Measuring incremental store visits shows whether location-based advertising created additional foot traffic or merely reached people who were already likely to visit. This guide explains how to design an exposed-versus-control study, define visit windows, estimate lift and cost per incremental visit, check for bias, and turn the results into a repeatable reporting process.
Overview
Basic foot traffic attribution answers a narrow question: did people exposed to an ad later appear at a store? Incrementality asks the more useful business question: did the campaign cause visits that would not otherwise have happened?
The difference matters because location-based advertising often reaches people who already live, work, shop, or travel near a location. A campaign can report many attributed visits while producing limited additional revenue if most of those visitors would have arrived without seeing an ad.
A practical store visit lift study compares a group exposed to the campaign with a similar control group that was eligible for the campaign but did not receive the measured exposure. The core calculation is:
Incremental visit rate = exposed visit rate − control visit rate
Then calculate:
- Incremental visits = exposed audience size × incremental visit rate
- Relative lift = (exposed visit rate − control visit rate) ÷ control visit rate
- Cost per incremental visit = campaign cost ÷ incremental visits
- Incremental revenue = incremental visits × average incremental revenue per visit
These formulas are simple, but their reliability depends on sound audience selection, consistent measurement, sufficient sample size, and privacy-safe data handling. For broader KPI planning, see Location Analytics Dashboard Metrics.
How to estimate incremental store visits
1. Define the campaign and measurement unit
Specify the campaign objective, locations, channels, audience, budget, and dates before launch. Decide whether the unit of analysis is a person, device, household, hashed identifier, or geographic area. The available unit should be documented because it affects deduplication and how results are interpreted.
Use a consistent store-visit definition. For example, a qualifying visit might require a location signal within a store boundary for a minimum dwell period, subject to the measurement provider's methodology. Avoid changing the boundary, dwell rule, or data source between exposed and control groups.
2. Create comparable exposed and control groups
The exposed group contains eligible users who received the campaign exposure. The control group should come from the same eligible population and have a comparable opportunity to visit the measured stores, but should not receive the ad being evaluated. Random assignment is generally the clearest design when the platform supports it. If randomization is unavailable, use matching or geographic controls and document the limitations.
| Measure | Exposed group | Control group |
|---|---|---|
| Eligibility | Meets the campaign audience and geography rules | Meets the same rules |
| Advertising exposure | Receives the measured ad or channel exposure | Does not receive the measured exposure |
| Visit rate | Qualifying visits ÷ exposed audience | Qualifying visits ÷ control audience |
| Incrementality role | Observed outcome after exposure | Estimate of expected visits without exposure |
| Main risk | Self-selection, frequency effects, or cross-channel exposure | Contamination, weak matching, or unequal access to stores |
3. Choose the visit window
Set a primary post-exposure window before reviewing results. A short window may capture direct response but miss delayed visits; a long window may include unrelated trips and dilute the campaign effect. The appropriate window depends on the product, purchase cycle, distance to stores, and campaign objective.
Consider reporting a primary window plus one or two secondary windows, such as visits within one, seven, and fourteen days. Do not select the most favorable window after the fact without labeling it as exploratory.
4. Calculate lift and uncertainty
Count qualifying visits and audience records for both groups. Calculate each visit rate, subtract the control rate from the exposed rate, and multiply the difference by the exposed audience size. Report an uncertainty measure, such as a confidence interval, when your analytics provider or statistician can calculate one appropriately.
A small difference can be commercially useful, but it should not be treated as conclusive when the groups are small or the interval is wide. Sample-size planning should account for the expected baseline visit rate, the smallest lift worth detecting, the desired confidence level, and the allocation between exposed and control groups. When expected visits are rare, a larger eligible audience or longer preplanned observation period may be necessary.
5. Test for bias before interpreting the result
- Pre-period balance: Compare historical visit rates for exposed and control groups before the campaign. Large differences may indicate poor matching.
- Geographic access: Check whether both groups have similar distances, store availability, and coverage across trade areas.
- Audience composition: Review device, customer, demographic, or behavioral distributions where lawful and available in aggregated form.
- Contamination: Determine whether control users received the campaign through another device, channel, publisher, or household member.
- Store and timing changes: Flag closures, openings, promotions, weather disruptions, holidays, and inventory issues that affected one group or period disproportionately.
