Does anybody know if targeted ads based on tracking even work? Are they worth all the extra cost and complexity compared to traditional ads? It doesn’t seem like it. Half the time I see a super-targeted ad, it’s for a product I already purchased.

Also, what ever happened to showing ads to people who aren’t already interested in your product to expand your brand and maybe bring in new customers? The current ad model feels overfitted to me.

Yes, yes they do work. That's why it's a billion dollar industry. It allows industries to micro-target specific ads for communities and speak to them directly. It gives you more ability to expand your brand to new customers, not less, as it allows to you to specifically target niche groups who previously had no interest in you. It's important to get a sense of the extreme level of refinement that firms have access to through data-driven marketing. Want to design a marketing campaign for dog-owning, outdoorsy lesbians? Subaru launched an ad campaign in the 90s using subtle coding in their wide-net ads. Now companies can do that much more effectively by directly targeting those communities.

As for the common complaint that you always see ads for products you already purchased, that's actually a very good time to make an impression. What are the odds that you are thinking about buying a new dishwasher at any given moment? Probably next to 0. You probably would completely ignore any dishwasher ad you saw. Now imagine you just replaced your dishwasher with a new one. You probably noticed that dishwasher ad now. You might have even clicked on it to see if you got a good deal on it. You probably care more right now about dishwasher specs than you ever have in your life up to this point. Maybe there's a better deal out there. This is the perfect time to send you more dishwasher ads.

Nothing of what you stated supports the theory that it works, you've only shown that people are willing to pay for the ability.

In the end I guess it boils down to whether you trust advertising companies to only help their customers sell stuff that nobody needs.

The last point is wrong: once I have purchased the dishwasher i care _less_ about a dishwasher at any previous point, because I am not most likely to have a functional dishwasher and even if yours is better I am not going to buy one more.

If marketers would only get this they would make so much more money, and I would get better ads for more relevant products. Instead I get ads that target me because I am in AGE_RANGE and live in country, or ads for scam products.

> because I am ... most likely to have a functional dishwasher and even if yours is better I am not going to buy one more.

Break the population into groups:

1. Have a working dishwasher / don't need one

2. Old dishwasher is failing, looking for a new one

3. Just bought a new dishwasher, it works great

4. Just bought a new dishwasher, going to return it

I suspect group 4 is who they're targeting.

>once I have purchased the dishwasher i care _less_ about a dishwasher [than] at any previous point //

Ever recommended something to a friend/relative, or bought a second one of something that works/fits/performs well? Or even ever thought you should. I've definitely bought a pair of trainers (sneakers) and then thought, oh I should have bought another pair. If the shop had sent me an email, "get a second pair postage free" a few weeks later then they'd probably have made a sale.

I know people who have second homes definitely would re-buy white-goods, for example.

I don't know that straightforward data on this will ever be forthcoming. And we can produce theoretical arguments every which way, not entirely unlike how classical philosophers were able to prove, through reason alone, that objects in nature tended to only travel in perfect circles and straight lines, and never shapes like ellipses and parabolae, and probably produce about the same volume of useful epistemological output in the process.

I'd think that the more interesting thing would be to try and find some proxies we can use as an ersatz empirical test. For example, what about ad prices? If personalization based on tracking really does work better than other forms of ad targeting, then one would expect that that difference would yield a noteworthy difference in ad prices.

In short: If it really works so well, then you'd expect personally targeted ads to cost significantly more per impression than ads that use content-based targeting. And I'd assume that that information is reasonably public.

Absolutely. A vegetarian/vegan restaurant being able to advertise exclusively to those people is one great example. In that example you're a new customer, but you've shown interest in similar products so you're much more likely of a customer(and better spend of advertising) than advertising to somebody on a carnivore diet.

It helps with being able to measure whether you're advertising to the right people. Traditional mass media advertising made a lot of money off of showing ads to completely irrelevant people. Targeted advertising makes even more money off of showing ads to mostly irrelevant people.

I'm surprised that most people don't understand this. It's not about being more effective. It's all about have sales or conversion attribution to be able to do stuff like A/B testing.

Nowadays cable TV targets ads too. If your diabetic grandma connects to your cable provided wifi router, you will get ads for glucose monitors and insulin pumps. Targeted advertising should be illegal.

The direct mail advertising industry was thriving in 1995. You could buy datasets of people by income, car ownership, shopping habits, etc even easier than you can now.