions to know that the property conversation and the market research conversation rarely happen in the same room. That gap is where expensive mistakes get made. A site that looks attractive on footfall data can be surrounded by four direct competitors already serving the same customer base. A site with modest footfall can sit inside a trade area with almost no competitive supply and strong demographic alignment.

The methods covered here are primarily market research methods applied to location decisions. They’re not about property valuation. They’re about understanding demand, competition, and the realistic revenue potential of a given location before you commit.

If you’re building out a broader market research capability, the Market Research and Competitive Intelligence hub covers the wider toolkit, including competitive monitoring, search intelligence, and how to structure research programmes that actually inform decisions.

What Is a Trade Area and Why Does Its Definition Matter So Much?

A trade area is the geographic zone from which a location draws the majority of its customers. Defining it accurately is the foundation of every other analysis. Get it wrong and your demographic data, competitive mapping, and demand estimates are all built on a flawed base.

The most common mistake is using a simple radius circle. A 2-mile radius sounds intuitive, but it ignores the fact that customers travel by roads, not as the crow flies. A motorway junction, a river, a train line, or a one-way system can compress a trade area dramatically in one direction while extending it in another.

Drive-time polygons are considerably more reliable. These map the area reachable within a set experience time, typically 5, 10, and 15 minutes for retail or quick-service businesses, and they respect the actual road network. For businesses where customers travel on foot, walk-time isochrones serve the same purpose. Most modern GIS platforms generate these automatically from postcode or coordinate inputs.

The other dimension of trade area definition is penetration rate by zone. Primary trade areas typically contain the customers who visit most frequently, often representing 60 to 70 percent of volume from the smallest geographic footprint. Secondary and tertiary zones contribute the rest at lower frequency. Understanding this layered structure matters when you’re modelling revenue potential, because a large tertiary zone with weak penetration can look impressive on a map while contributing very little actual volume.

Demographic Analysis: What the Data Tells You and What It Doesn’t

Once you have a defined trade area, demographic analysis tells you who lives and works within it. Age distribution, household income, family composition, employment type, and population density are the standard variables. For most consumer-facing businesses, these form the first filter in site evaluation.

The data sources vary by market. In the UK, Census data combined with commercial geodemographic segmentation products like ACORN or Mosaic gives you a reasonably detailed picture at small area level. In the US, Census Bureau data and American Community Survey outputs are the standard starting point, often enriched with commercial data from providers like Esri or CoStar.

The limitation worth keeping in mind is that demographic data describes the population, not their behaviour. A trade area with high household incomes and the right age profile for your offer doesn’t guarantee those households will choose you over existing alternatives. Demographic alignment is a necessary condition, not a sufficient one. I’ve seen businesses open in apparently ideal demographic pockets only to discover that the local market was already heavily served, or that purchasing behaviour in that area skewed toward channels the business wasn’t competing in.

Daytime population data adds an important layer for businesses that depend on workers rather than residents. A city centre location might have a modest residential population within its trade area but a daytime population three or four times larger. For coffee shops, lunch restaurants, or business services, that distinction is the difference between a viable site and an unviable one.

Competitive Mapping: Understanding Supply in the Market

Competitive mapping plots the location and characteristics of existing competitors within and adjacent to your trade area. It tells you about the supply side of the equation: how many alternatives exist, where they’re concentrated, and what gaps, if any, are present.

The inputs are typically a combination of business directory data, commercial property databases, and primary research. For categories with significant online presence, search data can supplement this. A category search on Google Maps for your target location gives you a fast, if imperfect, view of the competitive landscape as customers would see it.

What competitive mapping doesn’t tell you is how well those competitors are actually performing. A cluster of three competitors in a trade area could mean the market is oversupplied and margin-thin, or it could mean demand is strong enough to support multiple operators. You need demand-side data to interpret the supply picture correctly.

One approach worth considering is using web traffic data as a proxy for competitor performance. Tools like Similarweb can give you directional estimates of a competitor’s digital footfall, which, for businesses with significant online-to-offline conversion, can indicate relative strength. It’s an imprecise signal, but it’s better than treating all competitors as equivalent regardless of their actual market position. The broader point about using web analytics as a competitive signal is covered in more depth across the market research hub.

Gravity Models: Quantifying the Pull of Competing Locations

Gravity models are borrowed from physics and adapted for retail geography. The underlying logic is that a location’s attractiveness to customers increases with its size or offer strength and decreases with distance. Customers will travel further for a larger, better, or more distinctive offer than they will for a smaller or more generic one.

The classic formulation is Reilly’s Law of Retail Gravitation, which calculates the boundary point between two competing centres based on their relative sizes and the distance between them. Huff’s Probability Model extends this into a probabilistic framework: rather than drawing a hard boundary, it estimates the probability that a customer at any given location will choose each competing site.

In practice, gravity models require calibration. The distance decay parameter, which controls how sharply customer probability drops with distance, varies significantly by category. Customers will drive 45 minutes for a specialist retailer they can’t find elsewhere. They won’t drive 10 minutes for a convenience store. Using an uncalibrated model produces outputs that look precise but are built on assumptions that don’t hold in your specific category.

For businesses with existing locations, calibrating the model against observed customer origin data from loyalty programmes, card transaction data, or customer surveys gives you a much more reliable distance decay parameter than any industry default. This is one of the cases where proprietary data, even at modest scale, outperforms published benchmarks.

