Estimating cross-brand promotional effects with aggregate data requires knowledge of the joint distribution of each brand’s promotions. While such information is available in store-level scanner data, it is not included in more aggregated scanner datasets. This paper presents a technique for overcoming this difficulty and develops a retailer-level model that incorporates both own- and cross-brand promotions. Promotional activity is integrated into the specification in a manner consistent with the way store-level models control for promotions, thereby avoiding the problem of aggregation bias. The proposed methodology extends the usefulness of retailer-level scanner data by allowing it to answer important questions regarding how the promotions of competing products interact.