Choosing where to open is the highest-stakes decision a growing brand makes. A single new location commits roughly $1–10 million in capital across a decade-long lease, and the wrong site can’t be repriced the way a landlord reprices a building — it drags four-wall economics for the length of the term. In the second half of 2026, the way brands make that call is being rebuilt around data and AI.
Retail site selection is the process of deciding where a brand opens physical locations — which markets to enter, which trade areas to target, and which specific sites will perform. The shift underway in 2026: site decisions are moving from instinct and static reports to AI-driven scoring and predictive revenue models calibrated to a brand’s own stores.
The backdrop makes precision matter more, not less. The National Retail Federation projects U.S. retail sales will grow 4.4% in 2026 to $5.6 trillion— stronger than the 3.6% ten-year average — but warns the consumer is increasingly bifurcated, with higher-income households driving most of the growth. Demand is real; it’s just unevenly distributed across markets and trade areas. That’s exactly the kind of environment where a data-driven site decision separates a location that compounds from one that quietly underperforms for ten years.
Data-Driven Decisions Are Now the Baseline
Physical retail is not in retreat. In Placer.ai’s early-2026 survey of roughly 500 retail, CRE, and CPG leaders, more than 55% expressed confidencein brick-and-mortar performance over the coming year, against only about 20% who were concerned. And despite years of “everything is going online” narratives, e-commerce still accounts for just 16.2% of U.S. retail sales by dollar value— meaning roughly 84 cents of every retail dollar is still spent in a store. The store is where the money is; the question is which store, where.
What has changed is the standard of proof. Using mobile and demographic data for site selection is now table stakes, and the brands pulling ahead treat site forecastingas a repeatable discipline rather than a gut call. Backward-looking reports tell you which bet went wrong six months ago; the current generation of retail location analytics is forward-looking — estimating what a site will do before the lease is signed.
Modern site selection runs on two distinct engines. Site scoringscreens many candidate locations quickly against your brand’s criteria — it narrows the funnel. Predictive revenue forecasting then underwrites a shortlisted site by modeling expected sales and cannibalization. Scoring finds the contenders; forecasting decides the deal. The strongest platforms do both.
The split below is the simplest argument for why this discipline pays off: the overwhelming majority of retail spending still happens in physical locations, so the cost of getting a location wrong is paid in real, hard-to-recover capital.
The Technologies Reshaping the Map
AI and machine learning
AI has moved from a feature label to the core engine of site selection. It now screens sites automatically against brand-specific criteria, answers natural-language questions about a market, and improves as more performance data flows through it. Walmart’s demand systems already predict at zip-code-level precision; the same logic is reaching expansion decisions. The momentum is hard to overstate — the geospatial AI market is projected to grow from roughly $60 billion in 2025 to $472 billion by 2034, and Gartner reports that the large majority of retail IT leaders rank AI as their top technology priority. The practical takeaway: the gap between AI-equipped and AI-absent site selection widens every quarter.
GIS and geospatial analytics
Geographic information systems remain the canvas — the layer that stacks demographics, foot traffic, competition, and co-tenancy into a single view of a market. It’s a fast-growing canvas: the global location intelligence market is on track to roughly double from $25 billion in 2025 to $47 billion by 2030, and retail is its single largest end-user vertical at about a quarter of the market. When multiple data layers combine, they reveal a picture of a trade area no single source can produce alone.
Predictive analytics and revenue forecasting
This is where a shortlist becomes a decision. Predictive models identify analog stores — existing locations that most resemble a candidate site across key variables — and project performance from how those analogs actually perform. The most reliable forecasts are built on a brand’s own data, not industry averages, because different concepts are driven by different variables: a gym’s revenue tracks membership density, a frozen-dessert brand’s tracks foot-traffic seasonality. The model also estimates trade-area overlapso a new unit doesn’t quietly cannibalize an existing one.
A site score tells you whether a location is promising. A revenue forecast tells you whether it pencils.
To make this concrete, the model below is a simplified version of how a modern site score comes together. Move the inputs and watch the composite score and revenue band respond — the same logic Locate runs across many more variables, calibrated to a brand’s actual stores.
Site Score, simplified
Adjust the inputs for a candidate location
How Shopping Habits Are Redrawing the Map
The omnichannel, daily-route store
Shopping “mostly online” is not the leading method for any generation or category, according to McKinsey’s 2026 work with ICSC. Instead, hybrid work has reshaped where convenience lives: locations embedded along daily routes — near home, school, the gym, or essential services — now carry outsized value. Thirty-seven percent of shoppers rank in-stock reliability among their top reasons for choosing a retailer. For site selection, that elevates accessibility, visibility, and proximity to daily-life anchors over raw demographics.
Generational spending power is shifting
The customer is changing underneath the map. Gen Z and Millennials increasingly shop across both channels and gravitate to curated, showroom-style formats and pop-ups. Meanwhile Gen X — just 19% of the population — drives about 31% of retail sales, and Baby Boomers and Gen X together still control nearly two-thirds of U.S. retail dollars. The implication for site selection is blunt: a market-level average tells you almost nothing. What matters is whether a specific trade area’s customers resemble the customers at your best-performing stores.
Durable Locations Beat Trendy Ones
As capital gets more expensive and the consumer stays selective, the locations that hold up are the ones anchored in durable, everyday demand. Placer.ai’s 2026 outlook flags the strongest growth confidence in categories tied to routine — wholesale clubs, grocery, mass merchants, value retail — and a durable role for top-tier centers, with select Tier 2 properties opening up as space tightens. A site embedded in a community’s daily pattern, with a tenant mix that reinforces rather than competes, ages better than one chosen for a temporary hot corridor.
Preparing for What’s Next
The throughline for the rest of 2026 is simple: demand is healthy but uneven, capital is expensive, and the cost of a wrong location is unforgiving. The brands that win treat site selection as a repeatable, data-driven discipline — scoring to narrow the field, forecasting to underwrite the deal, and calibrating both to their own stores rather than industry averages. The intuition of an experienced operator still matters; it just works far better with evidence underneath it.
A weak location can’t be undone cheaply. It sits on the P&L for the length of the lease — and shows up in the numbers any buyer scrutinizes at exit. In a higher-cost environment, the return on getting site selection right has never been larger.
Common Questions
- What is retail site selection?
- It’s the process of deciding where a brand opens physical locations — which markets, which trade areas, and which specific sites will perform. Modern site selection pairs demographic, foot-traffic, competitive, and co-tenancy data with predictive models to forecast performance before a lease is signed.
- How is AI used in retail site selection?
- In two complementary ways: site scoring screens many candidate sites quickly against brand-specific criteria, and predictive revenue forecasting models expected sales and cannibalization for a shortlisted site using analog stores from the brand’s own portfolio.
- What factors matter most in 2026?
- Customer and psychographic match to a brand’s best stores, foot-traffic density and timing, competitive saturation, co-tenancy, drive-time accessibility, and trade-area overlap. Market-level averages matter far less than the specifics of an individual trade area.
- How do you forecast revenue for a new location?
- With analog modeling: the model finds existing locations that most resemble the candidate site, then projects performance from how those analogs actually perform. The most reliable forecasts are calibrated to a brand’s own sales data rather than industry averages.