3 ms·

> Obviously more targeted ads perform better. Do we honestly know if that's true in any meaningful way? It also depends on what you mean by targeted. If it's a

> Obviously more targeted ads perform better. Do we honestly know if that's true in any meaningful way? It also depends on what you mean by targeted. If it's advertising power tools to someone who's clearly searching things related to DIY, sure, make sense. If we mean: "Push this product to men between 30 and 40, who works in construction and have 2 kids" then I'd love to see an A/B test that suggested that a broader campaign isn't cheaper and equally effective. You're cutting of a lot of potential customers when you're targeting is to narrow and I don't think that's accounted for in general.

mumblemumble 3y ago
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. See, for example: https://dl.acm.org/doi/abs/10.1145/1963405.1963431 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.
- jklinger410 3y ago
  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?
  - mumblemumble 3y ago
    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.
  - Spivak 3y ago
    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."
    - jklinger410 3y ago
      > "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.
      - Spivak 3y ago
        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.
- jamiequint 3y ago
  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)
jklinger410 3y ago
> then I'd love to see an A/B test that suggested that a broader campaign isn't cheaper and equally effective Do you think no one has done this?
- mrweasel 3y ago
  I think a few large companies, say Coca Cola, have done it. What I doubt is if anyone has done it for small / medium sized companies, or for highly localized companies. My thinking is that any positive results they've seen from targeted ads are just a result of them running any time of advertising campaign and not specifically a targeted one. The targeted ads might also have been proved to be more effective, in at least some cases, for some companies, but are they significantly better, like 20, 50 or 100%? They might be 5% better, but is that enough to justify the amount of data collection?
  - jklinger410 3y ago
    > They might be 5% better, but is that enough to justify the amount of data collection? Now this is a nuanced take I am here for. I think if we put a value on personal data, much like a carbon tax, personalized ads would instantly become less profitable than contextual. In a heartbeat. No, the increase in efficiency is not worth the data collected. However with improvements in machine learning, behavioral ads will get even better (this is happening as I write this). And that value comparison may flip yet again.
chankstein38 3y ago
Especially when I block ads everywhere but my mom loves to screenshot things and send them to me lol it's entirely possible a 50-60s baker living in a different household can affect my spending habits as an engineer.
Aerroon 3y ago
Something you want to consider is language. If I open up Youtube without being logged in 6 of the 10 recommended videos at the top are in a language I cannot understand. I don't see how untargeted ads would be much different in that regard. Another factor to consider is location. Very few services/products that are advertised globally will be available where you actually are.
mocamoca 3y ago
There is a shelf-space issue. Let's assume most of high-quality ad-spaces are available for "broader campaigns". Once each small business broadcasts ads here, they will compete for space. Price of space-unit goes up. Then some ad-space-seller opens Pandora's box: "hey I can allow you to narrow the audience by location targeting. Each unit is more expensive but conversion rate is higher so you will end up with a better margin overall"

mumblemumble 3y ago

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. See, for example: https://dl.acm.org/doi/abs/10.1145/1963405.1963431 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.

jklinger410 3y ago
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?
- mumblemumble 3y ago
  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.
- Spivak 3y ago
  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."
  - jklinger410 3y ago
    > "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.
    - Spivak 3y ago
      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.
jamiequint 3y ago
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)

jklinger410 3y ago

> then I'd love to see an A/B test that suggested that a broader campaign isn't cheaper and equally effective Do you think no one has done this?

mrweasel 3y ago
I think a few large companies, say Coca Cola, have done it. What I doubt is if anyone has done it for small / medium sized companies, or for highly localized companies. My thinking is that any positive results they've seen from targeted ads are just a result of them running any time of advertising campaign and not specifically a targeted one. The targeted ads might also have been proved to be more effective, in at least some cases, for some companies, but are they significantly better, like 20, 50 or 100%? They might be 5% better, but is that enough to justify the amount of data collection?
- jklinger410 3y ago
  > They might be 5% better, but is that enough to justify the amount of data collection? Now this is a nuanced take I am here for. I think if we put a value on personal data, much like a carbon tax, personalized ads would instantly become less profitable than contextual. In a heartbeat. No, the increase in efficiency is not worth the data collected. However with improvements in machine learning, behavioral ads will get even better (this is happening as I write this). And that value comparison may flip yet again.

chankstein38 3y ago

Especially when I block ads everywhere but my mom loves to screenshot things and send them to me lol it's entirely possible a 50-60s baker living in a different household can affect my spending habits as an engineer.

Aerroon 3y ago

Something you want to consider is language. If I open up Youtube without being logged in 6 of the 10 recommended videos at the top are in a language I cannot understand. I don't see how untargeted ads would be much different in that regard. Another factor to consider is location. Very few services/products that are advertised globally will be available where you actually are.

mocamoca 3y ago

There is a shelf-space issue. Let's assume most of high-quality ad-spaces are available for "broader campaigns". Once each small business broadcasts ads here, they will compete for space. Price of space-unit goes up. Then some ad-space-seller opens Pandora's box: "hey I can allow you to narrow the audience by location targeting. Each unit is more expensive but conversion rate is higher so you will end up with a better margin overall"