It's possible that some combination of efficient markets, semi-inaccurate/incomplete tracking data and chaos theory combines to make it mostly irrelevant.

So Steve just got a promotion and likes the BMW 3 series, but advertising to him is useless because he has already decided to buy a BMW 3 series. Whereas Larry has had the same job for ten years, and the same Ford for ten years, but if you put a luxury car ad in front of him it may get him to take a test drive and actually create a new customer.

The data says Mary is single and goes to the gym every day, but it only thinks she's "single" because she's in a committed long-distance relationship and isn't interested in dating anyone else. And she's a fitness expert who is willing to spend her time researching fitness products, so she already knows everything there is to know about those products, already buys the ones she wants, and advertising them to her isn't going to create any new exposure. Whereas Jane never goes to the gym, so you might actually sell her a gym membership or a piece of fitness equipment because she doesn't already have one or know anything about it yet.

John has a 10th wedding anniversary coming up, but he has also known exactly what he's going to do for years. Whereas he has a third cousin whose wedding is coming up and has no idea what to get, so you should be showing him ads for toasters and flatware rather than jewelry and chocolate.

In general, you may do better to advertise your stuff to the people who aren't already interested in and knowledgeable about it. Which actually looks kind of a lot like random scattershot rather than targeting.

I don’t want to buy beef jerky when I’m reading about a military strike in the Middle East, no matter how much my profile indicates I like beef jerky — but I might be open to a book on politics.

Because reading about content is self-selected disclosure of interest, the NYT already has all the information they need to target me — they know I read politics, and when I’m in that context.

The only thing Google can provide is slight refinements on which political book to suggest — which isn’t far enough above the noise floor to matter. Anything else is just them giving the NYT statistical fuzz to pretend carrying their ads on beef jerky isn’t an all-around negative so they can fleece advertisers.

Targeted advertising isn’t about efficiency, it’s about raising the number of places they can (uselessly) place beef jerky ads so as to increase their cash flow.

This seems to me what "targeted advertising" should mean - NYT says "for all routes of nytimes.com/politics, show political ads. If it's local news, open it up to classifieds for local business." Etc.

If someone's on the page, is that not enough information?

For Google and Facebook it doesn't make sense, but for a newspaper i would expect some selection on the ads they show.

Instead of only political books being able to advertise on that page, they’re bidding against beef jerky and cars and so on. This raises the price for the ad slot by creating (false) demand.

Without those out-of-place, targeted ads there would be less demand for every ad slot, and people could bid less for them.

Heck, even Google AdSense scans your site to match ads with the content (source: I've tried using AdSense). They just don't rely exclusively on the content for ad targetting.

for actual comparisons: https://thomashunter.name/posts/2019-01-09-generic-banner-ad...

The whole surveillance capitalism nonsense we live under is a way for ad networks to justify their existence. If you're a company that has data on everyone, you need to convince people they need that data to accurately sell ads, and targeted advertising exists solely for that purpose. The amount of cross-site tracking that these companies are able to do outscales what any company could manage to accomplish themselves, or even for a smaller competitor to step in, so as long as targeting users across the entire Internet is believed to be a must-have, these ad behemoths stay on top.

> Dear Amazon, I bought a toilet seat because I needed one. Necessity, not desire. I do not collect them. I am not a toilet seat addict. No matter how temptingly you email me, I'm not going to think, oh go on then, just one more toilet seat, I'll treat myself.

It is easiest to understand it with coke. It is so prolific that people say "coke" instead of "soda". That we have "is Pepsi okay?" as a joke. Instead what these ads to is make you feel good for your choice in coke. Not to convince you to buy it, per say, but so that when you do buy it you get an extra kick from those sweet endorphins.

So is Amazon trying something similar? Perhaps the opposite? As in "Hummm... maybe I could have gotten a better one", causing you to return and replace (or just flat out replace). But that might be a weird strategy.

I was researching cameras. I saw very little camera ads in my research. After I bought a camera? Ads for the camera I just bought everywhere, for months afterwards. I have not once clicked on a ad and bought something. I maybe clicked on 10 ads in my lifetime and they were 75% stuff I THOUGHT I had no intrest in. I think targeted ads is the biggest con no of our age.

