4 ms·

This entire thesis hinges on targeted advertising being effective. There is a growing group of people who are increasingly doubtful of this [0]. I personally h

This entire thesis hinges on targeted advertising being effective. There is a growing group of people who are increasingly doubtful of this [0]. I personally have worked as a data scientist trying to assess the value generated by various advertising campaigns, and I personally found that the field is rife with egregious statistical misuse, usually because it was necessary to prove significant ROI on advertising. [0] https://thecorrespondent.com/100/the-new-dot-com-bubble-is-here-its-called-online-advertising/13228924500-22d5fd24 https://thecorrespondent.com/100/the-new-dot-com-bubble-is-h...

bluGill 6y ago
Which targeted is ineffective. Keywords can be useless while the whole field can be useful. Which is to say noting that I'm searching for C++ and so advertising your compiler or programing class is useless - I'm already a programmer (I just forgot the exact spelling or order of arguments to the thing I need) and my company has chosen my compiler. However if you know I my hobby you can target me with your new drill bit and be better yet. Though the largest advertisers don't care. Coke doesn't care that I don't like soda, they still want to target me just in case I'm called to bring drinks to some event. Ford can safely assume all Americans own a car and be close enough to right. Likewise everyone uses toilet paper (bidet users can be ignored) and soap (if you don't use soap you should be the highest target, though the ads perhaps should be different from those who use soap)
- godelski 6y ago
  Coke doesn't advertise to get you to buy the product. In many situations you don't actually have a choice (restaurant, theatre, etc). They advertise to make you feel a certain way about the brand. Car companies do something similar. They aren't advertising to get you to buy the car but associate a certain prestige with the vehicle which in turn makes their actual target buy their car. I'm not sure why everyone thinks ads are strictly about buying things. There's political ads, religious ads, pubic service announcements, etc. Ads are versatile.
  - bluGill 6y ago
    Thanks for making my point better than I did.
- jariel 6y ago
  The 'C++' keyword however, already narrows the target down to the 0.2% of the population, i.e. C++ devs. making it 500x more effective than a non-targeted ad. If only 10% of those typing C++ would ever be interested in a course, then those are not bad numbers. More nuanced: at the 'non-targeted' threshold the ad would not make sense at all, total inefficiency. At the targeted threshold of being able to target at least C++ devs, the ad probably starts to work. That is the difference between a viable business and not That means engagement, value creation, sales, C++ developers trained and ready for the market. This is extremely good for society. We definitely want aspiring C++ devs hooked up with quality courses. This anecdote very tangibly demonstrates the effectiveness of targeting for individual companies ... but it also points to the market efficiency that comes along with good advertising. If you have a startup, and you can't reach any of your audience, you're dead. This notion of 'word of mouth' is ridiculous as a business plan, it's exceedingly rare, and usually it's not that anyhow in reality - it's usually a form of effective social marketing by the early movers. Clubhouse for example is being helped by the 'celebrity' of the VCs behind it - they don't have mass market following, but a very avid following in a certain niche that will come onto the platform. I'm noticing a lot of Marc Andreseen on Clubhouse, too much for a busy VC, but not too much for someone who's hyping his own investment and bringing in a lot of viewers, helping out a lot of panels. The essential nature of basic targeting is not controversial, it's quite obvious at least at the most crude level.
jariel 6y ago
"This entire thesis hinges on targeted advertising being effective. " Targeted ads are unequivocally more effective than non targeted ads, on the aggregate - there is no dispute other than at the margins. Do you think that advertising makeup to the general population has the same effectiveness than advertising it to women? Or women who have shown an interest in makeup? Ads are complicated and nuanced, but everyone in the industry already knows this. There will always be science at the margins as we discover the means by which people truly engage, but otherwise, there is no arguing with core demographic targeting. It would be like completely non-technical people saying "Javascript is completely ineffective because of null ambiguity" whereas it's universally used, and the limitations of JS are recognized to all but the most junior developers.
- tomnipotent 6y ago
  The general consensus is that personalized advertising, specifically, is no better than contextual advertising that came before it. Everyone was fine and happy with contextual advertising.
  - jariel 6y ago
    Yes, this is true, but I suggest it's because despite the privacy invasion, we don't know that much about individuals. We don't need to though. Literally just basic demographics, plus a few interest points ... that's all I think we need to do well of those could be reliable.
    - cm2012 6y ago
      This is incredibly not true. For one, the best data point FB has on predicting personal likelihood to click an ad is which ads that person has clicked in the past.
      - jariel 6y ago
        I don't know what you are objecting to i.e. Social Media surprisingly not knowing a lot about us - or - the need only for relatively small amount of solid demo data. On the first point - they just don't know that much about us. That's what the whole post is about (!) They were lying advertisers ad to the probabilistic estimates of who we are. Go here: https://adssettings.google.com/authenticated?hl=en https://adssettings.google.com/authenticated?hl=en Google will show what they know about you if you have that on. It's not very accurate at all. It's very fuzzy. For me they have wrong age, a lot of wrong interests. So they surprisingly don't know a lot, just because someone clicks on an add ... that is not a lot of information - there's so many reasons we click on ads. Google thinks I'm into cars and I don't even drive, I hate cars! As for basic targeting - it's mostly what advertisers need. So age, education, income, zip code (or zip code classification), rough location would be great. Plus just a few, solid interests (real ones, not estimated) - and that would be a big step forward for ads. So yes - our privacy is fairly aggressively invaded at the same time they can only glean so much from it. Seems like a Paradox.
com2kid 6y ago
> There is a growing group of people who are increasingly doubtful of this Most targeted advertising, sure. Targeted ads based on interests? Diet? General age range? Of course those will be more effective. Showing me targeted ads for fitness products has been very effective, I have a yoga and a workout app I use that are both 100% due to targeted ads. I also have tried out various food products (keto cereal!) due to targeted ads. Interest based ads work, if done well. Now that said, ads for stuff I already bought, eh, less so. Even worse if I buy something from an ad, get it home, unbox it, then google for help. Now I get ads trying to sell me the same product again. Though I'm not sure how horribly invasive ads would have to be to avoid that scenario. I've generally found FB ads to be, honestly, useful at times. What I really don't trust are kickstarter campaigns running YT ads. Shouts 'scam!' to me almost instantly. Also that YT ad that was playing a few months ago for the $20 adjustable dumbbell set that went to a super shady site. (Normally adjustable dumbbells go for $500+...) It appeared to be a fake storefront, but YT kept happily running the ad day after day. It was well targeted though. :)
- Red_Leaves_Flyy 6y ago
  You're following trends. Of course targeted advertising works on people that chase the Jones's.

