It wouldn't be legitimately useful, but it can be abused by intelligence agencies. You can imagine a machine learning algorithm trained on the shopping history of terrorists and associates, which then scans the entire database to identify new potential terrorists.
For example, the US has a No Fly List with millions of algorithmically identified potential security risks. If you're on the list you can't get on a plane, but you are never notified or given a reason, and can't challenge the listing. Here's a paper that describes the issue I'm talking about: https://pure.uva.nl/ws/files/4284150/61150_Goede_M._de_Trans....
> This paper argues that the deployment of transactions data of many kinds has become the banal face of the war on terror’s preemptive strike. Because the failure to predict and prevent 9/11 is partly thought to be a failure to ‘connect the dots’ of available intelligence, post 9/11 policies seek to register, mine and connect ever more ‘dots’, or association rules, in the form of credit card transactions, travel data, supermarket purchases and so on. We argue that it is in these ordinary transactions that another spatiality of exception is emerging, one in which the traces of habits, behaviours and past practices become the basis of security decisions to freeze assets, to apprehend, to stop and search or to deport. As such, these developments constitute a relatively unacknowledged violence in the war on terror, which is in need of critical questioning.
> Supermarket checkout staff are being trained by the security services in how to detect potential terrorists. MI5 has been secretly advising food retailers, including Asda and Tesco, on how to identify extremist shoppers. Measures include [...] being alert to mass purchases of mobile phones, which can be used as bomb detonators. The awareness training for staff also covers bulk sales of toiletries which could be used as the basic ingredient in explosives.
With a full supermarket loyalty database, you can just scan for anyone with suspicious toiletry purchases and an ethnic-sounding last name and bring them in for questioning.
If you know the suspect purchased a bag of funyuns, pack of marlboro lights, and a diet dr. pepper 1 liter, having the shopping history of every 7-11 customer for the last X years could give you a high probability of identifying exactly which customer it was, or eliminating exactly which customers it wasn't, from a list you compile from a large set of information sources.
Using metadata and tracked information of known individuals can illuminate the lives of people who aren't tracked through process of elimination and correlation, which is why privacy rights are so crucial to legislate correctly. Right now, the US justice system is not at all equipped to properly handle the scale and scope of private industry's panopticon providing more or less total global surveillance.
We need to see some legislation with teeth, big and sharp enough to completely kill any business, no matter how large, if privacy isn't respected. But hey, let's all enjoy being tracked, logged, monitored, and surveilled every second of every day in the meantime.
