Hacker News Reader

Comment by Aardwolf | Hacker News Reader

> Additionally, there are features such as Protected Audience that can serve you ads for “remarketing” (for example, Chrome tracked you visiting a listing for a toaster, so now you will get ads for toasters elsewhere), and Attribution Reporting, that gathers data on ad clicks.

Why is this actually effective? When you bought a toaster (which is a likely reason why you visited the listing), why would ads for other toasters still be relevant?

View on HN

Because they don't know if you bought it or not. They are gambling that you didn't and are trying to capture a sale for the ad buyer. Presumably that is somewhat effective since it's so common.

Amazon still re-shows toaster ads when they know for certain you've bought one. Maybe the likelihood you're buying one for someone else as a gift is higher than their guess of what else you might buy?

One explanation is that targeted ads are so bad at what they claim to be good at, that they can’t use the added information that you bought a toaster to do any better than some tiny increase in likelihood of buying a toaster if you just bought one.

As shit as showing you ads for the thing you just bought is, it may be the best ROI they can manage. Which is pretty damning for the whole enterprise.

Maybe, but I'll suggest an alternative view that's probably unpopular: big ad companies walk a fine line between what knowledge they can share even from _inside_ the same company. Ad sellers cannot have perfect information for probably several reasons: auditing, privacy, safety, and legal. Evidence of this is that ads are reported as being encrypted between business units (e.g. if you work on ranking product listings, you may no idea what ad is injected at the top).

Obviously the result looks stupid to us as consumers which is quite memorable. I suggest it is a by-product of intentional choices. IMO this is a better explanation than "oh this company is just too stupid to do it right, but also they are all knowing about everything you do and are just so doggone evil." Maybe they are that evil, but I am sure they have learned how not to overturn their apple cart and this is an obvious result of that.

I see language in Amazon's ad seller pages like:

"We will always automatically exclude recent past purchasers of your promoted products."

So there's some notion of this data being available, at least internally.

My hot take may be 100% wrong. Thanks!

It doesn't make much sense in case of a toaster, but if you bought a pen (or socks, SSDs, lightbulbs, etc.), you may buy some more after you tried it and liked it, or maybe you will try another brand.

That's a good point. Now I'm curious if they have some internal tagging that differentiates between these two kinds of products.

I guess it's more like a continuum than two discrete categories. On one end of the spectrum there's a family house, on the other, there's toilet paper. That's why they repeat ads even for a toaster.

I suppose I'm wrong because it appears not to exist. But, I can't get past the idea that there's just a lot of things very few people have more than one of. Many of them being big ticket items too, where the cost (or opportunity cost) for a wasted low-conversion-rate ad is high. Like bbq grills, most types of power tools, air compressors, automotive battery chargers, refrigerators, washer/dryer, dishwasher, lawnmowers, etc. And many of those have obvious companion items that would be a better use of the ad space than trying to sell someone a second bbq grill, for example.

Some items on amazon shows on a "buy it again" list, I've some appliances appear there after a couple of years...

I’ve always thought it was because you’re more likely that you’ll buy another toaster than something in any other individual category. More likely that your next purchase will be another toaster than a pitchfork, for example. Still stupid though.

I always opt for the toaster-pitchfork bundle with monthly subscription from Prime, because a toaster would be nothing without its accompanying pitchfork!

you might still return it.

Me visiting a listing for a toaster means that I'm in the market for a toaster, doesn't necessarily imply I already bought one.

Also, with this data, they can potentially build a better model than Amazon's recommender model. Most consumers have a pattern when buying and there's a lot of money in figuring it out.

This is already a solved problem, you can exclude audience based on purchase.

They think you are in ‘research’ mode - so you are more likely to be looking at the various toasters in the market, reviews etc.

I expect most people research before buying a specific toaster.

There is a likely chance that you didn’t buy it, so bombarding you with other toaster ads(preferably much cheaper) raises the chance of you actually buying it.

There might be other cases where you bought one and is really happy about the benefits and want your friends to enjoy too but may be they can buy a different brand/price/featured ones to compare experience.

There are lots of possible such cases ad and sales people think up.

Src: worked in adtech for few years and yes am not proud.

When I have my ad defenses down (no ad blocking, some random browser), I see ads for mobile cases for my Samsung Galaxy Note 2 (released in 2012). I guess that was the last set of data points they have against my persona. That's the only relevant ad I see that I remember searching for.

I've had 3 other phones since then. Ads just show what they can. Still better chances for a click than random.

Because you'd likely want to buy a new toaster in a few months, after the one you just bought breaks due to planned obsolescence, or low quality parts, poor QA, and no possibility to fix it yourself. /s

Those advertisers, they think of everything. :)

Let's take the toaster out of context here and then if that's just a "product" and the brands/governments want you to purchase it frequently that's why.

Hot take (as it applies to toasters): many appliances are crap, they break quickly and are not user repairable.

The odds of buying a toaster a year later might be higher than it sounds.

you could still return it when you see a better toaster on an ad. It's probably a stronger signal than what they could otherwise get

Reply on news.ycombinator.com