That's the prevailing common-sense wisdom, but it's completely wrong.

These algorithms work on statistical foundations, and the probability of you buying a vacuum cleaner if you already bought one is always higher than the probability for a randomly sampled person from the general population.

Suppose you try it out, find you don't like it, and decide to return it within a week, after which you want to buy another one. How often does that happen? 5%? That might still make you significantly more likely to buy the next vacuum than a random member of the population.

Of course, if you bought a vacuum and returned one, that would be the ultimate targeting point. I don't know if "they just returned a vacuum cleaner" is a datum provided to advertisers.

It also might be that people usually don't return it until they've chosen one that they think would be better (possibly ordering the new one, even waiting until they receive it and trying it out before starting the return process).

Not all ad / search systems know you've made a purchase. Google probably does though. Of course they should probably shut up about it.