bluGill 6y ago

Which targeted is ineffective. Keywords can be useless while the whole field can be useful. Which is to say noting that I'm searching for C++ and so advertising your compiler or programing class is useless - I'm already a programmer (I just forgot the exact spelling or order of arguments to the thing I need) and my company has chosen my compiler. However if you know I my hobby you can target me with your new drill bit and be better yet. Though the largest advertisers don't care. Coke doesn't care that I don't like soda, they still want to target me just in case I'm called to bring drinks to some event. Ford can safely assume all Americans own a car and be close enough to right. Likewise everyone uses toilet paper (bidet users can be ignored) and soap (if you don't use soap you should be the highest target, though the ads perhaps should be different from those who use soap)

godelski 6y ago
Coke doesn't advertise to get you to buy the product. In many situations you don't actually have a choice (restaurant, theatre, etc). They advertise to make you feel a certain way about the brand. Car companies do something similar. They aren't advertising to get you to buy the car but associate a certain prestige with the vehicle which in turn makes their actual target buy their car. I'm not sure why everyone thinks ads are strictly about buying things. There's political ads, religious ads, pubic service announcements, etc. Ads are versatile.
- bluGill 6y ago
  Thanks for making my point better than I did.
jariel 6y ago
The 'C++' keyword however, already narrows the target down to the 0.2% of the population, i.e. C++ devs. making it 500x more effective than a non-targeted ad. If only 10% of those typing C++ would ever be interested in a course, then those are not bad numbers. More nuanced: at the 'non-targeted' threshold the ad would not make sense at all, total inefficiency. At the targeted threshold of being able to target at least C++ devs, the ad probably starts to work. That is the difference between a viable business and not That means engagement, value creation, sales, C++ developers trained and ready for the market. This is extremely good for society. We definitely want aspiring C++ devs hooked up with quality courses. This anecdote very tangibly demonstrates the effectiveness of targeting for individual companies ... but it also points to the market efficiency that comes along with good advertising. If you have a startup, and you can't reach any of your audience, you're dead. This notion of 'word of mouth' is ridiculous as a business plan, it's exceedingly rare, and usually it's not that anyhow in reality - it's usually a form of effective social marketing by the early movers. Clubhouse for example is being helped by the 'celebrity' of the VCs behind it - they don't have mass market following, but a very avid following in a certain niche that will come onto the platform. I'm noticing a lot of Marc Andreseen on Clubhouse, too much for a busy VC, but not too much for someone who's hyping his own investment and bringing in a lot of viewers, helping out a lot of panels. The essential nature of basic targeting is not controversial, it's quite obvious at least at the most crude level.