Even more egregiously, maybe sometimes the advertiser's metrics have some reverse-causality: you buy the camera, then they show you the ads for it; but something in their metrics isn't properly modeling the sequence of events (e.g. they correlate retargeted camera ads shown per week with camera sales per week). Then they mistakenly take credit for a sale that happened before the ad was shown.

That's not to say such mistakes always happen, but I wouldn't be surprised they didn't happen fairly frequently. Given that measurement is hard and both adtech and advertisers are motivated to present success stories, and may not be too motivated to second-guess positive-sounding numbers.

I can't even imagine the outrage that would happen if it emerged that Google was using its access to people's emails to measure ad effectiveness.

Incidentally, those weren't targeted ads. Those were ads hand-picked by the person producing the content about those books. From that very small sample I conclude that ads relevant to the context in which they're placed are more effective than targeted ads.

I will say that I'm thankful that Squarespace, Audible, and Skillshare support independent video and podcast creators. I click their referral links sometimes to check out new features etc, but would only ever sign up if I needed that particular service. Still, I think that form of repetition helps their mindshare in a positive way.

I did buy the Glif directly from an ad on Daring Fireball, albeit months later, and in combination with brand awareness from their free time-lapse / speed ramping app.

This is annoying and, when one thinks about the implication that everyone and their dog knows who you are, scary.

I have never bought off of ads otherwise and find that most of the time they are not relevant to my interests. I also started using an adblocker because of other reasons (mal-ads/autoplay) even though I previously did not do so because I felt text/banner ads were quite acceptable in order to help keep the lights on.

Mostly clothing...things I didn't even think did existed

And it may even be profitable to be the first to turn away from it when everybody else is on it, and instead use more of your ad spend for non-zero-sum customer generation. Then you get all the new customers to yourself while everyone else burns their margins fighting over the existing ones.

Winner: ad platform.

Except that they do have a choice.

Suppose there are five competitors. Everybody knows Steve is going to buy a BMW 3 Series. The margin is $5000. Assuming everyone else is going to do what you're going to do, how much does it make sense to bid to advertise to Steve? If you bid $1000, you have a one in five chance of making a $5000 margin. If you bid $5000, you have a one in five chance of making $5000.

So the optimal amount to bid is $0, because you have the same one in five chance at $5000 if everyone bids $0 than if everyone bids $1000 or $5000, but then you have a $1000 expected value rather than a $0 or $-4000 one.

Now suppose the others aren't using the same utility function as you and so may bid a different amount and you get a bidding war. So the first thing that happens is somebody bids $100 expecting to make $5000 -- bully for them, until the second one bids $200. Everyone raises their bid to $1000 because that's what a 20% chance at $5000 is worth. If one of the competitors drops out, the others will raise their bid to $1250 because now it's a 25% chance.

Which means there's never any profit in it for the advertiser. If anybody plays then at least one competitor will retaliate and everybody loses a total of $5000 to the ad platform. If nobody plays, everybody gets an expected value of $1000. Being the first to quit costs you nothing, but being the first to play costs you and each of your competitors $1000. Who is going to play this game?

In reality what happens is marketing runs an ad for $3000 offering a $1000 incentive, Steve makes the purchase, and making gets to (possibly correctly) claim they netted $1000 in sales.

Those marketing people want to keep their jobs, so they are going to work hard to justify their existence!

No, I am assuming that advertising works the same for everyone. If there are five dealers and fifty customers and with no advertising each dealer gets ten customers, then with each dealer spending $50,000 on targeted advertising, each dealer still gets ten customers. When each dealer spends the same amount they each get the same benefit, so they still split the existing customers five ways.

Where are the customers gonna end up?

Probably the clearest example I can think of is doing a no-adblock Google Search for "Coca-Cola." The first result is an ad, put out by the Coca-Cola Company. This might not seem to make any sense, since the first non-ad result is an identical link to the same website, but imagine what Cott Corporation or PepsiCo would do if they could by the first result for their competitor...

If I have a $20 margin, I am better off giving Google $15 and getting an incremental $5 profit than not advertising. Even worse if my customers have a lifetime value of $50 it may make sense to give Google $45 and not see a return from my spend until years down the road while Google gets the $45 right now.

