Online Advertising

Highlights

• 2023-12-05 15:34 And so if you were watching ITV or later, Channel Four, you would watch the same commercials. And they were designed for millions and millions of people to be watching the same thing at the same time. And they were crafted really beautifully by amazing musicians. And you’d have filmmakers, I mean, most notably, I suppose, Ridley Scott, the filmmaker, began in advertising. He did that famous hovis vorjack new World Symphony bread advert.

• 2023-12-04 15:04 It’s Tim Huang, who is the author of a really terrific book called Subprime Attention Crisis.

• 2023-12-05 15:35 Is it the financial engine of the Internet? Yeah. So it’s actually really fascinating. You divided history just there into Bi and AI, right? Before Internet and after Internet. What’s really interesting is that there’s also a period before advertising on the Internet, and after advertising on the Internet we forget it now, but there was a period of time when people said, the Internet has no real business model. And we had no idea that advertising really would be such a core fundamental financial engine for the Web. I remember that. I remember the very first time I went online, I think it was like CompuServe. Yeah, that’s right. I don’t remember any adverts at all. It was just that’s right. It was a subscription, was the original model for the Web. In some ways, advertising had kind of popped up on the web. But it really wasn’t until the invention of something that’s known as programmatic advertising, which really Google innovated in the early 2000s, that advertising became such a powerful force on the web and is really ad. Technology is the way in which the real big giants of the internet are funded. Right. You look at Google, you look at Facebook, they are majority funded through advertising. And this has actually happened later in the Internet’s history. Right.

• 2023-12-05 15:36 I remember it must have been sort of mid ninety s, am I right in thinking it was at and T had the first correct banner ad? That’s right. So the first yeah, banner ad, as far as we can tell right. Is this ad that ran in the mid ninety s at and T was running it on Wired, the Wired magazine’s website, as the first real kind of banner ad on the web. What’s interesting about that though, is it was very much kind of a traditional sort of ad, right? Like there was an advertising guy that worked with Wired to put up the banner. It was almost like a billboard on the internet. Right. And I think that’s worth remembering is that look, this is important historically in retrospect, but when it happened was this totally minor part of At T’s campaign called You Will. They were really focused on these very slick TV commercials. If you Google it and go onto YouTube, you can see those At T ads. And the You Will, as far as I can tell, was basically here’s all the amazing things the internet’s going to do for you in the future. Things like, you know, and all these things that don’t quite exist. Video phones, but video phones in phone boxes rather than on cell phones. That’s right. And essentially what happened is that At T paid Wired to put a banner up on their website. It was kind of this hilarious ad. It was just sort of this pixelated banner with the slogan for the ad encouraging people to click on the ad. I am looking at that very advert now and it’s basically just a banner. And it says have you ever clicked your mouse right here? And there’s an arrow and it says you will. That’s right. And you clicked on it in a remarkable rate. Right. So the rates of click through on that banner ad were something around the order of 44%. Right. Was almost one in two people clicked on the ad. And was that because it was a novelty? Because we weren’t used to seeing this thing and it said Please click on me. So everyone was like, oh crap. That’s right. Yeah. I mean, I think that’s the really fascinating thing is so if you know anything about advertising stats nowadays, right, that kind of banner ad is lucky to get 0.4% 0.2% click through rates. We’ve seen a hundred fold decline in the effectiveness of these ads really in only a few decades.

• 2023-12-05 15:37 It looks a little bit more like the stock market. Effectively, when you go click on a website, there is a split section auction that takes place where algorithms bid for. Your attention wait, hang on. Depending on which one pays the most. Hang on. So there’s no humans involved, is that right? Well, there’s humans involved in terms of setting up the algorithm, determining pricing, figuring out what’s going to appear, but the actual decision of which ad is displayed and who pays to get that ad. It’s by and large, automated by this point. So what do you mean by auction? So there’s a system in the industry that’s known as real time bidding. And the idea of this is Dallas. You would go to a website and say, I’m on the Facebook newsfeed, right, and they’re about to deliver you an ad. What happens at that point is that there’s a signal that goes out to a marketplace that says, who wants to give an ad to Dallas? And many, many people, businesses, brands, advertising agencies will compete through this marketplace to deliver the ad to you. And it effectively is delivered by auction. Whoever pays the most is effectively the one who gets the right to deliver that ad. So it’s like a mini auction going. On billions of times a day. Every time an ad is crikey, it’s an incredible engineering feat. We kind of forget about it because it’s like infrastructure. It’s almost like electric. Well, that’s the thing. It’s totally invisible. And that idea of an auction, of course, I immediately think of some guy with a hammer and lots of people waving. Waving? No, it’s just a server with a light that blinks and then the ad appears on your newsfeeds. So this is some code. This is someone that has coded this, someone’s written an algorithm. It’s by and large, software and extremely data driven is really where most of the advertising happens today. It’s not the guys with ponytails.

