Retail Site Selection Data Stack: POI to Spend

Retail site selection works when teams combine signals instead of worshiping one metric. POI data tells you what is there; foot traffic tells you who shows up and when; spend panels tell you whether visits translate to wallet share; mobility and origin-destination data tell you where demand comes from. The best 2026 programs use all four with explicit validation checks, especially when evaluating expansion markets, co-tenancy, cannibalization, and post-opening performance. Start with GSDSI POI data, POI and geofencing, Global Mobility and Location Data, and competitive benchmarking. Pair with POI data quality in depth and the geo-panel audit before licensing production feeds.

Key Takeaways

  • POI quality is the foundation. Bad polygons and duplicate venues corrupt every downstream traffic or spend read.
  • Foot traffic needs panel math. Daily unique devices, observation bias, and visit classification matter more than raw visit counts.
  • Spend panels validate monetization. They help distinguish busy sites from profitable trade areas.
  • Origin-destination data explains draw. A site with lower visits can be better if it attracts the right home/work trade area.
  • Do not overfit one quarter. Weather, tourism, roadwork, and competitor openings can distort short windows.

Expansion committees should publish a written decision standard before vendors present samples: minimum POI match rate on a labeled venue list, minimum daily unique devices in the trade area, spend-panel category coverage, and OD stability across four quarters. Without a written standard, bake-offs become storytelling contests. The standard can be updated quarterly, but it must exist so real estate, finance, and analytics grade vendors on the same rubric rather than on whichever deck animates best in the room.

When two vendors disagree on the same candidate site, decompose the disagreement: POI boundary differences, panel coverage differences, spend normalization differences, or OD window differences. Committees that pick a winner without decomposition often discover the error after lease execution, when the cheaper vendor's optimism was a polygon issue rather than a market insight.

Start With POI Quality and Geofencing

Every site-selection stack starts with the venue definition. If the POI record uses a centroid instead of a true polygon, if a mall tenant shares the parent polygon, or if multiple tenants occupy one address without clean unit boundaries, traffic attribution will drift. Require polygon provenance, category taxonomy, open/close dates, brand normalization, and duplicate-resolution rules before reading visit counts. The OpenStreetMap ecosystem, commercial POI providers, and property records can all help triangulate reality, but no single source is perfect: multi-source verification is the durable architecture.

For dense retail corridors, test geofences against satellite imagery and store lists before modeling. The cost of a bad polygon is not just a wrong map; it is a wrong expansion decision. See why POI data quality makes or breaks foot traffic analytics for the polygon-versus-radius decision framework.

Co-tenancy analysis requires unit-level POI boundaries inside parent polygons, not one mall outline for every tenant. Expansion teams that skip unit boundaries routinely overstate traffic for in-line spaces and understate anchor draw.

Foot Traffic and Mobility: Read the Panel, Not the Hype

Foot traffic can reveal daypart, dwell, repeat visitation, cross-shopping, and catchment patterns. But the signal depends on panel coverage, consent provenance, and visit classification. Ask for daily unique devices in the trade area, confidence intervals for small venues, and how the vendor handles employees, passersby, parking lots, and adjacent stores. The foot traffic panel sizing guide explains why current panel math should be evaluated market by market, not from a national headline count.

Spend Panels and Demand Validation

Spend panels answer a different question than traffic: whether a market monetizes. A busy corridor may underperform for your category if basket size, income fit, or competitor density is wrong. Card or receipt panels help validate category demand, share shift, and post-opening ramp. Use them with caution: panel composition, merchant normalization, online/offline splits, and refund handling all matter. Connect to commercial real estate signals when the decision is investor-grade.

Public anchors such as U.S. Census demographic tables and BLS QCEW employment data help sanity-check vendor panels. If every commercial panel says a trade area is premium but census and lease comps disagree, slow down before signing.

Origin-Destination and Cannibalization

Origin-destination matrices show where visitors actually come from: often a better trade-area definition than a drive-time ring. For portfolio operators, OD overlap between a candidate site and existing stores estimates cannibalization risk before lease execution. See how CRE investors use origin-destination data for the institutional workflow.

Require ≥2,000 attributed visits over the analysis window for tract-level OD reads; use zip grain for smaller-volume POIs. Stability quarter-over-quarter is itself a signal: material shifts usually mean a competitor opening, access change, or demographic drift.

A Practical Operating Model for Expansion Teams

  1. Define the decision: market entry, relocation, competitor benchmark, or post-opening measurement.
  2. Validate POIs and polygons before importing traffic or spend data.
  3. Read traffic, spend, and origin-destination as separate lenses, then score overlaps.
  4. Run holdout validation against known stores before applying the model to candidate sites.
  5. Refresh the model after openings, closures, road changes, and major competitor moves.

GSDSI packages POI, mobility, and adjacent commercial signals into buyer-specific pilots through the pilot process. Score vendors with the RFP matrix so real estate, finance, and analytics review the same evidence.

Finance teams should see the stack as a sensitivity model, not a single score. Run scenarios where traffic is strong but spend is weak, where OD draw is wide but visit counts are modest, and where a competitor opening shifts cross-shopping. Expansion committees that debate scenarios outperform teams that debate a single composite index from one vendor dashboard.

Post-opening measurement is the closing loop: compare forecasted traffic and spend panels to actuals at 30, 90, and 180 days. Feeds that fail post-opening validation should trigger vendor review even if the pre-opening bake-off looked excellent: panel composition and competitive dynamics change after you enter a market.

Governance Handoff to Real Estate and Finance

Package the final site memo with POI version IDs, panel vendor, spend panel vendor, and OD methodology window. When leadership revisits a decision twelve months later, those metadata fields explain why the read differed from a broker deck: without them, teams relitigate anecdotes.

Align incentives: brokers bring relationships, data brings measurement. The stack does not replace brokers; it gives finance a defensible second opinion when rent and capital calls are on the line.

Frequently Asked Questions

What is the most important data source for retail site selection?
There is no single source. POI quality is the foundation, foot traffic shows behavior, spend panels validate monetization, and origin-destination data explains trade-area draw. The best models combine them and validate against known stores.
Can foot traffic alone predict store revenue?
Not reliably. Foot traffic is a strong behavioral signal, but revenue depends on category fit, basket size, competition, pricing, store operations, and local demand. Spend panels and known-store validation help bridge that gap.
Why do POI polygons matter so much?
A wrong polygon assigns visits to the wrong venue. In malls, mixed-use properties, and dense retail corridors, polygon errors can make a weak site look strong or hide a strong tenant inside a parent property.
How do spend panels and foot traffic disagree?
High traffic with weak spend may indicate lookie-loos, wrong category fit, or panel bias. Weak traffic with strong spend may indicate a destination with a small but high-value customer base. Read both before committing capital.
How often should site-selection models refresh?
Refresh when major market facts change: store openings and closures, road or transit changes, competitor moves, seasonal shifts, tourism changes, or panel methodology updates. Quarterly refreshes are common for active expansion teams. Document the refresh trigger in the committee memo so future teams know why the model was rerun.