Hacker News Reader

Comment by rusk | Hacker News Reader

My experience of targeted advertising is Amazon trying to sell me things I’ve already bought ...

This is a point I bring up when casually talking with non-tech minded friends when they start talking about machine learning taking over everything...

I bought a Dyson last summer. Amazon is still suggesting that I buy more vacuum cleaners.

"We showed him an ad. He bought the vacuum cleaner."

'In that order?'

"Shhhh, just give us money."

Half-joking there -- in a previous discussion, the claim was that the ad-buyers don't care enough to check for whether these product-buyers were ones that had already seen an ad, or it was after the fact.

This makes sense. While the "you might need a second vacuum cleaner" argument maaybe checks out, the "you might need a second sous vide machine/espresso bean grinder/welding machine" definitely doesn't make sense.

There are some people who end up buying all the best coffee bean grinders in search of perfection; there are people who can't resist a classic MIG welder. But those people were going to buy them anyway, and represent too few sales to justify any expenditure.

What you need are complementary goods. These people bought coffee bean grinders? Show them ads for travel mugs, for high-end coffee subscriptions, and fancy grinder-cleaning brushes.

Agreed; just to be clear, when I've had this happen to me it's showing an ad e.g. for the exact same model coffee grinder from the exact same store that I already bought one from. Clearly no intelligence involved there.

Dyson users probably disproportionally have second homes, holiday homes, etc.. You're more likely to recommend to a friend of you see a deal "hey, I bought this awesome vacuum, you should get one, I saw a deal" ...

It's not as silly as it seems.

I buy repeat items as gifts (I like this, X will like this), or for my parents (home stuff), or when I bought something for one child I know and it turned out to be well made and now another child needs/wants one; or tech stuff that turned out well when someone else needs the same repair (eg new HDD).

It's not as silly as it seems.

I'm not sure I agree. I mean, I see the theory you're getting at, but the Dyson example is built on a few assumptions we'd need more data on -- starting with the notion that Dyson owners are disproportionately more likely than owners of other vacuum cleaners to have second homes. And even if they do, that's, well, one more sale at most, and one that's more likely to be influenced by their experience with the first Dyson rather than ongoing advertising. When you say that you buy repeat items as gifts because they turned out well, you're tacitly confirming that further advertisements for that product aren't necessary to reach you. At best, the ads can sway you if they happen to be running an unusually good deal on the product you've already decided you want to buy.

The big problem targeted ads have now is, as other people have pointed out, that they seem to be targeted with knowledge of what you've recently been looking to buy, but not knowledge of what you've recently bought. If I search for polo shirts, I'm in the market for polo shirts, but once I buy polo shirts, I'm probably not going to be in the market for them for a few months. Once I buy a car, or a television, I'm probably not going to be in the market for another one for years.

>further advertisements for that product aren't necessary to reach you //

I'm one if those "advertising doesn't affect me" people. Except, then I realised that advertising acts subconsciously and uses human psychology against us, and I'm not as immune as I thought.

I buy a Dyson, every ad I see for vacuums is a Dyson, it confirms I made the right choice, everyone is buying them, they're everywhere, etc.. Everytime I see a shiny new vacuum it's a chance for my brain to compare it with the old rubbish one; why are all those people "enjoying" vacuuming when I have to suffer it.

With polo shirts it's like "these new ones look smart/fashionable/etc." vs the old one.

Yes there is likely a lag in "this person was looking for ..." signals; but I still don't buy (heh!) that as _entirely_ silly.

>Once I buy a car, or a television, I'm probably not going to be in the market for another one for years. //

Car, probably, depends on the person's wealth; TV, I've v heard people say "we liked it so much we got another one for our bedroom" or whatever.

FWIW I'd never buy a Dyson ...

No one has the data for both sides of the equation. Google dominates search advertising, Amazon dominates online buying. If those companies merged, one company would have all the data needed to use deep learning to take all our money. They'd be able to distinguish products that, once purchased, means the buyer won't buy again for years, and items that are regularly bought, and they'd have the timing between seeing ads and buying. It wouldn't be perfect, because some people would buy elsewhere, but it might be frighteningly effective.

If Amazon of all companies, who regularly ship things to my home, can't figure out that I only have one home then maybe all this targeting stuff is hopeless after all.

And selling Hawaii vacation packages to people after they have been there. And I live on the east coast.

This also puzzles me. If I buy a household appliance that's not a consumable product (most shouldn't be), why does it recommend more of the exact same? Why doesn't it instead recommend other types of appliances?

The key take-away should be that you're comfortable with buying household appliances online, usually not that you need another vacuum cleaner within the next year.

It's data, and a lack thereof. It's easy to know you're browsing a site. It's almost impossible to know you bought something.

Banks and merchants and publishers and advertisers and all the layers in between create dozens of data silos and nobody is interested in sharing. Add to that the weird expectation that people don't want to be tracked but yet want adtech to know when you've bought something.

Well surely Amazon knows what I bought on Amazon...

Tbf Amazon has the least experience with ML of the FAANGs.

It seems that it would pay out a lot for them to do a few simple heuristics by hand. For example: if just bought something do not show a thing from the same category. At most show a thing from accessories category, which I think they do.

Up until recently i used to work at a major Indian eCommerce and can confirm that most companies use a mix of the simple heuristics you suggest together with some data mining.

