Personalized ads are more effective than non-personalized ads, to try to argue that personalized ads are ineffective is incorrect and the "without any proof" claim is absurd seeing the amount of specific data they collect on effectiveness measures. I used to work for ad tech companies and while that led me to hate them more than most people I'm not gonna say the data isn't their supporting their effectiveness.
Edit: I'm not familiar with data on context based ads but I'm very skeptical they are significantly better in the general case. They are already used in things where it makes sense like when you're searching for something.
I don’t have compelling evidence either way. But, I’d be a little skeptical of the data collected by the ad company. They are specifically and organization who’s entire skill-set includes convincing people to pay more, and that they might need some new service. I mean it isn’t some dirty secret, it is exactly what their value proposition is.
The internal data you were viewing and the metrics they track are, in part, to show people and convince them to buy the ad service. That’s like pure uncut ad-guy ad-material.
There is no grand conspiracy, the ad industry is massive, and advertising works. Companies would find out pretty quickly that advertising is a waste of money, yet here we are decades later and they still ad spend like crazy.
It definitely works, and the more tailored the ads, the better they work.
The key is remembering we are talking about average people, not nerdy techno anarchists with router level ad blocking and a pavolonian vomit reflex to seeing the word "sponsored".
It doesn’t require a grand conspiracy, just nobody deciding to rock the boat. It seems that when academics try to measure how effective the ads are, the effect sizes are much smaller than expected or it turns out the companies haven’t run any actual experiments.
> The key is remembering we are talking about average people, not nerdy techno anarchists with router level ad blocking and a pavolonian vomit reflex to seeing the word "sponsored".
Sure, dump everyone who is skeptical of ads into this niche weird person case, and it makes it easier to ignore them. Have you actually talked to these “average people,” though? My experience has been that most people just find ads annoying.
Having worked for these companies some of the data is murky (e.g. did these ads they saw earlier lead to them buying the product later, perfect attribution is obviously impossible) but a lot of it is unambiguous where they tracked people straight from clicking on the ad directly to a purchase. People have their conspiracies but I've seen it in black and white, it's very very clear. The only way I could see the data not being clear is in the case of outright fraud, which I'm fairly certain wasn't happening within our own metrics (as it would not only lead to legal liability but even more importantly fuck up the machine learning models).
Edit: to be clear I would believe the effect of ads is overstated, it's just the idea they are ineffective is wrong and people claiming that you can get more effective ads without tracking people at all doesn't seem plausible based on what I know of the industry. I could see contextual ads working in niche use cases (which again we already see when searching for products. YouTubers have relevant sponsors all the time. We even have affiliate marketing, where it's not only contextual but part of the content).
Tracking somebody from clicking on the ad, through to the purchase doesn’t prove that the ad added any value, though. The ad only added value if the person wouldn’t have otherwise found the product.
Look, none of these criticisms are novel or unknown in the industry and it's well known that measuring the exact impact of ads is impractical but we're not talking exact numbers here, just whether personalized ads are generally effective or not. This criticism effects how effective it is but not the fact that it's generally effective, unless you think all ads are doing is shifting sales forward in time (which doesn't really make sense to me).
I think this is more of an effect for things like search or ads on an e-commerce platform (somewhat ironically the contextual ads people here are advocating for are much more susceptible to this) but less so for a lot of the more random ads, especially for niche products.
Edit: For me they are obviously effective. I think the more interesting question is exactly what the return on ad spend generally is but that would take very specific data that I don't have access to.
Yeah I feel you on this. Anecdotal, but I’ve had plenty of Google searches that ended in what technically counts as ad conversions, but the exact same link is only three items below. The only difference for them is that I clicked on the ad version because it is near the top typically.
Keyword advertising for your company’s own name is a well known mistake: there’s no reason to serve ads to people who are already searching for you. If anything the high conversion rate on that sort of keyword is an argument against… using conversion rates as a metric, haha.
The "without any proof" part can be debunked even without the deep data, just looking at sales figures and conversion rates of personalized ads vs traditional "scatter-shot" approaches.
Who are these folks doing this "scatter-shot" approach? How do we get some insight into their practices?
The major company doing context sensitive advertising nowadays is Amazon. When you search on Amazon, they display relevant "sponsored" products that are clearly labeled as such.
So how is Amazon's "context sensitive" advertising business doing? By most accounts, pretty good actually.
The real problem in my opinion is the lack of competition to the "personalized" approach. Everyone (except Amazon) just accepts "personalized" as the default --- mainly because there is no credible, large scale, organized, generally available alternative to compare it to.
I don't think I want to argue againsr these ads on the basis that there's some alternative form of advertising that's more effective.
The problem is with data mining and tracking and nudging behavior. I want the things driving my behavior to be originating from my own intentions or from my preferred sources of inspiration (e.g. friends, family, media I'm most interested in consuming.)
