I've noticed several cases where an ad or post closely matched something I had just been thinking about, even though I hadn't searched for it, typed it into my phone, or talked about it near the phone. For example, while taking a shower, I imagined making a video about overpriced bikes on Facebook Marketplace. When I opened Facebook afterward, the first sponsored video I saw featured someone calling a bike a "POS" while recording themselves browsing Marketplace. I use Marketplace frequently, but I had never noticed that exact video format before. How could the recommendation system produce such a specific match without directly receiving the thought?
4 Answers
If you want to test this, keep a dated record of the thoughts and the first few posts you see before checking anything related. Repeat it across different topics and compare the number of meaningful matches with the unrelated recommendations. You can also review app permissions, reset advertising identifiers, limit personalized ads, and log out of connected services. Those steps will not eliminate tracking completely, but they can show how much targeting changes when available signals are reduced.
There are several less dramatic possibilities too: the ad may have been broadly targeted to Marketplace users, the content could have been newly promoted to a large audience, or you may have encountered similar material before without consciously remembering it. Recommendation systems also test content on groups of users and quickly promote posts that perform well, so a coincidence can look unusually precise.
Using Marketplace regularly is already a strong signal. A video about browsing Marketplace may be uncommon overall, but it is a logical piece of content to show someone who spends a lot of time there. Also, people tend to remember the striking matches and forget the many irrelevant posts they scroll past, which can make the timing seem even more extraordinary.
The phone cannot directly read your thoughts, but the platforms can build remarkably detailed profiles from your behavior. They track things like what you view, pause on, search for, click, purchase, skip, and how often you use particular features. They also use information from websites and apps connected through advertising tools such as tracking pixels. That data can make a prediction feel personal even when the specific thought was never recorded.
Thinking about the idea without typing or saying it means that particular thought probably was not directly collected. The recommendation could still be based on your long history of browsing Marketplace and interacting with related content.
That explains why the topic appeared, but not necessarily why the format seemed so specific. Still, the system may have inferred that format from viewing habits, similar users, or the creator’s existing audience rather than from the thought itself.