1 00:00:10,119 --> 00:00:13,520 Speaker 1: Lord, Welcome to another episode of The Odd Laws Podcast. 2 00:00:13,600 --> 00:00:17,800 Speaker 1: I'm Joe Wisenhall and I'm Tracy Alloway. Tracy, do you 3 00:00:17,960 --> 00:00:26,960 Speaker 1: know what powers the internet? Um? Hamsters on wheels? No, Um, electricity. 4 00:00:27,200 --> 00:00:31,319 Speaker 1: I don't know. Data. Is this like a metaphor? I 5 00:00:31,360 --> 00:00:34,120 Speaker 1: guess you could probably say. I have a lot of 6 00:00:34,159 --> 00:00:38,160 Speaker 1: different answers. Electricity is probably one. Someone could say data. 7 00:00:38,760 --> 00:00:43,720 Speaker 1: I was gonna say online advertising. Ah, now this is interesting. Um. 8 00:00:43,840 --> 00:00:46,840 Speaker 1: Do you mean in the sense that online advertising pays 9 00:00:46,880 --> 00:00:51,199 Speaker 1: for websites to actually run? Yeah. Basically, it's like if 10 00:00:51,240 --> 00:00:56,040 Speaker 1: you think about virtually anything we do on the internet, 11 00:00:56,520 --> 00:00:59,960 Speaker 1: if you think about at least some aspect of our 12 00:01:00,200 --> 00:01:03,720 Speaker 1: careers having written on the internet, if you think of 13 00:01:03,760 --> 00:01:07,679 Speaker 1: all the social networking, literally everything in some way, it 14 00:01:07,760 --> 00:01:12,479 Speaker 1: was probably paid for the online advertising. Yeah, I guess 15 00:01:12,480 --> 00:01:15,080 Speaker 1: you're right. I mean, none of these things are really 16 00:01:15,160 --> 00:01:19,360 Speaker 1: provided for for free. And I guess the cliche is 17 00:01:19,400 --> 00:01:22,240 Speaker 1: always if you're not paying for the service, Uh, then 18 00:01:22,440 --> 00:01:24,920 Speaker 1: you know the product they're selling is is probably you 19 00:01:25,080 --> 00:01:28,200 Speaker 1: and your personal data, and they're selling that to advertisers, 20 00:01:28,280 --> 00:01:31,120 Speaker 1: right exactly. And of course, as we know, a lot 21 00:01:31,160 --> 00:01:34,800 Speaker 1: of the practices and online advertising these days are pretty controversial, 22 00:01:35,560 --> 00:01:38,480 Speaker 1: and people are concerned about privacy and as you said, 23 00:01:38,520 --> 00:01:41,440 Speaker 1: your personal data and the reader or the user being 24 00:01:41,480 --> 00:01:46,120 Speaker 1: the product. But have you ever wondered, really how they 25 00:01:46,319 --> 00:01:49,040 Speaker 1: figure out what ad to serve you? In the split 26 00:01:49,080 --> 00:01:51,120 Speaker 1: second you go to a website and there's already an 27 00:01:51,120 --> 00:01:53,840 Speaker 1: AD that's perfectly tailored to your interest. Have you ever 28 00:01:53,880 --> 00:01:59,280 Speaker 1: really explored how that happens? I'm not lying when I 29 00:01:59,320 --> 00:02:02,760 Speaker 1: say I have actually wondered this specifically, I've wondered why 30 00:02:02,800 --> 00:02:04,960 Speaker 1: the pair of shoes that I looked at for like 31 00:02:05,280 --> 00:02:08,600 Speaker 1: five seconds two weeks ago follow me around the internet 32 00:02:08,680 --> 00:02:12,040 Speaker 1: from months later begging to be bought. That really annoys me. 33 00:02:12,120 --> 00:02:14,200 Speaker 1: So if this is what we're going to talk about, 34 00:02:14,480 --> 00:02:16,480 Speaker 1: then I'm on board, Joe. But what does it have 35 00:02:16,560 --> 00:02:20,160 Speaker 1: to do with markets exactly? Well, it's uh, first of all, 36 00:02:20,320 --> 00:02:22,280 Speaker 1: this is exactly what we're gonna be talking about, So 37 00:02:22,360 --> 00:02:24,760 Speaker 1: you're in luck. And of course it's a market thing 38 00:02:25,080 --> 00:02:29,560 Speaker 1: because someone has to buy and sell that ad inventory 39 00:02:29,600 --> 00:02:32,240 Speaker 1: and essentially you go to a website or go to 40 00:02:32,280 --> 00:02:35,760 Speaker 1: Facebook or whatever, and there has