Separating Brand-Loyal From Brand-Indifferent Shoppers With Behavioral Metrics

Assume you work on a large online store and want to know, for a given customer, whether they are brand agnostic or brand conscious when they shop. A brand-agnostic shopper chooses mainly on the product itself, its price and convenience, while a brand-conscious shopper gives real weight to which brand makes the item. How would you tell the two groups apart? Describe the use cases where the distinction would matter and the metrics you would use. ```hint Look for costly choices What shoppers say about brands matters less than what they give up for one. Find the moments in search and purchase data where picking a particular brand costs the shopper something. ``` ### Clarifying Questions - Is brand consciousness a trait of the shopper across the whole store, or should it be judged per category or even per brand, since someone can insist on one phone maker yet buy unbranded batteries? - Should the output be a label per customer, a score, or an overall share of shoppers, and which team will act on it (search ranking, store-brand strategy, advertising sales)? - Can we use only on-site behavior and purchase history, or can we also run surveys and controlled experiments? - How much history is needed before we classify someone, and what happens with new customers? ### What a Strong Answer Covers - A precise working definition that separates loyalty to one brand, preference among a few known brands, and indifference, scoped by category - Concrete metrics drawn from search, browsing, filtering and purchase behavior - Controls for confounders such as price, availability, delivery speed and categories dominated by a single brand - A way to validate the classification, for example with experiments or surveys - Use cases that connect each segment to a business decision ### Follow-up Questions - A customer always buys the same detergent brand, and it is also the cheapest option in stock. How do you tell loyalty from price-seeking? - How would you use these segments in search ranking without harming shoppers or brands that pay for sponsored placement? - How would you check that your classification is right if you could never ask a customer directly?