Maybe a payday loans company finds that people like Steve are less likely to be interested short term loans - so they avoid advertising to him.

I have been running my online business for 5 years now. I learn (and keep learning) all I need to know from the same, high quality source.

I get targeted all the time by adds on landing pages, email funnels, Facebook ads, etc. Some of these are legit, but thanks to my experience I can see that most of them are of poor quality.

They will never sell me anything, while they might sell something to someone that hates his job and is looking for another way to make money.

Unless you have a promo, you really don't want to advertise to them... But that is exactly what you end up doing, because you can't target "people that visited my website 6+ months ago".

When someone stops visiting your site, make sure to send them a email. Ask them to return! You have their email address, right?

Also Google could place a survey question instead of an ad, asking "Which of these products have you purchased in the past month?"

I think a message's context matters - and this is almost self evident from examples like ads for airline tickets being shown on plane crash articles. I think targeting is great, I really do, but I think the value of trusted brands is likely just as, if not more, strong.

If context matters, and I think that argument was the entrenched idea pre-internet and has not been disproven, then where is the positioning of targeting? My hypothesis would be that targeting is likely best utilised when people stray from self selecting brands, e.g. NYTimes needs no targeting, but a smaller publication likely does, as people aren't there for the brand's known positioning, but because of their targetable interests, e.g. they ended up on a site about coding, or computer games, or knitting or whatever.

If true, that would be easy for an ad network to detect. If the click through rate for ads on a topic not matching the users interests (as determined be browsing history) are higher, then your theory is correct.

It's basic stuff, and I'm sure they already do that.

But the ads I see are way off target.

Hell Google's news ... AI or whatever they're doing has been absolutely convinced I'm a Nebraska Cornhuskers fan, probabbly because my team plays them and I google something about them once in a while.... but in the past even when I told Google directly "I don't want to see stories about Nebraska Cornhuskers" ... after a while it again becomes convinced I'm a huge Cornhuskers fan.

I'm just not sure these systems are ... that effective.

Not to say it is easy, the systems have to be really complex / interesting.

The two sides see "off target" quite differently. If the baseline is 0.1% accuracy, and an ad platform offers 1% accuracy, then you see relevant ads showing up at 1 in a 100 rates rather than 1 in a 1,000 rates. You still feel like you're being spammed by useless nonsense. As an advertiser, my cost to reach my audience just dropped by 90%.

I'm really curious how those happen.

At one point it was sure that I was female (am not, have never been) too.... to their credit they figured it out after a year or so, but man the ads. But that was at least a limited run of confusion.

But probably the truth is that these systems aren’t that good :)

Funnily enough when I'm discussing the dangerous nature of large datasets in the hands of companies/bad actors (sometimes the one and the same...) and the "creepy" effects this is one of the examples I use, I've found that if you know a little about the person and couch the argument in a way they can personally respond to, they understand even if they don't know the first thing about technology generally.

EDIT: https://www.forbes.com/sites/kashmirhill/2012/02/16/how-targ... for anyone who hasn't seen it.

For example they may show you an ads about hairdryers after you searched for computer keyboards. There is absolutely no relationship between the two, but Google doesn't know it yet, so they try. After all, there may be a correlation. In the likely event where there isn't, Google will simply stop showing hairdryer ads and show you computer mice instead (a known correlation) or running shoes (another experiment).

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

'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.

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.

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).

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.

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 ...

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.

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.

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.

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.

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.

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

It's called lying.

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.

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.

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

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.

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

I browsed retailers to copy the specs into my ad.

Hint to Amazon: not everything is a consumable ;)

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

We are far away from a machine classifying you as being single and going to the gym every day or liking BMW and having gotten a promotion.

Not that I don’t think we couldn’t get there...but today these system are far away from it!

Just go to your Facebook ad settings to understand how primitive the information is they have about you.

But I agree that targeted ads (as primitive as they are) are more efficient then I untargeted ads

If Google's ability to discern my interests for ads is as bad as is their ability to discern my interests for news, then I would not be surprised.

I have never been aware of listening to any music by the band "Foo Fighters". I've almost certainly heard their songs in passing, but never in a context that gave me the name of the band.