to be some sort 41 00:02:35,760 --> 00:02:40,840 Speaker 1: of process for allocating that advertising space and how much 42 00:02:40,880 --> 00:02:42,720 Speaker 1: are people going to pay for it? And how much 43 00:02:42,800 --> 00:02:45,320 Speaker 1: is that worth and how much are uh they is 44 00:02:45,360 --> 00:02:47,840 Speaker 1: the website going to sell it for? And that is 45 00:02:48,040 --> 00:02:52,120 Speaker 1: a marketplace. So the Internet is fundamentally run by market structure, 46 00:02:52,160 --> 00:02:56,000 Speaker 1: is what you're telling me. I love this episode already exactly. Well, 47 00:02:56,200 --> 00:02:59,240 Speaker 1: I'm very excited about it. So without further ado, I 48 00:02:59,240 --> 00:03:02,239 Speaker 1: want to bring in our guest. His name is Option Bdelli. 49 00:03:02,360 --> 00:03:06,919 Speaker 1: He is a systems engineer for online ad platforms, and 50 00:03:06,960 --> 00:03:10,519 Speaker 1: we are going to talk about market structure as it 51 00:03:10,680 --> 00:03:20,200 Speaker 1: pertained to the world of online advertising. Thank you very 52 00:03:20,240 --> 00:03:24,200 Speaker 1: much for coming on. Thank you, and hi Tracy Hija. 53 00:03:25,440 --> 00:03:29,240 Speaker 1: So what happens when I go to a website? Wow, 54 00:03:30,320 --> 00:03:33,040 Speaker 1: let's just let's just jump right in. Great, so today 55 00:03:33,520 --> 00:03:36,280 Speaker 1: on average for the average website, let's go through what happens. 56 00:03:37,080 --> 00:03:40,200 Speaker 1: So you load up your favorite web page, whatever it is, 57 00:03:40,240 --> 00:03:44,800 Speaker 1: let's say bloom Bloomberg dot com of course, and chances 58 00:03:44,840 --> 00:03:46,800 Speaker 1: are you visited Bloomberg dot com before. This is not 59 00:03:46,840 --> 00:03:49,840 Speaker 1: the first time you visited this site, so that's important. 60 00:03:49,880 --> 00:03:52,200 Speaker 1: We'll get back to that in a bit. What happens 61 00:03:52,720 --> 00:03:55,560 Speaker 1: the moment you load the page there is a race. 62 00:03:55,840 --> 00:03:58,320 Speaker 1: People want to find out who you are and whether 63 00:03:58,360 --> 00:04:00,880 Speaker 1: it's worth their time to show you an AD, And 64 00:04:01,000 --> 00:04:03,600 Speaker 1: the hope is if they show you an AD, maybe 65 00:04:03,680 --> 00:04:07,320 Speaker 1: you will in some way engage or convert or respond 66 00:04:07,360 --> 00:04:10,360 Speaker 1: to that ad in different ways, and there are different 67 00:04:10,520 --> 00:04:15,440 Speaker 1: methods for measuring exactly how you convert. So the whole 68 00:04:15,800 --> 00:04:18,320 Speaker 1: game at this point is to show you an AD 69 00:04:18,839 --> 00:04:23,240 Speaker 1: as quickly as possible, that is, as revenue impacting as 70 00:04:23,279 --> 00:04:26,599 Speaker 1: possible for both the person purchasing the ad and the 71 00:04:26,640 --> 00:04:30,040 Speaker 1: purchase the person selling you the ad, the publisher and 72 00:04:30,120 --> 00:04:33,120 Speaker 1: the advertiser. So I have a bunch of questions already 73 00:04:33,200 --> 00:04:36,520 Speaker 1: about how the buyers and the sellers kind of come together. 74 00:04:36,720 --> 00:04:39,440 Speaker 1: But I guess before we get to that, can you 75 00:04:39,560 --> 00:04:43,480 Speaker 1: describe what kind of personal data is available if Joe 76 00:04:43,560 --> 00:04:46,600 Speaker 1: or I, you know, actually click on a website, be 77 00:04:46,640 --> 00:04:51,200 Speaker 1: at the Bloomberg website or something else. Sure, so we 78 00:04:51,240 --> 00:04:55,040 Speaker 1: can divide this generally into three buckets of data. Let's 79 00:04:55,080 --> 00:04:57,520 Speaker 1: say there's the data that we know about the user 80 00:04:57,800 --> 00:05:01,000 Speaker 1: as as soon as they visit the page, the data 81 00:05:01,000 --> 00:05:03,560 Speaker 1: that is live that is part of