Price Elasticity from Scanner Data
Also known as: SCAN*PRO Model, Store-Level Sales Response Model, Multiplicative Sales Response Model, Promotion Sales-Effect Model
Estimating price elasticity from scanner data means fitting a store-level sales-response model to the weekly unit-sales, price, and promotion records that retail checkout scanners generate, in order to recover how sensitive demand is to price. The canonical specification is the SCAN*PRO model developed by Dick Wittink, Peter Leeflang, and colleagues: a multiplicative model in which a brand's unit sales in a store-week are a product of relative-price terms raised to elasticity powers and promotion multipliers for feature and display. Taking logarithms turns this into a linear regression whose price coefficients are directly interpretable as own- and cross-price elasticities, while the promotion coefficients become multiplicative lift factors. Pooled across many stores with store-specific intercepts, the model delivers stable, managerially usable elasticities and quantifies the sales lift from promotions. Later work, such as Van Heerde, Gupta, and Wittink, decomposed the promotional sales bump into brand switching, purchase acceleration, and category expansion, refining the interpretation of what an elasticity captures. It is the standard aggregate demand model in retail analytics.
Key highlights
- Price coefficients are directly interpretable as constant own- and cross-price elasticities, requiring no further transformation.
- The multiplicative form matches how demand actually responds to price (proportionally) and keeps predicted sales positive.
- Using relative rather than absolute price normalizes cross-store price-level differences and isolates response to temporary changes.
- Pooling across many stores with store intercepts yields stable, precisely estimated elasticities suitable for decision support.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use store-level scanner price-elasticity modeling when you have store-by-week unit-sales, price, and promotion data and need own- and cross-price elasticities and promotion lift factors for pricing and trade-promotion decisions. It is the right tool for frequently purchased, scanner-tracked retail categories where temporary price cuts, features, and displays drive substantial sales variation, and where managers want a parsimonious, interpretable aggregate demand model rather than individual-level choice. It suits marketing-mix analysis, promotion evaluation, and what-if pricing simulation. It is less appropriate when the research question is individual brand choice or loyalty dynamics (use a scanner-panel logit), when price is set in a way that creates serious endogeneity you cannot control for, when the category exhibits large stockpiling that makes weekly sales a poor demand proxy without dynamic correction, or when you need to attribute the promotion bump to its sources without the additional decomposition step.
Strengths & limitations
- Price coefficients are directly interpretable as constant own- and cross-price elasticities, requiring no further transformation.
- The multiplicative form matches how demand actually responds to price (proportionally) and keeps predicted sales positive.
- Using relative rather than absolute price normalizes cross-store price-level differences and isolates response to temporary changes.
- Pooling across many stores with store intercepts yields stable, precisely estimated elasticities suitable for decision support.
- Aggregate store sales mask individual heterogeneity, so the model cannot separate who responds, only the average store-level effect.
- Managerially set prices can be endogenous, biasing elasticity estimates unless instruments or controls are used.
- Weekly sales conflate genuine demand with stockpiling and pull-forward, so a raw elasticity may overstate incremental effect.
- The measured promotion bump mixes switching, acceleration, and category expansion and needs decomposition to interpret correctly.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
Why is SCAN*PRO multiplicative and estimated in logs?
Demand responds to price in proportional terms — a ten-percent cut tends to change sales by a roughly constant percentage rather than a constant number of units — so a multiplicative model where sales equal a base times price and promotion factors captures the response better than an additive one, and it guarantees positive predicted sales. Taking logarithms converts that product of factors into a sum, giving an ordinary linear regression. The convenient payoff is that the coefficient on log price is exactly the price elasticity, so the model is both behaviorally sensible and directly interpretable without further calculation.
Why use relative price instead of absolute price?
Stores differ persistently in their price levels for reasons unrelated to short-run demand response — local competition, format, cost structure. If you regress sales on absolute price across stores, those persistent differences contaminate the estimate and you no longer measure how sales respond to a price change within a store. SCAN*PRO uses price relative to the brand's own regular or mean price, which normalizes away the cross-store level differences and isolates the response to temporary deals. Combined with store-specific intercepts, this is what lets the pooled elasticity reflect genuine price response rather than store heterogeneity.
Is the whole promotion sales bump incremental profit?
No, and this is a crucial caveat. The measured bump combines three sources: brand switching (sales captured from competitors), purchase acceleration or stockpiling (the same buyers simply buying earlier, which borrows from future weeks), and category expansion (truly new category demand). Van Heerde, Gupta, and Wittink showed that the switching share is much smaller than the long-quoted three-quarters, and that acceleration is substantial. Because stockpiled and pulled-forward sales are not net new business, a promotion that produces a big bump can still be unprofitable. Decomposing the bump is therefore essential before treating an elasticity or lift as a measure of incremental return.
Sources
- 1.Leeflang, P. S. H., Wittink, D. R., Wedel, M., & Naert, P. A. (2000). Building Models for Marketing Decisions. Kluwer Academic Publishers.ISBN 9780792377726
- 2.Van Heerde, H. J., Gupta, S., & Wittink, D. R. (2003). Is 75% of the Sales Promotion Bump Due to Brand Switching? No, Only 33% Is. Journal of Marketing Research, 40(4), 481-491.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 23). Price Elasticity from Scanner Data. ScholarGate. https://scholargate.app/marketing/price-elasticity-scanner-data