A friend on Facebook posted a link to a YouTube video called "Dave Grohl brings kid on stage in Kansas City to rock out", in which Grohl, who is apparently the lead of the Foo Fighters, invited a kid from the audience at a Foo Fighters concert up to play a song, and I watched this video. (The kid asked to play a Metallica song, which Grohl and the Foo Fighters knew and played, so I still have not knowingly listened to any Foo Fighters music!)

Google has latched onto this and decided that I am interested in Dave Grohl news. Every time I go to Google News, for the last several weeks, the top of the "recommended by your interests" section is a story about Grohl or the Foo Fighters. Even though I hit the "show fewer stories like this" link for most of them, still they come.

Furthermore, they are almost all negative stories. The one thing they have of me showing any interest in Grohl and/or Foo Fighters, that video of the kid on stage, was a positive thing. But what they are giving me is a parade of stories about Grohl screwing over current or former bandmates, Grohl misbehaving while drunk, etc., from what appear to be trashy gossip publications. Yet a bit of research shows that Grohl is apparently actually a nice guy, well regarded. If I click the link to tell Google to not show me anything else from one of these trash publications...it obeys and just turns to other one to find Grohl gossip for me.

What the heck, Gooogle!?

There doesn't appear to be a way to filter out sports I'm not interested in, or even all sports.

Apple News has a hard time telling the difference between Paris, France and Paris Hilton.

I don't think so.

A human would understand context. AI isn't supposed to be simple string matching.

Also, there is no "Paris Hilton" hotel. There is Hilton Paris Opera, Hilton Paris La Defense, and Hilton Paris De Gaulle.

"Letter from Foo Fighters frontman to Cornwall Council goes viral again"

When a targeted ad is creepily accurate and on point, people flip and think that the machines have figured us out. But the ads that are irrelevant just fly by us without a hit. The one-out-of-a-hundred ads that get it right, likely due to an ad targeter's lucky strike, are the ones that get under our skin, and the only ones we really notice.

It was Target in the US:

https://www.nytimes.com/2012/02/19/magazine/shopping-habits....

> As Pole’s computers crawled through the data, he was able to identify about 25 products that, when analyzed together, allowed him to assign each shopper a “pregnancy prediction” score. More important, he could also estimate her due date to within a small window, so Target could send coupons timed to very specific stages of her pregnancy.

> ...

> “With the pregnancy products, though, we learned that some women react badly,” the executive said. “Then we started mixing in all these ads for things we knew pregnant women would never buy, so the baby ads looked random. We’d put an ad for a lawn mower next to diapers. We’d put a coupon for wineglasses next to infant clothes. That way, it looked like all the products were chosen by chance.

I can see how this could cause embarrassment and, maybe, awkward questions if one's partner also uses the same account.

I'm always surprised when I see a company wasting money on adwords when they are also ranking many #1 keywords organically, specially on brand keywords. I may be outdated but I doesn't make any sense to me.

It's clearly confused, and so my wonder is how do they correct that over time? How do you un-scramble a scrambled machine learned profile? Can you even detect that it's gone awry?

The models react strongly to that...i went from years of poor ads to fantastic ads within 2-3 purchases

They are very efficient in telling people that are intending to buy your product that they are going to buy your product. Or even buy your advertised product on the website that you're advertising.

They are not efficient in acquiring new customers, as most(90%+++) of the ad impressions go to the people that are already aware about your product.

The only truly efficient ad targeting is restricted by geography, culture, age and timezone.

It may be that targeted ads are more effective, but some counters to your points:

> John has a 10th wedding anniversary coming up.

A wedding anniversary is a significant thing–some John's may be swayed by an advertisement, but many would consider it a bit of a cheat not to put more "novel" thought into it.

> Mary is single and goes to the gym every day.

Frankly, the former of those two data-points could mean anything. Much of what advertisers believe about consumer habits of single people could easily be correlation.

As for the latter, anyone with a daily gym routine is less likely to want to change it. Selling gym membership to someone speculatively hoping to start going to the gym is a better bet, but much harder to track.

> Steve just got a promotion and likes the BMW 3 series.

Going back to the anniversary gift—I seriously doubt many people are much swayed by one-off ads in buying something as significant as a car (unless Steve has 10+ cars, in which case the promotion is less relevant). This is case where untargeted ads definitely have a much bigger role (Steve sees many BMW ads passively over a number of years).

