Understanding the effectiveness of a sales promotion.
Type of Report
Quantitative modeling of consumers' decisions of brand choice, purchase time, and purchase quantity in a promotional environment; empirical analysis using scanner data.
To determine whether the sales increase occurring during a promotion period is "incremental" or "borrowed" by decomposing the sales "bump" during the promotion period into three components: sales increase due to (1) brand switching, (2) purchase acceleration, and (3) stockpiling.
The impact of sales promotions on consumer decisions of when, what, and how much to buy is modeled. These consumer decisions in turn determine the overall sales of a brand. An elasticity analysis is then performed to decompose the increase in overall sales of a brand into the three components mentioned above. The models are estimated and validated using scanner data for ground caffeinated coffee.
Although technical details of this report will appeal to quantitative modelers, the overall approach and the substantive results should be of interest to marketing managers.
A sales promotion can affect at least three different decisions: (1) when consumers buy, (2) what brand is bought, and (3) how much is bought. The combined effect of these three decisions, summed across consumers, gives us the total sales of a brand in a week. But simply looking at the total sales increase during a promotion period does not tell us whether the promotion has generated incremental sales or has simply borrowed future sales.
To address this question, we need to decompose the total sales increase during a promotion period into, say, x percent increase due to switching, y percent due to acceleration, and z percent due to stockpiling. For example, consider two hypothetical cases:
Case 1: x=80%, y=15%, z=5%
Case2: x=5%, y=10%, z=85%
In case 1, the majority of sales increase is due to brand switching, whereas in case 2 most of the sales increase can be attributed to stockpiling. It is then easy to say that the promotion is more effective in the first case.
However, in order to make such conclusions, we need a method which can empirically derive the values of x, y, and z. This paper presents such a method. Conceptually the method is simple. We first develop models of purchase time (when to buy), brand choice (what to buy), and purchase quantity (how much to buy). These models explicitly incorporate competition, and differences in consumer purchasing patterns (e.g., the probability of choosing a brand decreases if the competitive brands are on promotion). For any given week, these models give us the probability of buying the product, probability of choosing a brand given that a purchase is made, and the expected purchase quantity of the product category.
We then represent the total sales of a brand as a product of these three components. This gives us the total elasticity of brand sales as the sum of the elasticities of choice, timing, and quantity components. Since elasticity tells us what percent increase is expected (in total brand sales, choice, timing, or quantity) due to a 1 percent increase in promotion, the above formulation allows us to decompose the sales bump into the percentage increase due to switching, time acceleration, and stockpiling. In other words, it gives us the value of x, y, and z. The models are estimated on scanner data for ground caffeinated coffee.
Results and Implications
The results show that promotional variables (feature, display, price cut) play a strong role in consumer brand choice decisions. The effect of these variables is limited on purchase time and purchase quantity decisions. Specifically, of the total sales increase due to promotion, more than 84 percent is accounted for by brand switching, 14 percent or less by purchase time acceleration, and less than 2 percent by stockpiling. This indicates that promotions for this product are very effective in drawing consumers from competitive brands. On the other hand, promotions have a limited success in making consumers buy early. Although feature and display have some impact on the purchase time decision, price cut and regular price have almost none. This suggests that if consumers are not planning to buy coffee in a given week, they may not go out of their way to check prices or price discounts on coffee brands unless their attention is attracted to them through features or displays. The small effect of promotion on stockpiling can be attributed to three possible phenomena: consumer perceptions that stockpiling coffee may affect its freshness, storage constraints, and high promotion intensity in the marketplace. This suggests that stockpiling will perhaps be a bigger factor for other product categories such as tuna fish (smaller cans) or soap (freshness not important).
These results can help managers understand the effectiveness of a sales promotion. The effectiveness of alternative promotional offerings under various competitive scenarios can also be compared to determine the most suitable and effective promotions.
Professor Gupta comments:
"This paper presents a comprehensive model which captures three main components of consumers buying decision: when, what, and how much to buy. The model incorporates price, and promotions of the competitive brands as well. Empirical illustration is provided using scanner data, which suggests the relevance and applicability of the model to real life problems.
"This study takes an important step toward a better understanding of promotions effectiveness. However, more work is needed to include (1) the lead effect of promotions—consumers can anticipate future promotions, (2) long term impact of promotions—frequent dealing may damage the brand image, (3) asymmetric affects of promotions—some brands may gain more from promotions than others, and (4) threshold levels of promotions—a minimum discount, say 50 cents, may be needed before consumers switch brands.
"Good managerial insights, sophisticated research skills, and detailed disaggregate data (e.g., scanner data), should make it possible for us to better understand sales promotion for what it is—a very important marketing tool."
Related Titles from MSI
Effective Sales Promotion Lessons for Today: A Review of Twenty Years of Marketing Science Institute-Sponsored Research by Dudley M. Ruch, Monograph, 1987, 47 pp., 87-108.
Measuring and Evaluating Sales Promotions to the Trade and to Consumers, summary of an MSI conference prepared by John U. Farley, 1985, 25 pp., 85-113.
Triggers to Customer Action—Some Elements in a Theory of Promotional Inducement by Eugene R. Beem and H. Jay Shaffer, Working Paper, 1981, 70 pp., 81-106.
Research on Sales Promotion: Collected Papers edited by Katherine E. Jocz, Monograph, 1984, 123 pp., 84-104.
Factors Influencing Grocery Retailers' Support of Trade Promotions by Ronald C. Curhan and Robert I. Kopp, Working Paper, 1986, 26 pp., 86-104.
Trade Promotion by Grocery Products Manufacturers: A Managerial Perspective by John A. Quelch, Working Paper, 1982,43 pp., 82-106.
The Impact of Price Promotions on a Brand's Market Share, Sales Pattern, and Profitability by Leigh McAlister, Working Paper, 1986, 33 pp., 86-110.
About the Author
Sunil Gupta is Assistant Professor of Marketing at the Anderson Graduate School of Management, University of California, Los Angeles.
Comments from membersPlease login to view and/or submit comments
|Please login for full access and discounted pricing.|