Market basket analysis consists of using data mining techniques to analyze customer shopping data to find patterns and relationships among purchased products. This information may help a retailer design onsite or e-commerce shopping spaces. It may also be used to develop marketing campaigns. The evaluation can be expensive, so it may be used only for popular items.
Both affinity analysis and cluster analysis, may be used when evaluating market baskets. An affinity analysis tool may be used to find activities, such as purchases, that happen at the same time. Cluster analysis is a type of statistical technique that arranges raw data into categories, and it is often used for more complex analysis. When evaluating market baskets, sales of an item or group of items will be examined in relationship to sales of one or more other items or groups.
Other concepts in market basket analysis include frequency, support, and confidence. Frequency refers to the number of times customers buy product A and B at the same time. Support is the number of times products A and B were bought together compared to the total number of sales. The term confidence refers to the number of times product A and B were bought together, compared to the number of times only product A was bought.
For example, a swimwear retailer may use market basket analysis to compare the sale of swimsuits to suntan lotion. The retailer finds that many of its customers purchase swimsuits, and that these buyers also often purchase suntan lotion. He may use the results of this analysis to organize his onsite or e-commerce store in a way that increases the chances that customers will buy these related products. In the onsite store, he may place suntan lotion displays next to the swimsuits. Suntan lotion may be listed as a related item on in an e-commerce shop.
Although market basket analysis is frequently used for items bought at the same time, it can also be applied to purchases made over a series of time. The retailer may target advertising to take advantage of a customer’s likelihood to purchase products in a particular order. For example, a furniture store finds that customers who purchase cribs, then purchase twin beds approximately two years later. The furniture store may obtain customer contact information, and email promotions for twin beds about 18 months after the crib purchase.
Market basket analysis can become expensive and difficult to implement when retailers have a large number of goods to analyze. To be most cost-effective, retailers often focus their efforts on those items that have both high support and high confidence. These items are likely to generate enough profit to make the cost of the analysis worthwhile.