Methoden vergleichen

	Price Elasticity from Scanner Data	Scanner Panel Analysis
Fachgebiet	Marketing	Marketing
Familie	Regression model	Regression model
Entstehungsjahr	1988	1983
Urheber	Dick R. Wittink, Peter S. H. Leeflang and colleagues (SCAN*PRO)	Peter M. Guadagni & John D. C. Little
Typ	Multiplicative store-level sales-response regression	Disaggregate multinomial-logit brand-choice model
Wegweisende Quelle	Leeflang, P. S. H., Wittink, D. R., Wedel, M., & Naert, P. A. (2000). Building Models for Marketing Decisions. Kluwer Academic Publishers. ISBN: 9780792377726	Guadagni, P. M., & Little, J. D. C. (1983). A Logit Model of Brand Choice Calibrated on Scanner Data. Marketing Science, 2(3), 203-238.
Aliasnamen	SCAN*PRO Model, Store-Level Sales Response Model, Multiplicative Sales Response Model, Promotion Sales-Effect Model	Scanner Panel Logit, Guadagni-Little Model, Household Panel Choice Model, Loyalty-Variable Logit
Verwandt	3	3
Zusammenfassung	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.	Scanner panel analysis models individual households' brand choices using the purchase histories captured by UPC scanner panels, in which the same households are tracked occasion by occasion with the brand chosen and the prices and promotions they faced. The defining method is Guadagni and Little's 1983 multinomial logit of brand choice, the first model to put scanner panel data to serious analytical use. Its signal innovation is the loyalty variable: an exponentially smoothed measure of each household's past brand purchases that enters the utility function and captures persistent brand preference and state dependence. Alongside loyalty, the model includes price, promotion, and brand intercepts, and yields the probability that a household buys each brand on a given occasion. From the fitted model one recovers price and promotion elasticities at the individual level and can simulate how marketing actions shift choices. The framework launched the modern era of disaggregate choice modeling and remains the reference point for scanner-based brand-choice analysis.