jariel 6y ago

"This entire thesis hinges on targeted advertising being effective. " Targeted ads are unequivocally more effective than non targeted ads, on the aggregate - there is no dispute other than at the margins. Do you think that advertising makeup to the general population has the same effectiveness than advertising it to women? Or women who have shown an interest in makeup? Ads are complicated and nuanced, but everyone in the industry already knows this. There will always be science at the margins as we discover the means by which people truly engage, but otherwise, there is no arguing with core demographic targeting. It would be like completely non-technical people saying "Javascript is completely ineffective because of null ambiguity" whereas it's universally used, and the limitations of JS are recognized to all but the most junior developers.

tomnipotent 6y ago
The general consensus is that personalized advertising, specifically, is no better than contextual advertising that came before it. Everyone was fine and happy with contextual advertising.
- jariel 6y ago
  Yes, this is true, but I suggest it's because despite the privacy invasion, we don't know that much about individuals. We don't need to though. Literally just basic demographics, plus a few interest points ... that's all I think we need to do well of those could be reliable.
  - cm2012 6y ago
    This is incredibly not true. For one, the best data point FB has on predicting personal likelihood to click an ad is which ads that person has clicked in the past.
    - jariel 6y ago
      I don't know what you are objecting to i.e. Social Media surprisingly not knowing a lot about us - or - the need only for relatively small amount of solid demo data. On the first point - they just don't know that much about us. That's what the whole post is about (!) They were lying advertisers ad to the probabilistic estimates of who we are. Go here: https://adssettings.google.com/authenticated?hl=en https://adssettings.google.com/authenticated?hl=en Google will show what they know about you if you have that on. It's not very accurate at all. It's very fuzzy. For me they have wrong age, a lot of wrong interests. So they surprisingly don't know a lot, just because someone clicks on an add ... that is not a lot of information - there's so many reasons we click on ads. Google thinks I'm into cars and I don't even drive, I hate cars! As for basic targeting - it's mostly what advertisers need. So age, education, income, zip code (or zip code classification), rough location would be great. Plus just a few, solid interests (real ones, not estimated) - and that would be a big step forward for ads. So yes - our privacy is fairly aggressively invaded at the same time they can only glean so much from it. Seems like a Paradox.

com2kid 6y ago

> There is a growing group of people who are increasingly doubtful of this Most targeted advertising, sure. Targeted ads based on interests? Diet? General age range? Of course those will be more effective. Showing me targeted ads for fitness products has been very effective, I have a yoga and a workout app I use that are both 100% due to targeted ads. I also have tried out various food products (keto cereal!) due to targeted ads. Interest based ads work, if done well. Now that said, ads for stuff I already bought, eh, less so. Even worse if I buy something from an ad, get it home, unbox it, then google for help. Now I get ads trying to sell me the same product again. Though I'm not sure how horribly invasive ads would have to be to avoid that scenario. I've generally found FB ads to be, honestly, useful at times. What I really don't trust are kickstarter campaigns running YT ads. Shouts 'scam!' to me almost instantly. Also that YT ad that was playing a few months ago for the $20 adjustable dumbbell set that went to a super shady site. (Normally adjustable dumbbells go for $500+...) It appeared to be a fake storefront, but YT kept happily running the ad day after day. It was well targeted though. :)

Red_Leaves_Flyy 6y ago
You're following trends. Of course targeted advertising works on people that chase the Jones's.