Not every incrementality test needs a complex control group design. Sometimes the simplest approach is the most practical: run the campaign in some stores and not in others.
Geo-fencing tests use geographic boundaries to create natural test and control groups.
Campaign stores get the media. Holdout stores don't. Compare sales across the two sets. The difference is your estimate of campaign impact.
How it works
Step 1: Select test and control stores. Match stores by sales volume, category performance, demographic profile, store format, and competitive environment. The goal is to create two groups that are as similar as possible pre-campaign.
Step 2: Run the campaign in test stores only. In-store screens, digital targeting limited to test store shoppers, any store-level activation. Control stores receive no campaign activity.
Step 3: Measure the difference. Compare category and brand sales in test stores vs.
control stores during the campaign period. Adjust for any pre-existing differences using the pre-campaign baseline.
Step 4: Estimate incrementality. The sales difference between test and control stores, adjusted for baseline, is the estimated incremental impact of the campaign.
Why geo-fencing is practical
It's the simplest store-level experiment available. No complex identity matching required. No audience-level randomization. Just stores with the campaign and stores without.
For in-store media, screens, radio, point-of-sale, geo-fencing is often the most natural test design. The media is physically deployed in specific stores, which creates a natural test/control split.
It's also easy to explain. "We ran the campaign in 200 stores and compared to 100 similar stores that didn't have the campaign. Sales were 7% higher in campaign stores."
That's a story anyone can understand, procurement, finance, category management.
The limitations
Geo-fencing has known limitations:
Cross-contamination. Shoppers don't respect geographic boundaries. A shopper exposed to the campaign in Store A might make their purchase in Store B (a control store). This dilutes the measured effect in test stores and contaminates the control.
Store-level variation. Even well-matched stores have differences, local competition, manager effectiveness, staffing, local events. With small store counts, these differences can overwhelm the campaign signal.
No individual-level attribution. You know the store-level effect, not the shopper-level effect. You can't measure new-to-brand rate, repeat rate, or share of wallet, because you can't track individual shoppers across the test/control split.
Digital complexity. If the campaign includes digital channels (offsite, CRM, website), limiting exposure to test-store shoppers only is technically challenging. Digital exposure leaks across geographic boundaries.
When to use geo-fencing
Geo-fencing is best for: - Pure in-store campaigns where exposure is physically contained within test stores - Large store counts where statistical power is sufficient (50+ stores per group minimum) - Quick reads where a directional result is needed fast
- First tests where the brand wants evidence before investing in a full measurement program
It's not ideal for: - Omnichannel campaigns with digital components - Precise individual- level measurement - Small-scale tests with limited store counts
Combining with shopper-level measurement
The most robust approach combines geo-fencing with shopper-level measurement.
Geo-fencing provides the store-level view: did campaign stores outperform control stores? Shopper-level control groups (using loyalty data) provide the individual-level view: did exposed shoppers outperform matched unexposed shoppers?
When both methods agree, confidence is high. When they disagree, it reveals something interesting, maybe the in-store effect is strong but the digital effect is weak, or vice versa. The combination produces richer insight than either method alone.
The bottom line
Geo-fencing is the simplest store-level experiment in retail media. Compare exposed stores to non-exposed stores. Measure the difference.
It's not the most precise method. It has contamination risks, store-level noise, and no individual attribution. But it's practical, intuitive, and often the fastest path to a directional incrementality estimate.
Use it as a starting point. Build toward shopper-level measurement for precision. And always be transparent about what the method can and can't prove.
Related Reading
- Cost-to-Serve: The Hidden Reason Unbundled Pricing Breaks Operations
- Omnichannel ROAS: One Outcome Across Onsite, Offsite, and In-Store
- Synthetic Testing: Proving Retail Media Impact When Control Groups Aren't Possible
- Minimum Proof Package: The Measurement Bundle That Makes Buyers Renew
- Category Index: How to Tell If You Beat the Category, Not the Calendar
Ready to see how this works in practice?
Footprints AI helps brands and retailers measure what matters. See our customer success stories or get in touch to discuss your retail media strategy.
