A shopper at the self-checkout scans a phone number to knock a few dollars off a grocery bill, barely registering the transaction as a transaction. It feels like a coupon. It is closer to a contract.
In exchange for that discount, the shopper has just attached a name, and a purchase history, to a stream of data that used to be anonymous.
That distinction is the whole business.
Retailers have always known what sold, in aggregate, on any given day. What they did not know, until loyalty programs became standard, was who bought it, how often, in what combination, and how a price change or a coupon shifted that one person's next visit. A loyalty number turns a river of anonymous receipts into a filing cabinet of individual customer files, each one updated in real time.
Explaining the mechanics takes one sentence: the retailer trades a small, visible discount for a large, invisible asset. The discount shows up on the receipt. The asset shows up on a balance sheet, if it shows up anywhere at all, as a cluster of things executives now call "customer insights" or, more candidly, a retail media business.
That business has a specific customer of its own, and it usually is not the shopper. It is the consumer-goods company that wants to know whether a shelf-tag price cut actually moved a particular household from one detergent brand to another. Loyalty data answers that question in a way store-wide sales figures cannot, because it follows the same person across visits. Retailers package that answer and sell it, along with targeted digital ad space inside their own apps and websites, back to the brands stocking their shelves.
Kroger built a subsidiary out of exactly this. Amazon folded it into the logic of Prime from the start. Airlines figured out the adjacent version decades ago, when they noticed that selling miles to a co-branded credit card issuer was more reliably profitable than selling seats. In each case the loyalty program stopped being a marketing expense and became a second business riding inside the first one, often a more profitable one.
There is a second, quieter revenue stream built into the same programs: points that never get redeemed. Every rewards scheme issues more points than customers cash in, because thresholds are set high, expiration clocks run, and ordinary people forget balances exist. Accountants call the unredeemed portion breakage, and it gets recognized as revenue rather than sitting as a liability forever.
A program engineered to be slightly annoying to redeem is not a design flaw. It is a feature with a number attached to it.
Who pays for all this is less visible than who profits, which is rather the point of the arrangement. Shoppers pay in the currency of exposure: a purchase history that can be sold, subpoenaed, breached, or used to justify a personalized price that a stranger buying the identical item next to them does not see. Non-members pay too, in a subtler way, since everyday shelf prices at many chains have crept upward as the discount now flows through the loyalty app instead of the sticker.
Card-issuing banks profit as well, collecting interchange fees on every swipe of a co-branded card while paying the retailer a cut for the privilege of owning that relationship. The retailer gets a financial-services income stream with none of the merchandising risk. None of this requires the loyalty program to make the shopper's life better, only to make it legible.
Consistent incentives fall out of this arrangement, and they point away from what a rewards program sounds like it should do. A retailer optimizing for data and ad revenue wants shoppers enrolled and tracked more than it wants them to redeem generously; it wants purchases concentrated under one identity more than it wants prices genuinely low; it wants the app open more than it wants the discount simple. None of that is hidden exactly, just rarely the pitch printed on the sign-up screen.
Two things could push this arrangement in a different direction, and neither looks close. One is regulation treating loyalty data with the same seriousness applied to health or financial records, which would raise the cost of building these programs enough to change how aggressively they collect. The other is genuine competition from reward mechanisms that do not require identity at all, such as flat cash-back cards or anonymized discount networks, which would force retailers to compete on price rather than on the value of the dossier a purchase history builds.
Neither pressure is strong today.
Privacy law in the United States remains a patchwork, and the retail media business has grown large enough that some chains now report the ad and data arm as more profitable, per dollar, than the stores generating the traffic. That ordering, revenue from watching shoppers exceeding revenue from selling them things, is fairly new, and it is not obviously stable.
What a store becomes once the thing that keeps it profitable is no longer what it sells, but what it knows about who's buying, is the open question now.
Retailers have not had to answer that question in public, because the loyalty card still looks, to the person scanning it, like nothing more than a coupon.