It may be there's some cases were targeted ads are extremely effective, and some where they're not. And—as you've pointed out—it may be that Google et al. aren't really incentivised to actually be good at targeting as long as their advertisers believe they're good at targeting (which goes hand in hand with market dominance).

Combine this with marketers wanting to be more data-driven - and needing to report numbers to their own bosses - and you've got a situation ripe for exploitation. I expect the adtech companies could sell pretty much anything, as long as it smells like numbers.

> A wedding anniversary is a significant thing–some John's may be swayed by an advertisement, but many would consider it a bit of a cheat not to put more "novel" thought into it.

No idea what you mean here by cheating or novel thought. We're talking about whether knowing that info allows you to have more successful advertising compared to random chance. Serving John ads for roses, chocolates, and vacations will be more effective than serving those ads to a random person.

>Frankly, the former of those two data-points could mean anything. Much of what advertisers believe about consumer habits of single people could easily be correlation.

Correlation is the whole point. If someone who is single and goes to the gym is correlated to certain purchases or behavior, you can advertise those. And being single, a woman, and a gym-goer isn't super valuable by itself but combined together you can advertiser women's athletic clothing that is functional but also attractive.

> As for the latter, anyone with a daily gym routine is less likely to want to change it. Selling gym membership to someone speculatively hoping to start going to the gym is a better bet, but much harder to track.

Again, you don't understand advertising. If I know someone goes to the gym every day I'm not going to advertise gyms to them. I'm going to advertise water bottles, protein, healthy meal kits, athletic clothing, etc.

> Going back to the anniversary gift—I seriously doubt many people are much swayed by one-off ads in buying something as significant as a car

We're not selling Steve a car with targeted ads. We're taking his promotion and propensity towards bmw to assume hes wealthy and likes luxury goods. You advertise more expensive goods to him, rather than cheap ones.

That statement makes sense intuitively, but what we're discussing here is whether our intuition reflects reality. It may seem to self-evident to you, but do you have data?

All of your examples "make sense" from the same intuitive perspective, but all assume the subjects are positively influenced by the advertisements more than a randomly selected subject would be. That's an assumption.

Please don't condescendingly remark that I "don't understand advertising" when you've missed the point entirely. This isn't about how the ad industry works—ad companies are clearly economically successful—it's about whether (as another commenter eloquently put it) that emperor is wearing clothes.

> Correlation is the whole point. If someone who is single and goes to the gym is correlated to certain purchases or behavior, you can advertise those.

The point is that the correlation we're talking about here is a proxy. If a lot of people buy a product and are single, but most aren't buying that product because they're single, it means they're buying it for another reason, which may or may not change relative to their relationship status.

Your deep bias is breaking your logic. These are assumptions that marketers test. If they work, you keep doing it. If it doesn’t, you try something else. You clearly have no idea how the industry works, otherwise you’d understand that you’re arguments have been addressed.

Correlation is the entire point of that advertising. No marketer cares whether being single causes a purchase, simply that being single is correlated with a purchase more than a random amount.

I don't have any data on my own actually behavior, only my feelings. If I search for "React" on Google and ads for react based services or other software services appear it doesn't bother me. If I'm on stackoverflow or jsfiddle and there are ads for software dev related products and services great! If I'm on Polygon and see ads for games, perfect!

It's when I see targeted ads unrelated to the activity that I feel angry, annoyed, upset. For example I viewed some apartments on an apartment site. Then I went and checked movie reviews on yahoo (japan) and every ad was for that exact apartment I looked at. It's not just creepy it's anger inducing. I'm not interesting in looking at apartments. I'm looking at movies. It's like the sales person from the last store I visited followed me into the movie theater and is badgering me to buy the clothing I looked at

So, I'd basically like to believe there's 3 tiers

1. 100% un-targeted, random ads

2. Content targeted. Video game ads on a video game site. Or BMW ads on an article about BMWs

3. User targeted. Being tracked all over the net to try to divine what I want and showing me ads for that.

It's easy to believe 2 and 3 are better than 1. I want to believe 2 is as good as 3, maybe better since it's not creepy.

