Did you choose the deal, or did the algorithm choose you?

Imagine you are standing in the checkout aisle at the Shaw's on Admiral Street near Providence College. You were not planning to buy Powerade. Then you see the familiar 28-ounce bottles: normally $1.29, but today the sign says '2 for $2.' You pause. Two bottles suddenly seem like a better deal than one. You put them in your cart. That small moment is nothing new. The '2 for $2' sign is a traditional promotion: everyone who sees the display gets the same offer. But another kind of promotion works differently – and this is where artificial intelligence and machine learning enter the picture. Before you check out, you may enter your phone number. That simple action turns your purchase into data. Over time, a retailer can learn what products you buy, how frequently you buy them, and how you respond to different promotions. At Shaw's, customers may see storewide weekly deals as well as personalized 'For U' offers through the app. Unlike the '2 for $2' sign that everyone sees, a 'For U' offer is presented to an individual customer. This makes personalized offers particularly interesting from a machine-learning perspective. At first glance, this looks simple: the store gives you a personalized coupon, and you save money. Suppose you regularly buy energy drinks. Your purchase history may indicate that you are interested in that category. A personalized coupon might then be offered to you. If you use the coupon and purchase more, that behavior provides another piece of information. Machine-learning models can analyze patterns like these to help retailers understand which promotions are more likely to influence different customers. The store is learning about you. Once you realize that, you can learn from the store, too. But there is another way to look at it. When you receive a personalized coupon, you might ask: Why did I receive this particular offer? Perhaps the retailer has noticed that you buy this product frequently. Perhaps it believes a discount will persuade you to buy more. Either way, the coupon gives you a clue about what the retailer may have learned from your shopping history. Now the relationship becomes more interesting. The retailer knows that it is learning from your purchases. You know that it is learning from your purchases. And you know that the retailer may use that information to decide which offers to show you. This does not mean consumers need to 'beat' the algorithm. Instead, we can become more thoughtful participants in the process. When a personalized coupon appears, ask yourself: Was I already planning to buy this, or did the coupon create the desire to buy it? If you were going to buy the product anyway, the coupon may simply be a useful discount. But if you buy something only because the app suggested it, the promotion may have changed your behavior. In this sense, a personalized coupon is more than a discount. It is also a signal. It gives the retailer information about you, while giving you a glimpse into what the retailer may have learned about you. This creates a two-way learning process: the retailer learns from consumers, while consumers can observe personalized offers and learn something about how the retailer sees them. AI is becoming increasingly good at learning our shopping habits. We do not have to become machine-learning experts to respond intelligently. We simply need to recognize that the algorithm is learning from us – and that, once we understand that, we can learn from the algorithm, too. The next time you open your Shaw's app and see a personalized 'For U' offer, ask yourself: Did you choose the deal, or did the algorithm choose you? Wen Chen is a professor of Operations Management at Providence College.