- Attribution consistency: Confirm that exposure logs and visit measurement use compatible time zones, identifiers, retention rules, and deduplication logic.
Location analytics should be aggregated and privacy-conscious. Use consented data and provider thresholds where applicable, suppress small cells, and avoid attempting to identify individual visitors. The principles in Privacy-First Digital Identity for Marketing provide useful context for consent and first-party measurement design.
Inputs and assumptions
Build the study around a small input sheet that can be reused for every campaign. Record both the value and its source or owner.
- Campaign cost: Include media spend and any measurement fee you want reflected in efficiency calculations. Keep production or fixed platform costs separate if they are not attributable to this test.
- Exposed audience: Use the deduplicated number of eligible, measurable records, not impressions.
- Control audience: Record its size and assignment method. Note whether the control was randomized, matched, or geographically selected.
- Exposed visits and control visits: Apply the same store boundary, dwell rule, visit window, and data-quality filters.
- Average revenue per visit: Use a conservative, clearly defined estimate. Ideally separate gross sales, net sales, margin, and repeat purchases rather than treating them as interchangeable.
- Confidence or uncertainty measure: Record the method supplied by the measurement provider or analyst.
- Exclusions: Document stores, dates, audiences, or channels removed from the analysis and why.
Do not assume that every attributed visit is incremental. Do not use click-through rate as a substitute for store visit lift, and do not compare campaigns with different control designs as though their results were directly equivalent. For campaign setup and geography choices, review Geofencing vs. Geotargeting vs. Geo-Conquesting.
Worked example
Assume a retailer runs a location-based advertising campaign across selected stores. These figures are illustrative assumptions, not a benchmark:
- Campaign cost: $12,000
- Exposed audience: 100,000 measurable records
- Control audience: 100,000 comparable records
- Exposed visits during the defined window: 4,200
- Control visits during the same window: 3,600
- Estimated revenue per incremental visit: $38
The exposed visit rate is 4.2% and the control visit rate is 3.6%. The estimated incremental visit rate is therefore 0.6 percentage points, or 0.006. Applied to the exposed audience, the campaign generated an estimated 600 incremental visits.
Relative lift is calculated as 0.6% ÷ 3.6%, or approximately 16.7%. Cost per incremental visit is $12,000 ÷ 600, which equals $20. Estimated incremental revenue is 600 × $38, or $22,800.
This result is not automatically profitable. Compare incremental revenue with media cost, product margin, fulfillment costs, discounts, and any measurement fees. Also review uncertainty: if the exposed and control rates are statistically indistinguishable, describe the outcome as an estimated positive difference rather than a confirmed causal effect.
Reusable reporting template
Use the following structure in a campaign readout:
- Objective: What decision should the study inform?
- Population and geography: Which audiences, stores, and trade areas were eligible?
- Design: How were exposed and control groups created?
- Visit definition and window: What counted as a visit, and when?
- Results: Exposed rate, control rate, absolute lift, relative lift, incremental visits, and uncertainty.
- Efficiency: Campaign cost, cost per incremental visit, and incremental revenue or margin.
- Bias checks: Pre-period balance, contamination, geography, timing, and data-quality findings.
- Decision: Scale, modify, repeat with a stronger design, or stop pending more evidence.
When to recalculate and update the study
Recalculate after every material campaign cycle, especially when budget, audience, store footprint, bid strategy, creative, or visit window changes. A result from a dense urban trade area should not automatically be used to forecast a rural or low-frequency market. Recheck assumptions when average order value, margin, store hours, promotions, or inventory conditions change.
Review the methodology whenever a measurement vendor changes its identity resolution, location signals, store-visit model, privacy thresholds, attribution window, or control-group process. Platform reporting capabilities and available privacy-safe aggregation can also change. Keep a dated methodology note with each result so future studies remain comparable.
Before the next campaign, complete three actions: lock the primary outcome and visit window, approve the control design, and create the input sheet with owners for cost, audience, visit, revenue, and uncertainty data. After the campaign, publish the exposed-versus-control table, run the bias checks, and make the budget decision from incremental outcomes rather than attributed visits alone. For a broader return-on-investment model, use the Proximity Marketing ROI Calculator alongside this incrementality framework.