The targeting only looks bad because you don't know how valuable a lead is in the relevant business. If you're a tenth as likely to click on something with ten times the customer acquisition cost, then from the platform's perspective these are equally profitable to show.
Case in point: do some searches on mattresses and see what ads you see over the next weeks. Huge value per customer, so if you can plausibly be lumped into mattress firms' targeting options they'll win the auction.
> do some searches on mattresses and see what ads you see over the next weeks.
This sounds like an interesting experiment. I mean, the result seems obvious for a search on mattresses, but I think we can expand on this to search for more controversial or political topics. I'm going to start planning this experiment and I'd be very interested if anyone knows of an existing study along the same lines.
Case in point: do some searches on mattresses and see what ads you see over the next weeks. Huge value per customer, so if you can plausibly be lumped into mattress firms' targeting options they'll win the auction.