No, they literally could not, because the data flow between Gmail and Ads has ben severed a while ago:

https://blog.google/products/gmail/g-suite-gains-traction-in...

These kind of things are treated pretty seriously. There are teams whose job is literally making sure that your data does not leak from one Google product to another in a way that would violate your privacy. Source: I'm an engineer in Google, but in none of the mentioned teams.

Virtually all google products are governed by a shared google privacy policy, which allows Google to combine data across products and services. This change was made years ago, despite some end user protest.

Most apps and products are also optimized to maximize data harvesting - ex. Android OS and Maps will nag if users don't give up data. So I don't really believe what you're saying.

However, once you allow the idea that there are powerful people inside these companies, that genuinely don't want the company to create a panopticon, things can get interesting. I can't exactly prove that to you, but you've seen hints of that in the news over 2018. Or, you might have noticed e.g. that safe browsing only sends hashes of the URLs you type. Or, now this Gmail not telling Ads which languages you speak thing.

Now, again I can't offer more than my word for it, but the Gmail-Ads situation is the rule, not the exception. The only reason this made it to the news, is that the exception in place before has been lifted. But in Google, by default, if a product wants data you shared with another product, it has to make a very strong case that that's covered by intent of a permission you already gave. That, or ask you for permission, or ask you for the data again. If you ever wondered why Google keeps asking for the same basic immutable data, that might be the reason.

Funnily enough, this whole privacy thing is hard to tell from the outside. As you said, there is a single privacy policy, that governs the relationship between Google and yourself. All the internal data siloing is, well, internal.

Edit: I won't be continuing this thread. The cognitive load to not leak anything non-public is a bit unpleasant.

It allows you to sell more ads to the marketing people since it sounds very compelling. So there is your model.

The fact that I am still shown ads for items of the complete opposite (football) team that I am clearly (online visible, on FB, in my gmails, from my google search history, from my chrome browsing habits) fan of; show the targeting is still... moderate to say the least.

The thing is if there is something I want to buy, or own at some point in my life, I don't tend to forget about it or need some reminder from a targeted ad. If it's something I really want I'll usually think about it daily.

On the other hand, there could be something that I need, and have never searched for, that I have forgotten about such as going out and buying more toilet roll.

They would be the ones making the ad, targeting someone X miles from them, who is interested in BMW's.

Car just won an award, which would direct him to look up more info about the car and why it's considered good.

Maybe they just made an update to the car model he likes, but is not aware of this change.

We tend to think that these companies know us very well, but (anecdotally) that's not my experience on Facebook.

There is a page where you can see a list of things Facebook thinks you like. In my case, it's full of things I don't care about and I would never buy. They are there because, according to that page, I clicked on some ad at some point.

Which I don't remember, so it might well be five years ago. So I get targeted by ads I am not interested in at all. A random ad has more chances to capture my attention than something targeted on a non-interest.

I think there are a few factors:

1. Targeted ads are more likely to show people things they're already familiar with. The ads are more "expected" so there's less novelty. Untargeted ads can inadvertently hit wants and needs that aren't part of the user's tracking profile.

Closely related to that: targeted ads can induce their own special kind of fatigue. If all you see are diet ads because your profile says you want to lose weight, you're going to get really sick of seeing diet ads. The targeting actually works against effectiveness.

2. Targeted ads are much more exploitable by scammers. See: https://www.theatlantic.com/technology/archive/2018/01/the-s.... This leads to a greater erosion of trust. The expense of traditional and untargeted advertising weeds out such scams.

3. Building a good user profile is hard. If you get it right, you have the problems I described in #1. If you get it wrong, you're blasting someone with ads that are completely off the mark (like weight loss ads for a very healthy and fit person).

4. If a user sees one targeted ad a week, that ad is likely to punch above its weight. If they see 50 targeted ads every day, the effect is much different. They're more likely to develop negative emotional reactions to the ads, as they gain awareness and get creeped out by the tracking and attempted manipulation.

https://www.forbes.com/sites/kashmirhill/2012/02/16/how-targ...

- I already bought that recently, why would I need another one?

- I already decided which variety of thing you're showing me I will buy, this ad isn't helpful to me.

- Why are you spying on me? That's creepy, I'm not going to click that ad even if it was exactly the thing I wanted to buy right now.

