McDonald’s Keeps a Dossier on Every Customer

A WIRED journalist requested a copy of his personal data from McDonald’s and received a document that turned out to be far more extensive than a standard order history. In addition to purchase information, the company compiled a behavioural profile of the customer and used algorithms to predict when he would return, how much he would spend and what he would most likely order next.

According to the 515-page report received by the journalist, McDonald’s predicted 2.16 visits over the following six weeks, an average order value of around $13.49 and total spending of approximately $29.15. The system also identified the most likely next favourite item — a large Diet Coke. The probability that the customer would leave the chain, by contrast, was assigned a value of 0. In other words, the algorithm hardly expected this person to stop visiting McDonald’s.

What is particularly revealing is that the system analysed more than purchases alone. The customer was classified into behavioural categories such as «afternoon snack» and «rushed lunch on the go». In a separate ranking system, McDonald’s compiled a list of products most likely to appeal specifically to this visitor: alongside a large Diet Coke, it included a Snack Wrap, fries, a cheeseburger and McNuggets.

This isn’t «tracking every burger» but a much more interesting technology

From a marketing perspective, this is a classic shift from analysing past behaviour to predicting the future. The company knows what a person has bought before, how often they visited, which offers they opened or used, where they made purchases and how they interacted with the app. This data is then transformed into a probabilistic model: what the customer might want next and how likely they are to return.

That is why a loyalty programme today is not simply a «collect points → get a free burger» mechanism. In practice, it is a personalisation tool that allows the company to better understand the habits of a particular customer.

McDonald’s itself explicitly states in its current privacy policy that it collects data about MyMcDonald’s Rewards members, including purchase history, actions in the app, interactions with digital technologies in restaurants and, when device settings allow it, location data. The company also says it may use this data to create profiles and draw inferences about a person’s preferences, characteristics and behaviour, as well as use the information to personalise offers and train algorithms and AI models.

The most important thing here isn’t the 515 pages

The length of the report looks alarming, but the figure of 515 pages does not mean that McDonald’s wrote 515 pages of text about one particular person. The document contains a large collection of records: transactions, interactions with the app, rewards, offers and other data.

What is much more interesting is that the company does not merely store facts but draws conclusions from them.

A single purchase of a large cola says practically nothing about a person. But if it is combined with dozens of previous orders, visit times, purchase frequency, average order value, responses to promotions and the geography of visits, the algorithm can already build a sufficiently accurate model of behaviour.

This is precisely the point that privacy specialists consider most important. According to experts, the problem is not necessarily each individual piece of information, but the fact that combining them makes it possible to draw far more detailed conclusions about a person.

Why this benefits businesses

For McDonald’s, such a system has obvious commercial value. If the algorithm understands that a particular customer usually comes in after lunch and almost always orders a certain drink, the company can show that person a relevant promotion at exactly the right time.

Instead of a generic message saying «Try our new product», the customer may receive an offer tailored specifically to their habits.

This increases the likelihood of a purchase while also helping the company retain the customer. That is why the «churn probability» figure in the journalist’s report is particularly interesting: the business is trying to assess not only what a person buys, but also how likely they are to stop buying altogether.

Customers have rights too

McDonald’s says it provides users with ways to manage their personal data. In the United States, Rewards programme members can opt out, and when a request is made to delete personal information, the points and rewards associated with the account also become unavailable.

The WIRED story illustrates a rather striking paradox of the digital economy: the more convenient the app and loyalty programme, the more data the company obtains about your behaviour.

The user sees discounts, bonus points and the ability to place an order quickly. The company sees a sequence of actions, purchase frequency, preferences and the likelihood of the next purchase.

And while a person is thinking about which burger to order today, the algorithm may already be calculating what they are most likely to order several weeks from now.

In the journalist’s case, the prediction was particularly amusing: the system was virtually certain that he would not go anywhere. To test the algorithm, after receiving the report he requested the deletion of his data and decided to stop visiting McDonald’s — solely to prove to the system that it could, after all, be wrong