It's good to point out this possibility, but outside of an institutional purchaser, people don't buy 2 mattresses with the purchases separated by a week.
Knowing that you bought a mattress is useful for the ML to find the mattress-purchase-likely cohort. Filtering those people out just saves the advertisers money. Hence the grift.
What if they bought a mattress and found out they didn't like it after one night? A lot of stores let you return them if you don't like them within a number of days. That would mean they are in the market for another one.
I'm not saying it's likely, but there are scenarios.
Really 0%? It seems like some of the buyers would return their mattress and buy another. If 10% return and 90% of those buy another brand, that seems to me like an above average audience to market mattresses to.
>The platforms aren't as dumb as the post author thinks they are.
Maybe the platforms are not that dumb, it is the mattress seller that doesn't properly make use of that feature, or there could be other reasons why they insist on suggesting over and over again a mattress.
The experience described has happened to everyone for any (not very often replaced) item bought online, be it a printer, a wash machine or whatever, so if the platform is not dumb, it lacks data and the sellers are dumb, in any case the final effect of this "targeted" ads is 0 added sales.
It's more likely that the mattress company couldn't be bothered creating a separate segment from their purchase order completion page and then excluding that from ads, because 95% of people who come to the page don't buy a mattress, and 95% of people who have bought a mattress don't click on ads when you show them, and the cost for AdRoll retargeting is per click not per impression.
Well, according to Douglas Adams, somewhere in the Galaxy is a swampy planet populated by sentient mattresses named Zem:
“No one really knows what mattresses are meant to gain from their lives either. They are large, friendly, pocket-sprung creatures that live quiet private lives in the marshes of Sqornshellous Zeta. Many of them get caught, slaughtered, dried out, shipped out and slept on. None of them seems to mind this and all of them are called Zem.”
Maybe we should boycott mattress ads for the sake of preserving the species.
I am willing to bet that statistically (with the number of mattress companies offering X night returns) that a recent mattress purchaser is more likely to purchase a different mattress within X days than the random public.
Some % of the population buys more than one mattress in a 3 month period, intentionally. Perhaps they are landlords, or have 2nd homes, or want to stock their new home with new furniture.
Maybe, but (thank goodness) the mattress ads bombarding does not last 3 months, only a week or so, after 2 and 3/4 months have passed that % of population will either look where they bought the previous one (if they are satified with it) or make a new google search, in both cases the week long targeted campaign won't have much effect.
Knowing that you bought a mattress is useful for the ML to find the mattress-purchase-likely cohort. Filtering those people out just saves the advertisers money. Hence the grift.