I expect that this experience isn't extraordinarily unique. Having less targeted ads is actually somewhat helpful because it maintains the pretense that ad companies aren't spying on you constantly.

I contend it is due to untargeted ads.

If you could actually create ads for things that I would want, personalized ads are probably worth more. I just don't think it is possible with current technology.

http://en.wikipedia.org/wiki/Reactance_(psychology)

People know when they're being manipulated and we usually don't like it. All the data in the world won't help you if using it creates antipathy.

John has a 10y anniversary - will make John focused in buying stuff, he'll need exposure to new products... Not ads for stuff that he already looked at. John is probably not an amnesiac.

Mary is single and likes gym - doesn't make her more interested in gym clothes, she's already well aware of the gym clothes...

Steve getting a promo bundled with him liking BMW will be a total waste of a ad target. He already knows the cars and will buy a BMW with or without an ad from any car manufacturer.

Going on a diet implies that you know what diet you're going on.

In short retargeting is great at reminding people of what they like. It's really not great at acquiring new customers.

I worked on this stuff and it's flashy, but ultimately just a great way to siphon off marketing budgets out of corporations.

Don't sell your stock just yet. The product is useless, but it's a scam that is as prevalent as organized religion

It would be really nice to see those people buying to understand it more.

Amazon for example still shows me washing machines after i purchased one through amazon. But that feature is still here on a high traffic user landing page!

Perhaps it matters for products which are clearly build for a very specific target (like a startup selling a shopping service to 30 year olds which work all the time) but there might also be a lot of standard products which everyone would buy?

Dog food, dog toys etc.

Also i often enough see companies paying (at least those are ads) for there own name. When you enter 'Miele' (well known german company for washing machines and other stuff). You get an ad for miele.de but miele.de is already the first hit. The same when googling 'Samsung S9'.

I would be astounded if revenue dropped at all if Google stopped providing targeted ads, at least in the short term, because there is no serious competition. Longer term it may well be an advantage as you think, but I don't think it is a sure thing. Unless targeting gets an awful lot smarter, I think companies may choose to spend their dollars on cheaper, dumber services that provide the same conversion rates. Especially if privacy concerns and law changes enable a lot more people to drop out of the targeting.

If the unknowns have a greater influence. So, a little knowledge makes google overconfident.

Half the money I spend on advertising is wasted; the trouble is I don't know which half.

>>> User1 has a 10th wedding anniversary coming up. User2 is single and goes to the gym every day. User3 just got a promotion and likes the BMW 3 series. User4 is overweight but has just gone on a diet.

There, no identities and you can still advertise. Is it hard?

Perhaps the question is whether that turns to sales and engagement, or if people who don't click ads don't click ads regardless of content, and those that do, do.

The vast majority of ads I see are of no relevance to me. It makes me assume that the companies that would have relevant products just aren't spending as much as broader interest stuff, bringing us right back to square one.

If only one could 'scan' the product they just bought to no longer see any ads about it...

Often I see targeted ads for products I looked up and did some research on. I either bought the product or decided not to buy the product. Unless the ads convince me to buy another one or change my mind about whether or not I need a first one, they're not going to do any good.

Targeted ads can also be creepy and offputting. It's weird to look something up and then see ads for that specific version of that specific product everywhere.

Our team's recommender algorithms were marginally effective. Despite having significant data on our customers.

The new new thing is "personalization". We'll see if that helps.

If I had to guess, broadcast and individual targeting both (can) work, but the middle ground is fool's gold.

IMHO, I'd like to see renewed emphasis on improving search, browse, foraging. I generally know what I want, but still struggle to find it.

But they'll also still be expensive because everyone else also thinks the data is super-targeted so they throw a lot of money at it, increasing bids.

Really valuable - but is it best to be done when they're reading something completely irrelevant? If they're reading a piece on the latest political issue of the day, are they really in the mindset to explore that purchase?

I have used Twitter for a decade. They know where I live, and that I’ve never once indicated that I know Spanish or Japanese or whatever, yet somewhow their targeting setup seems to be “you are in Japan therefore here are Japanese adverts” which is fucking baffling for a company with that amount of engineering resource.

They've been wrong 100% of the time.

Maybe I'm just weird.