You'll never be able to fully control the range of things that influence you, but you can still be intentional to a meaningful degree. For me that means supporting free and open source culture, and using subscription-based model rather than an ad-supported model for content. I'm not perfectly consistent but I am somewhat, and I think I'm operating from a coherent vision of what I believe my interests are, which is no small thing.
Amazon is not a good example of contextual ads, though. It doesn't generalize:
1. You can easily argue that these "context sensitive" ads are actually personalized ads: They're personalized based on the search query you just made! Amazon context ads are the same as Google/Apple App Store "context ads". Suppliers are paying for higher ranking.
2. It's a shopping website! Of course those context ads are going to have high ROI because they're showing an ad relevant to the thing you're shopping for!
When people talk about context ads, they mean "Why doesn't Facebook or the local newspaper use context ads?" They don't mean "Why doesn't Target put up a coupon for beans in the beans aisle?"
Not an apples to apples comparison really. Amazon owns the entire user journey on its platform, which the "ads" are an integral part of. They are not analogous to Google showing you ads in banners and searches for target pages it doesn't own, on platforms it doesn't own. If you want to compare Google to those who actually advertise with the scatter-shot approach, you compare them with traditional advertising providers - ad spaces on TV & radio channels, billboard companies etc. That'd be a fair comparison because Google is also essentially a seller of ad spaces it "rents" from other websites - just in this case those ad spaces can simultaneously show different advertisements for different clients to each user, based on that user's best-match profile. It's a no-brainer that Google's approach will yield more leads.
There are smaller examples too. The Register was one such example the last time I checked. They sell space on articles and also run Sponsored Content features.
>Personalized ads are more effective than non-personalized ads, to try to argue that personalized ads are ineffective is incorrect
I basically agree with this. I think because people don't like personalized ads, there's a temptation to argue they don't work.
But I think it's motivated reasoning in this case. And I actually think the argument against them is stronger when you acknowledge that they are more effective. The privacy issue goes hand in hand with the effect that ads collectively have to socialize people into consumer behaviors.
The entire podcast and youtube channel industry relies on contextual ads right?
Havent almost everyone including MKBHD said youtube ads doesnt give them enough to be used as the only revenue.
Contextual ads are more effective. You type shoes, you get shoes ads. It doesnt first need the shoe data and then later show shoe ads after you started searching for socks. And with no middlemen,more profitable. Duckduckgo employs this IIRC.
Behavioural ads are easy cos you are setting up an api. Contextual ads would mean you need a worthy product and having to handle your ad folks yourself. You cannot buy a domain and immediately start showing ads.
Behavioural ads breakeven because they sell your data. Not ads.
The whole reason why new media outlets moved to subscription model is bizarre to me. They could've just started doing it old school and it would have made news open and more privacy friendly.
In-video sponsors are a form of contextual ad but ads inserted by YouTube are personalized (that doesn't mean context is not also a factor).
Channels like MKBHD (and LTT) need more revenue than what they get from YouTube ads because their expenses have greatly increased, particularly staff.
You can't automate contextual ads in news media, otherwise you get airline ads next to stories about airplane crashes. Or travel ads for places experiencing natural disasters or political upheaval. People pairing ads with stories increases the labor costs and there's already not enough money being paid for actual journalism to increase the cost of having ads.
This is an interesting point you make. But didn't we solve all of these context issues already? I don't remember getting any ads like his in Duckduckgo since I've started using it. Nor do for the ads we used to get when everybody used contextual ads.
The only issue is going to be that you will have to handle this when you implement ads for your website/app. And each of them will have to do it.
Of course e.g. MKBHD wants more ad revenue. To do so his only option is to put additional contextual ads as part of the video itself, so he does. MKBHD has no way to make a section of the video target individual viewers based on their history. YouTube does, so they do - because they know it makes them more money to do it that way.
Yes. It makes more money for the middle-man. Neither the advertisers nor user gets enough value.
There are so many articles on why your FB or Google ads are not doing well. They show ads the way THEY can make money. Not value for you. This is theh same going when you use adwords.
The ad blockers are for behavioural ads. The api end-points which tracks you through tabs, listens to what you are saying through your phone's microphone.
Not contextual ads which you will setup for just your website / app. They are just <img> tags or equivalent. The entire reason why people use ad-blockers are because it is bad UX, anti-privacy and just sheer garbage amount of data gathering. Use a website with and without ad-blockers. You will see the difference. With middlemen comes problems for users.
Behavioral ads transfer revenue away from publishers and to spam sites and the ad platform (google).
Targeted ads are definitely better for the publisher, but hard to automate (the matchmaking between publishers and advertisers is less automated), but the percentage of ad spend that goes to the publisher is much higher, and the quality of each ad impression is higher.
There’s some win for targeting on the margins, where there’s no good place to buy ads.
Also, there’s an infinite inventory of targeted ad slots (like invisible windows displayed by malware or redirect spam), which could be better than display ads, where you might not be able to spend your marketing budget, at least in theory.
Edit: I'm not familiar with data on context based ads but I'm very skeptical they are significantly better in the general case. They are already used in things where it makes sense like when you're searching for something.