Index Scoring Models: Comparing Multiple Sites on a Common Scale

When you’re evaluating multiple candidate sites simultaneously, index scoring models give you a structured, comparable framework. The approach involves selecting the variables that predict performance in your category, assigning weights to each based on their relative importance, scoring each site against those variables, and combining the scores into a single index.

The variables typically span three categories: demand indicators (demographic alignment, population density, daytime population), supply indicators (competitor density, gap analysis), and access indicators (footfall, visibility, car parking, public transport proximity). The weights assigned to each reflect your business model. A destination retailer might weight demographic alignment and trade area size heavily. A convenience format might weight footfall and access much more heavily than demographics.

The discipline in building a scoring model is resisting the temptation to reverse-engineer it toward a site you’ve already decided you want. I’ve seen this happen in planning processes where the model starts as an objective tool and ends up being adjusted until it confirms a decision that was made for other reasons, usually because someone senior likes the location or has already had a conversation with the landlord. When that happens, the model stops being analysis and starts being theatre.

The most useful check is to validate the model against your existing estate if you have one. Score your current locations using the model and see whether the index correlates with actual performance. If it doesn’t, the model needs recalibration before you use it to evaluate new sites. This validation step is often skipped, which is why many scoring models produce confident outputs that don’t hold up against observed results.

Analogue Analysis: Learning From Comparable Locations

Analogue analysis uses the performance of existing locations with similar characteristics to forecast the potential of a new site. If you’re opening a 12th location and you have 11 operating, the performance data from those sites is among the most valuable inputs you have.

The method involves identifying analogue sites, existing locations that closely match the candidate site on key variables: trade area demographics, competitive environment, access characteristics, and format. You then use the revenue and performance data from those analogues to build a range of forecasts for the new site.

The quality of analogue analysis depends entirely on the quality of your site classification. If you’ve been rigorous about recording the characteristics of each existing location, finding close analogues is straightforward. If your existing estate data is inconsistent or incomplete, the analysis becomes much harder to do reliably.

For businesses opening their first or second location, analogue analysis using competitor data is the alternative. This requires primary research: identifying competitors in comparable markets, estimating their revenue through whatever proxies are available (employee counts, queue observations, transaction volume estimates, planning application data), and using those estimates as a benchmark range. It’s less precise than using your own data, but it’s considerably better than forecasting from nothing.

Footfall Data and Mobility Analytics: The Newer Layer

Mobile device location data has added a significant new layer to site selection analysis over the past several years. Aggregated and anonymised mobility data from smartphone signals can tell you how many people pass through or visit a location, where they come from, how long they stay, and how their behaviour varies by time of day or day of week.

Providers like Placer.ai, Unacast, and StreetLight Data have built commercial products around this data. The use cases in site selection include validating footfall estimates, understanding the catchment area of existing competitor locations, identifying underserved corridors, and comparing candidate sites on observed traffic patterns rather than modelled estimates.

The limitations are worth being clear about. Mobile location data is a sample, not a census. Panel composition varies by provider, and the methodology for extrapolating from panel to population differs. The data is strongest in high-traffic urban environments and weakest in rural or low-density areas where sample sizes get thin. For directional comparison between candidate sites, it’s a useful input. For precise traffic forecasting, it needs to be treated with appropriate caution.

There’s also a question of what footfall actually predicts in your category. For impulse-driven formats, footfall is highly predictive of revenue. For destination businesses where customers plan their visit, it’s much less so. The correlation between passing traffic and conversion varies enormously by business type, and assuming a direct relationship without testing it against your own data is a common analytical error.

Integrating the Methods: What a Rigorous Site Selection Process Looks Like

The methods described here are not alternatives to each other. They’re layers that address different aspects of the same question. A rigorous site selection process typically works through them in sequence, using each layer to either confirm or challenge the conclusions from the previous one.

A reasonable sequence for a retail or service business might look like this. Start with trade area definition using drive-time polygons. Run demographic analysis against your customer profile to assess alignment. Map the competitive supply within the trade area. Apply a gravity model to estimate realistic market share given the competitive environment. Score the site using your index model and compare it against other candidates. Validate the top candidates using analogue analysis and, where available, mobility data. Then make the decision.

The process sounds linear, but in practice it’s iterative. Competitive mapping sometimes reveals that the trade area definition needs adjusting because a major competitor sits just outside the boundary and draws from the same population. Analogue analysis sometimes surfaces variables that weren’t in the scoring model. The value of working through multiple methods is precisely that they create opportunities to catch errors and challenge assumptions before they become expensive commitments.

One thing I’d push back on is the idea that more data automatically produces better decisions. I’ve worked with businesses that had access to sophisticated geodemographic platforms, mobility data feeds, and custom scoring models, and still made poor location decisions because the analysis was never seriously interrogated. The methods are only as useful as the quality of thinking applied to them. A simple, well-reasoned analysis using publicly available data will outperform a complex model that nobody in the room really understands or is willing to challenge.

For those building out a broader research and intelligence capability, the methods discussed here sit within a wider framework. The Market Research and Competitive Intelligence hub covers complementary approaches including competitive monitoring, demand analysis, and how to structure research programmes that connect directly to commercial decisions.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions