Hacker News Reader

Comment by paulie_a | Hacker News Reader

Oddly enough Facebook has the lowest quality of ads in my opinion. For the large amounts of data they have on me, they don't seem to utilize it effectively

View on HN

Have you seen anyone using it effectively?

Absolutely not. I'm really amazed by how bad Amazon is sometimes.

I buy cat food once for a friend - eternally and constantly get recommendations for more cat stuff.

I buy coffee filters every 6 months - I get "buy it again" a week later. This lasts for a month. Then it goes away. 5 months later, end of my regular buying period, and they seem to have forgotten all about them.

Ads/recommendations are shown to you at an 'auction', companies bid in micro-second level timing to show you an ad/recommendation.

So, when you say that you get low quality ads, you are telling us that the bidders think you aren't worth much. Hence, you get the same ad over and over, because you are being targeted by the 'long tail' of the bidders. Also, they 'forget' you for the same reason, because you aren't worth enough to them to become a 'regular'.

It's not the bidders, it's that you wisely don't respond well to ads/recommendations on the internet.

No offense ;)

Not all the recommendations are paid ads.

But they are equally terrible in quality

So then why is that? I'd assume general stupidity, we're humans after-all. But there may be something else going on. Any ideas?

Ah well that put a positive note on my morning coffee. I didn't know that.

But still wouldn't it be valuable to the bidders for Amazon to fix this? Or shouldn't Amazon value better targeting because relevant ads convince users to engage with the platform more? The people selling the coffee filters are still guaranteed my business, just not before 6 months. I always buy from the same seller too so it could be a good opportunity for a competitor to sneak in right at the 6 month mark and make a new repeat customer.

I guess the answer to that is probably "well the system works well enough elsewhere so that would just be hyper-optimization"

Yeah, with large companies like Amazon, never discount bureaucratic log-jams and just regular stupidity. I agree with you that optimizations can be made, but the system may be in a 'not broken, don't fix' local maximum. The inertia and complexity may be too hard to over come. Twitter only recently upped the character limit after nearly a decade because the old code-base was spaghetti and didn't allow improvement.

But these are the reasons why the large CEOs have always been terrified of 'two guys in a garage'. They know they have issues like this, and that the little scrappers don't. If anything, these little issues are the camel's nose under the tent of Amazon.

I’m surprised how it can’t even stop advertising a you-only-need-one product like an ironing board to me a week after I definitely purchased one. This isnt even machine learning level problem space.

Very true. Oh you just spent $400 on a top of the line product that'll last a decade? Here, check out these $100 bottom of the barrel alternatives that'll surely break in a month.

I get that Amazon is doing auction advertising but it seems like the value of their ads would be higher if they could correctly target them.

Reply on news.ycombinator.com