the request. Then 82 00:05:03,600 --> 00:05:06,520 Speaker 1: there's the data the Bloomberg in this example has from 83 00:05:06,560 --> 00:05:11,600 Speaker 1: your past visits. And then there's the data that the 84 00:05:11,640 --> 00:05:16,200 Speaker 1: advertiser has purchased from a third party that has been 85 00:05:16,200 --> 00:05:19,599 Speaker 1: collected from you over the course of months and years, 86 00:05:19,600 --> 00:05:23,600 Speaker 1: from all sorts of different sources. The party that is 87 00:05:23,640 --> 00:05:25,520 Speaker 1: in charge of providing this last piet of data, which 88 00:05:25,520 --> 00:05:28,080 Speaker 1: I'm guessing is the most interesting to you, is called 89 00:05:28,320 --> 00:05:32,080 Speaker 1: a d MP, a data management platform. It's the job 90 00:05:32,080 --> 00:05:35,720 Speaker 1: of these DMPs to find out as much data as 91 00:05:36,000 --> 00:05:38,080 Speaker 1: they can possibly get about you, and aggregate from all 92 00:05:38,080 --> 00:05:41,000 Speaker 1: sorts of different databases and data sources, and correlate it 93 00:05:41,080 --> 00:05:44,240 Speaker 1: to the information that is returned to them by the publisher, 94 00:05:44,279 --> 00:05:47,560 Speaker 1: in this case, Bloomberg. So I have a bunch of 95 00:05:47,760 --> 00:05:53,359 Speaker 1: potentially different interests. I'm interested in financial market, I'm interested 96 00:05:53,640 --> 00:05:59,839 Speaker 1: in boxing, I'm interested in barbecuing. So in theory, there's 97 00:05:59,880 --> 00:06:03,240 Speaker 1: a lot of different types of advertisers, very different from 98 00:06:03,279 --> 00:06:07,440 Speaker 1: each other, that would might want to compete for my attention, 99 00:06:07,480 --> 00:06:10,000 Speaker 1: that might want to put up some product that they're 100 00:06:10,040 --> 00:06:13,440 Speaker 1: selling in front of me. So how what is the 101 00:06:13,480 --> 00:06:17,680 Speaker 1: process by which that collection of data that the d 102 00:06:17,880 --> 00:06:22,839 Speaker 1: MP has on me turns into someone putting an atom 103 00:06:22,880 --> 00:06:26,160 Speaker 1: from ah? So that is the job of the DSP, 104 00:06:26,360 --> 00:06:28,920 Speaker 1: the demand side platform. Are you enjoying the outfits? I 105 00:06:28,920 --> 00:06:30,600 Speaker 1: I am already I have a feel like there's more 106 00:06:30,640 --> 00:06:33,880 Speaker 1: to come to there are Okay, So the demand side 107 00:06:33,880 --> 00:06:37,680 Speaker 1: platform partners with several d MPs, and it's their job 108 00:06:37,760 --> 00:06:41,360 Speaker 1: to look at your incoming data, your page visit data 109 00:06:41,440 --> 00:06:43,560 Speaker 1: that happens when you visit bloomber dot com in this 110 00:06:43,600 --> 00:06:45,920 Speaker 1: one moment and correlated to all the different data it 111 00:06:45,960 --> 00:06:47,920 Speaker 1: gets from the d MPs. And how they go about 112 00:06:47,960 --> 00:06:50,640 Speaker 1: that is sort of proprietary. It's the special sauce. But 113 00:06:50,760 --> 00:06:54,280 Speaker 1: imagine they have just about any imaginal data stores that 114 00:06:54,320 --> 00:06:57,320 Speaker 1: they can possibly call on you based on your email 115 00:06:57,320 --> 00:06:59,239 Speaker 1: address and how who you've given to you in the past, 116 00:06:59,279 --> 00:07:01,760 Speaker 1: cookies that you've often websites in the past, even location 117 00:07:01,800 --> 00:07:04,920 Speaker 1: begins from from brick and mortar real life stories that 118 00:07:04,960 --> 00:07:07,560 Speaker 1: you visit in the past, and they're correlating it all 119 00:07:08,120 --> 00:07:10,720 Speaker 1: and that single moment while you're viewing the page in 120 00:07:10,800 --> 00:07:12,680 Speaker 1: order to figure out which ad is best to serve 121 00:07:12,680 --> 00:07:15,320 Speaker 1: to you. It's weird to think there's like a behind 122 