• 2023-12-05 15:38 Like, you normally think about Silicon Valley, you think about nerds and garages. A lot of people involved were former Wall Street people, economists, right. So Hal Varian, who is an economist at Google for many years, helped to develop a lot of the kind of mechanics of this auction system that would later kind of fund big parts of the internet. Did someone make a decision? Did someone say, okay, we don’t have advertising on the internet. I’ve got a really good idea. We’re going to set up this thing called programmatic advertising. And I’m just wondering kind of where the kind of line was crossed, or was it that first banner ad that at and T had? Or was it well, I think it was really the invention of a product that Google had called AdWords, which was again, kind of in the mid 2000s. So the way to think about this is you’re going to go look for a plumber. So you type in plumber on Google. What happens there is a little bit like we’ve talked about. There is a bidding process for plumbers to advertise to you when you go for that search term. And so AdWords refers to the infrastructure that facilitates that buying and selling of ads in search results. It’s very interesting.

• 2023-12-05 15:38 You look at the early days of Google in their investor decks, they say, we’re mostly going to make money through licensing our search algorithm and then maybe we’ll make a little bit of money on advertising on the side. And in fact, the co founders have a paper that they wrote while they were at Stanford where they said a search engine should never rely on advertising because the incentives would be simply too perverse. Right. But they found out that this business model made huge amounts of money, willy Wonka waterfalls amount of money. And I think it had a gravitational well that pulled the company increasingly in that direction over time. So I think in some ways while advertising was experimented with, it wasn’t really a predominant business model that people anticipated. It really just proved to have worked so successfully that it kind of exerted a force on the internet as a whole. That’s really interesting. So was it kind of an accident in a way or it wasn’t a kind of conscious decision. It just sort of happened as a bit of a byproduct. That’s right. It was something that they invested in. But I don’t think anyone who was involved understood just how successful it would be. Right. That it would turn a company like Google, which was essentially like this academic science project, to one of the most powerful companies in the world. I think that was unanticipated by many of the people who were involved. If it was so successful, you say the returns on an advert are vanishingly small, like, in terms of clicks, where does the wealth come from? Where does the money come from? Why is it so successful if no one’s clicking on ads? Everyone just skip ads.

• 2023-12-05 15:39 And one of the things that some researchers have discovered is that the causal effect of that ad is very unclear. So it actually turns out they’re so good at targeting you, Dallas, that you would have bought the product anyways had they not even delivered the ad to you. But what the advertiser sees is, Dallas is going to buy this product, I deliver an ad to him, and he buys the product. And I think there’s a sense that, oh, okay, the ad really worked. But when you actually do randomized control trials, it’s actually very unclear whether or. Not that’s well, that’s the thing. And very often, let’s say I buy a pair of shoes, or I click on it and I look at a pair of shoes and I like the shoes, and then I go and buy the shoes for about six months afterwards, I just get nothing but adverts for shoes. Yeah, exactly. So it’s just not that clever. It’s like, I bought the damn shoes. Like, why are you still trying to sell me? You should know I’ve got the shoes. That’s the last thing I’m going to buy. That’s right, yeah. My friend has this wonderful story of buying some numbers for the front of his house, and after purchasing it, the algorithm was like, oh, you want more numbers? And so he was advertised brass numbers for his house for months afterwards.

• 2023-12-05 15:40 And I think that’s also what we have here as well. Right. There’s some arguments and some studies that suggest that one out of every $3 spent in the online advertising ecosystem is lost to fraud. Right. It just doesn’t go to anyone. It goes to a bot or someone who’s paid to just click on an ad. You think of any other market that operates with that level at fraud, it’s hard to imagine that that sort of thing is something functional, in the very least that we’re okay with or should be okay with. Is this because we’ve kind of removed humans from the equation and everything now is just purely run by algorithms so we just don’t get to see it and it’s incredibly complicated and no one person really understands all the so yeah. That is part of know. So there’s this thing that they talk about often in advertising called brand safety.

• 2023-12-05 15:41 That’s the kind of thing I think would really bring about a cris in the marketplace, right. Which is like, what have we been spending money on all these years? And I think what it looks like know, ultimately huge turmoil in journalism and media, for sure, probably in the near term, a lot of consolidation. There are a lot of weaker companies that bigger companies like Google and Facebook would buy. Right. And then I think really a bigger question, I think about the accessibility of services. Right. You imagine a world where this is really the case. Would you pay a subscription to Google? If so, how much would that be? Right? And you imagine the number of people that would lose access to that kind of service. There’s some real equity questions there. If I invented the internet, I’d kind of think, I keep thinking of all these things I do. Like I would charge everyone 50 p or a dollar to send an email, and that way I wouldn’t get so many emails. And actually that used to be the case, actually. Early compute, very early confuserve. Yeah. You both paid to send and receive like postage. Yeah, I think let’s bring that back because everyone’s inboxes would suddenly be uncluttered and you wouldn’t get nonsense emails. It’s a terrible idea. I mean well, yes, I have this idea that because everything has been free on the internet for so long, we actually have not had to confront very serious questions about guaranteeing access to people. If there was a kind of collapse, an attention crisis collapse, would that be a way for people like yourself to sort of come in and sort of rebuild it from scratch and go, okay, the whole thing’s collapsed. We’re going to have to think about better ways of doing stuff and could we create a better internet as a result? Is there a happy ending?

• 2023-12-04 15:29 And it was just this board game whose jingle is just still in my head after 20 years. See, jingles work. See that? I don’t know. That kind of advertising kind of works.