So, they would take your recent purchases, clicks, searches and wish listed products and recommended products from some accessory category or related products. ie. Traditional mining combined with heuristics.

It also helps to have separate widgets for similar products and for cross-sell products.

I always wondered why that happens at all but then I thought maybe it is because of their historic origins starting out as book seller - where showing another book after you bought a book makes absolute sense. While for refridgerators it makes no sense, they should put stuff in categories to distinguish that but then they are a super big company and I am single selftaught developer trying to make my living out of it.

Maybe that just goes to show just how much money they're making that they don't really need to feel bothered!

Nah. That's not AI enough!

It depends on how you measure that. They certainly were the first to make serious money by using simple regression to suggest 'people also bought'.

In general I think that all of the personal surveillance / categorization will end up being more truly profitable (as opposed to 'twinkie calorie' profits of ad networks) as they are used to suggest price points for products and services rather than to direct attention to products.

And Booking pushing advertisements for the exact same hotel I've just checked on their service. I can't understand how that would seem like a good idea to anyone.

And Booking pushing advertisements for the exact same hotel I've just checked on their service. I can't understand how that would seem like a good idea to anyone.

This happened to me just last week. Unfortunately, it's the least of Booking's user-hostile traits.

Like sending me e-mails pestering me to review a hotel I checked into less than an hour before.

Or flashing "Hurry! Only 1 room left!" when I know for a fact that the hotel I'm inquiring about is 90% empty.

Or flashing "23 people booked this hotel in the last 24 hours!" when I know the inn only has 4 rooms and is in probably the lest popular destination in North America.

Unfortunately, there's a particular place I have to travel to which only has one motel within 70 miles, and the only way to book a room in advance is through Booking.

A hotel can be 90% empty and Booking could have only 1 left in their inventory to sell.

A hotel can be 90% empty and Booking could have only 1 left in their inventory to sell.

There's a difference between "Only one room left!" and "We only have one room left!"

It's called lying.

The opposite works as well.

On one occasion, I've tried booking straight from the hotel, it told me no rooms available for the selected dates. I've tried Booking, and sure enough, that same room I had my eyes on was available.

Larger hotels have blocks of rooms reserved for the different booking services.

The opposite can also happen, Booking shows no rooms, and calling the hotel results in a room. It just means Booking's block of rooms is all rented out, not that the hotel itself is all rented out.

These are so obvious that I find them mostly hilarious, even if they are clearly dark patterns. I can see how less experienced hotel bookers can fall to this trap though. I do end up booking with them, as for some reason they usually are cheaper and have more stock listed in Europe than the main competitor Hotels.

All those little tricks you described are part of a strategy usually called Conversion Rate Optimization. They do this because it works. What you described was likely A/B tested. Some visitors to the site saw the nagging message, while others didn't.

Turns out you can grease your numbers by manipulating user sentiment.

Who cares if you need to lie a little bit, right?

Actually... That one works, if certain conditions are met.

They also don't update that fast.

You looked at a hotel in London and a certain hotel is the best revenue generator in the area - for a little bit you're going to see that ad... or maybe you're going to see that ad for a long time, if there are no better options to advertise.

I've thought the same (and heard the same booking "blunder"), until I visited a city I had visited before, and went for the exact same hotel...

But did you need an ad to get you to do that? Seems like that's the one ad you don't need to see.

That depends on where you see that ad. If you search for that particular hotel on Google, it's going to have an ad from N online travel agencies(OTA).

If you do a broad search on OTA website, then you're likely to book the same hotel that you stayed last time... Thus pushing it up is a great way of getting you to book... But that depends on how much the hotel is contributing in marketing

Surely that would make sense the next time you visit the city, not for the same visit.

There’s some marketing theory about brand reinforcement near a purchase creating a more likely customer later. I’m not a marketer though.

The easiest way to get "brand reinforcement" is to deliver a good product. I will remember them. Right now it seems they want to get "brand reinforcement" through constant nagging which already causes me to dislike the brand.

This seems to be ingrained deeper than targeted ads. Orders give me the option to "buy it again" for things that I only ever need one of.

Or, sometimes in my experience, something I was selling.

I browsed retailers to copy the specs into my ad.

In fairness, this is an understandable mistake for any automated system to make, even more so when compared to Amazon thinking everything is a consumable I repurchase every month.

Hint to Amazon: not everything is a consumable ;)

Me too - but I do wonder sometimes if there's a long game involved, where it's more likely to make you recommend it, or give it as a gift. I can't quite see how that would work in practice, but it just seems so utterly stupid otherwise.

> Me too - but I do wonder sometimes if there's a long game involved, where it's more likely to make you recommend it, or give it as a gift. I can't quite see how that would work in practice, but it just seems so utterly stupid otherwise.

I'm going to go with Occam's razor here: the advertisers are not scarily competent but instead typically bumbling and stupid.

Or the slightly more prosaic British version: cock-up, not conspiracy. Me too, on balance. But it seems like it should be easy to mark entire categories - particularly large purchases like vacuum cleaners - as "recently bought one of those, try something else."

I've heard this is done sometimes to make you feel good about your purchase and the decision you made by showing you inferior options.

can you elaborate more on this or perhaps drop a link?

Reply on news.ycombinator.com