What other kind of explanation do you expect? People have tested it by playing a Spanish soap opera on their TV for a few hours, leaving their phone in front of it to listen. They then started getting Spanish advertisements. I realize that there is a confirmation bias in that they are then actively looking for Spanish-language ads (to test their theory) but don't you think they would have noticed them at other times? They explicitly ran an experiment and got the verification.

In fact, some people were incredulous and re-ran the experiment thinking there's no way they would start getting Spanish ads: they did too. That's all it takes to convince me. I mean what other mechanism would get back to Facebook? As far as I know TV's don't report back which specific channel they're tuning into (all channels are streamed concurrently.)

I'm all for a more scientific test but at the moment I find the anecdotal evidence quite convincing. I just outlined how a scientific test might work here:

https://news.ycombinator.com/item?id=15782323