We probably don't. Empirically demonstrating the effectiveness of an advertising campaign is notoriously difficult and expensive and few people in the adtech industry have any interest in paying for that kind of research.
There was a point in an episode of the podcast You're Wrong About where one of the hosts made a comment about how much you can learn about people just by observing what kinds of things we don't require evidence to believe.
I think that the effectiveness of tracking-based advertising might be one of those things. It seems to be tied to an assumption that, where data is concerned, bigger is always better. Ironically, statisticians and philosophers of science have already known that this isn't actually true for about a century now.
Why do you assume that the statisticians and analysts who work in the adtech sphere are incompetent?
That they know the whole thing is a big lie, and contextual ads definitely work better, but they are either stupid or corrupt and forcing this concept down everyone's throats?
I don't. But I know from experience as someone who gets paid to do statistics that you work with the constraints that are set by your employer, and the person who gets to have the job is the person who's willing to do the job.
Oh they're definitely not incompetent, it's that "do the ads actually work and how well" doesn't matter all that much to the people running the auction house. What does matter is identifying situations where advertisers are willing to bid higher for a particular ad spot. If I'm an online record store and Google shows me a man in his early thirties wearing a band t shirt, who's been to multiple live concerts in the last 6 months, and who is in the 90th percentile of Spotify listeners that looks a juicy lead compared to an 18 year old woman whose only experience with live music was a 1975 concert.
But is it? Maybe? It could also be the case that our thirty year old is perfectly content with Spotify and has no interest in records, or it could be that he has been going to the same local shop for the last decade and can't be tempted away. And it could also be that our 18 year old actually loves the 1975 and thinks records are a vibe but isn't connected to her local music scene and is would love a way to browse and buy records that doesn't involve the social anxiety of going to in IRL store where she feels like an outsider. But you can't really capture this kind of data which is why we still get ads like "bought a vacuum -> you must be a vacuum enthusiast."
> "do the ads actually work and how well" doesn't matter all that much to the people running the auction house
The people running the ads are measuring the results in different ways. Whether or not they see measurable increases attributed to their ad spend is what keeps them coming back, or not.
Auction houses are inextricably linked to the results they are driving. It's most of the reason they get most of their business.
This is why companies like Google and Facebook control so much of the ad market. Not just because they are popular publishers, but because they have some of the most sophisticated ad delivery and measurement platforms.
I can transfer an ad click ID through to a sale and into my CRM, showing a 100% direct attribution. This attribution determines my likelihood to choose an "auction house" in the future.
It's kind of crazy the level of disrespect and outright dismissal the HN community gives to the ad world.
You are describing the idealist of ideal attribution situations which was possible with banner ads in the 90s and did not at all address the fundamental issue. The reason it's an auction house in the first place is because these ad networks have no clue in the slightest how valuable the ad space or any particular impression is and only care about user data collection inasmuch as it convinces publishers to bid up the price. Whether the ads actually work, the user data is useful to your targeting or deliver any value at all is up to you to figure out.
These ad networks own the market because they're the only suppliers of digital ad space with enough billboards to meet the massive demand; that's no small feat. And yes, they do have very far reaching tendrils to determine attribution which is also technically impressive but that's also not without a sea of muddy data especially when it comes to VTA.
Google and Meta aren't magic and you're giving them far far too much credit for what are really your own (or the agency you paid) successful ad campaigns because you did all the work.
> The reason it's an auction house in the first place is because these ad networks have no clue in the slightest how valuable the ad space or any particular impression is and only care about user data collection inasmuch as it convinces publishers to bid up the price.
You're completely clueless, just stop talking.
First, the "ad network" in this case and the "publisher" are the same thing. Second, nearly all of effective modern performance ad buys on the major publishers (Meta/Google) outsource most of the targeting to the publishers via tools like lookalike audiences (Meta) or dynamic search ads (Google), all of which are powered by user data. In fact, the publisher's loss of performance data via IDFA was the entire reason why losing IDFA was so bad for Meta. It wasn't that the buyers couldn't close the loop on who was converting (they could), it was that the publisher couldn't close the loop on who was converting (super important for targeting). The idea that buyers drive conversion and the publishers have not much to do with is wrong. It's blatantly obvious you have no direct experience in this area, and no idea what you're talking about.
Sources:
- Spent over $100m on performance ads
- Started and ran an ad attribution company
- Ran the team that built the Reddit ads platform and know how hard it is to make a performance ads platform work as well as Meta or Google (basically impossible)
lol, there are literally hundreds of billions of performance focused ad dollars a year chasing returns. If there is some hidden secret to running high-ROI ad campaigns without using the data that is locked up by Meta/Google etc, it's a miracle nobody has figured it out yet.
(p.s. Citing 12 year old research on advertising makes you look clueless, most of ad targeting barely existed then and the performance was absolute shit compared to modern ad platforms)
See, for example: https://dl.acm.org/doi/abs/10.1145/1963405.1963431
There was a point in an episode of the podcast You're Wrong About where one of the hosts made a comment about how much you can learn about people just by observing what kinds of things we don't require evidence to believe.
I think that the effectiveness of tracking-based advertising might be one of those things. It seems to be tied to an assumption that, where data is concerned, bigger is always better. Ironically, statisticians and philosophers of science have already known that this isn't actually true for about a century now.