The McDonald’s Loyalty Program That Turned the App Into the Price

McDonald’s has nearly 220 million active loyalty members, and that number tells you almost nothing about whether anyone is loyal.

On August 4, McDonald’s reported that sales to loyalty members across its 70 loyalty markets rose more than 20% to $40 billion over the trailing twelve months. Ninety-day active loyalty users rose 13% to nearly 220 million. The company is targeting 250 million active users by the end of 2027.

The less celebrated line was about volume. Global comparable sales rose just 1.3%, and comparable guest counts were negative in the United States, in International Operated Markets, and in International Developmental Licensed Markets. The company is selling more to fewer customers, at higher prices.

Members spend more and visit more. On the fourth-quarter 2025 earnings call, as quoted by a financial analysis site, CFO Ian Borden told investors a customer visited 10.5 times in a year. A loyalty member visited 26. Member spending over the trailing twelve months is $40 billion. That is one quarter of the company’s revenue in its 70 loyalty markets, and the company says it is growing.

Then came October, and the clearest illustration of what the program has become. Monopoly returned to the United States on October 6, this time as an app game built into MyMcDonald’s Rewards. Customers collect Chance Cards by ordering, scan them, and play digitally, with a $1 million prize among the rewards. The promotion runs through November 1, and its best-known feature requires the app.

McDonald’s isn’t only rewarding its customers’ loyalty. It is making the app the way you get the price.

UK retailers announced similar moves in the same week. Tesco expanded Clubcard Points Partners from 13 outside partners to more than 500, including Nike, Boden, and Ticketmaster. Morrisons upgraded its More Card app with six personalized weekly offers and a shopping list that lets customers scan a product barcode at home. JD Sports, a week earlier, let JD Status members earn JD Cash with more than 40 third-party brands, and extended benefits to more than half a million JD Gyms members.

McDonald’s own UK program moved the same way. Points are earned and redeemed through the app. Tesco shortened how long points last, with points earned before June 30, 2025 expiring on January 1, 2026. It added partner rewards in the app for the first time and changed redemption thresholds from March 17, 2026. JD gave members another way to earn and spend JD Cash, with a new way to lose it: JD Cash that sits unused for 180 days is forfeited.

Look at what each company reports. Tesco has published no figures for incremental sales, enrolments, or retention from its expansion, according to trade coverage. JD cites four million users and more than £120 million in JD Cash distributed. Morrisons cites eight million active users and record redemption rates. McDonald’s reports members and member sales. Every one of those numbers measures reach or cost. None shows what the program adds.

The first problem is what McDonald’s has chosen to count.

Goodhart’s Law: Once a measure becomes a target, it stops being a reliable indicator of the thing it was meant to track, because people and systems adapt to hit the number.

McDonald’s counts sales to customers who identify themselves as members when they pay. It has set a target for active users. And it has designed promotions, like this year’s Monopoly bonus play, that reward customers for identifying themselves. Each choice is reasonable. Together, they mean the reported number can rise because McDonald’s made identification more rewarding, whether or not any customer visits more often.

The second problem is the headline evidence.

Self-selection bias: When people choose to join a group or program, differences between the group and everyone else may reflect who chose to join and why, not what the program did.

The 10.5-to-26 comparison is a claim about a group of people who chose to give McDonald’s their data, their location, and their attention. No claim follows from it about what the program does to those who receive it, because no counterfactual was measured.

The third problem is what the app has become.

Price discrimination by self-screening: Offering a lower price only to customers who bear a cost for it, whether time, effort, or personal data, sorts buyers by price sensitivity. Price-insensitive customers skip the effort and pay full price, and price-sensitive customers do the work and pay less.

In this light, the app isn’t a reward layered on top of the price. It is part of the price, and personal data is part of the payment. McDonald’s will be able to use that data to sell advertising or target the customer later. Once a price is framed as a loyalty reward rather than a fee, it cannot be presented as a discount in either form.

Which brings us back to the claim that started this. A LinkedIn post this week argued that the Tesco, Morrisons, and JD announcements show customer loyalty doesn’t exist, because truly loyal customers wouldn’t need offers. That inference doesn’t follow, and the mistake runs in both directions.

Behavioral versus attitudinal loyalty: Repeat purchasing (behavior) is not the same as genuine preference for a brand (attitude). A customer can buy repeatedly out of habit, convenience, or price with no attachment, and a customer can feel attached yet buy elsewhere when it’s cheaper or closer. True loyalty requires both (Dick and Basu, 1994).

McDonald’s metrics capture the behavioral side, and only for identified customers. The “loyalty is a myth” argument treats the need for offers as proof that the attitude is absent. Both treat behavior as the whole of loyalty. Offers can reward, retain, or monetize attachment as well as buy it, and the number that would tell you which is happening is what customers would choose at the same price. That is the number none of these announcements report.

Monopoly used to be something you peeled and played at the counter. This year the whole game runs through an app that requires registration. For customers who value the game, that is a small trade. For customers who don’t want to hand over their data, the message is clear: the full price is what you pay for staying anonymous.

The UK expiry rule sends a related signal. Making points vanish faster turns a balance into something customers can lose, which usually speeds redemption. It also tells them the balance is a promotion with a clock on it, not a stake in a relationship. When a program’s own design treats rewards as expiring offers, customers learn to treat the program the same way.

The signal reaches investors too. When the company’s best-known evidence is a before-and-after comparison of people who chose to join, it tells analysts reach is the story, because reach is what’s being measured.

These are the things any organization with a loyalty program can change, whatever its size.

• Measure incrementality, not reach. Randomly withhold offers from a matched group of members and compare their behavior over time. That is the only way to estimate what the program adds beyond who joined.

• Separate migration from lift. Track how much of “member sales” was already happening before customers identified themselves. A metric that rises when you gate promotions behind identification can grow without a single extra visit.

• Report guest counts and retention for members versus matched non-members. Sales to members is a revenue measure. Visits, retention, and share of wallet are behavior.

• Test usefulness separately from discounts. Morrisons’ barcode shopping list is a feature that helps customers without costing margin. Irrelevant offers are the top loyalty frustration in one UK survey of 2,000 shoppers, which suggests relevance is worth testing directly.

• Measure the attitude as well as the behavior. Ask customers whether they prefer the brand or just the price, and track whether they keep buying when the offer is withdrawn.

• State the bargain plainly. If the app is the price, say so, and the customer can decide whether the trade is worth it.

• Public data can’t settle incrementality. McDonald’s may run internal holdout tests that show real lift. I’m working from public figures, and the 10.5-to-26 number is as quoted second-hand from the earnings call.

• Gating can be a fair exchange. There is nothing wrong with a customer paying for a lower price with time, effort, or data, as long as they know it.

• Falling US guest counts have many causes. They are also the thing the program is meant to reverse, and they make the case for the program harder to evaluate.

• Not everything has to be loyalty. A program can be a channel, a discount, or a product, and a company can run it well without calling it loyalty.

• The survey evidence on declining loyalty is weak. The research cited is often about consumer preference, not about loyalty, and the evidence is mixed.

• The UK announcements are too new to evaluate. They are announcements, not results, and the companies have not published outcomes.

If your loyalty program disappeared tomorrow, which of your members would still choose you at the same price, and how would you know?