A few comments here claim that aggregated data is fine. An interesting read is [1]. It discusses how trajectories of individuals can be recovered from aggregated mobility data with high accuracy. It's a great read because it breaks down the approach into small logical steps, but the end result (recovering individual's trajectories from aggregated data) sounds bizarre at first.
People are often able to get personally identifiable information from aggregated/anonymized data. At this point, I think people should be automatically skeptical that their personal habits are protected when a company claims they only collect/use/sell aggregated/anonymized data. In practice, it might not be protected at all.
[1]: https://blog.acolyer.org/2017/05/15/trajectory-recovery-from...