00:07:15,360 --> 00:07:19,200 Speaker 1: the scenes battle happening every time you open up a 123 00:07:19,200 --> 00:07:22,720 Speaker 1: new web page. I'm wondering how how did this structure 124 00:07:22,880 --> 00:07:25,720 Speaker 1: actually come into place? Like who made the decision that 125 00:07:25,760 --> 00:07:27,880 Speaker 1: this was going to be the way um that the 126 00:07:27,920 --> 00:07:32,040 Speaker 1: Internet essentially works and that advertising is sold online. So 127 00:07:32,120 --> 00:07:34,400 Speaker 1: let's go through a brief history of how the market 128 00:07:34,440 --> 00:07:38,400 Speaker 1: micro structure of online advertising evolved. Those are words I 129 00:07:38,400 --> 00:07:40,600 Speaker 1: never thought I'd be saying before, but here we are. 130 00:07:41,120 --> 00:07:43,360 Speaker 1: So in the beginning that this is why they welcome 131 00:07:43,400 --> 00:07:48,600 Speaker 1: to the outlaws. So in the beginning there were advertisers 132 00:07:48,600 --> 00:07:52,440 Speaker 1: and publishers. So in this case, there's let's say Bloomberg 133 00:07:52,520 --> 00:07:56,440 Speaker 1: and someone who wants to advertise on Bloomberg, and around 134 00:07:56,600 --> 00:07:59,240 Speaker 1: let's say twenty years ago, was fairly easy for these 135 00:07:59,240 --> 00:08:01,440 Speaker 1: two parties to find each other and match up. There 136 00:08:01,440 --> 00:08:03,320 Speaker 1: were not that many people who were advertising, and they're 137 00:08:03,360 --> 00:08:06,960 Speaker 1: not many people publishing websites. But more and more advertisers 138 00:08:06,960 --> 00:08:10,360 Speaker 1: and more and more publishers makes this market microstructure really inefficient. 139 00:08:10,400 --> 00:08:13,880 Speaker 1: Imagine the equipment of an open outcry pit, but only 140 00:08:13,920 --> 00:08:17,080 Speaker 1: about five people in each pit, scattered all around the world. 141 00:08:17,440 --> 00:08:19,480 Speaker 1: How to buyers and sellers match up with each other 142 00:08:19,520 --> 00:08:22,800 Speaker 1: after they've reached a certain volume. The solution to that 143 00:08:23,000 --> 00:08:28,679 Speaker 1: first became advertising networks. This would be an aggregation of 144 00:08:29,000 --> 00:08:35,000 Speaker 1: advertisers under one umbrella being presented to publishers. That solves 145 00:08:35,040 --> 00:08:39,080 Speaker 1: one problem, how do you get more supply to satisfy demand, 146 00:08:39,679 --> 00:08:44,000 Speaker 1: but it exposes another problem. To give an example where 147 00:08:44,000 --> 00:08:46,440 Speaker 1: that problem is, let's put some color on what ad 148 00:08:46,440 --> 00:08:48,720 Speaker 1: we're serving on Bloomberg dot com. Let's say we're serving 149 00:08:48,720 --> 00:08:53,240 Speaker 1: an ad for men's shoes. So if I present that 150 00:08:53,280 --> 00:08:56,000 Speaker 1: add to Joe when he's being a boomer dot com, 151 00:08:56,040 --> 00:08:59,240 Speaker 1: there's probably a higher percentage of a chance that he's 152 00:08:59,240 --> 00:09:01,200 Speaker 1: going to convert actually engage with that ad. Then if 153 00:09:01,240 --> 00:09:03,240 Speaker 1: I were to show those same men shoes to Tracy, 154 00:09:03,480 --> 00:09:05,840 Speaker 1: who I'm guessing is not in the market for Adidas 155 00:09:06,000 --> 00:09:11,640 Speaker 1: right now, so we need to wait now. So we 156 00:09:11,679 --> 00:09:13,200 Speaker 1: need a way to solve that problem. We need a 157 00:09:13,200 --> 00:09:15,720 Speaker 1: way to actually show Joe the ads that he cares 158 00:09:15,760 --> 00:09:19,720 Speaker 1: about and Tracy the ads that she cares about. So 159 00:09:20,320 --> 00:09:24,960 Speaker 1: that led to add exchanges, and that in turn became 160 00:09:25,040 --> 00:09:32,360 Speaker 1: demand side platforms. The evolution how we went from ad 161 00:09:32,480 --> 00:09:35,920 Speaker 1: networks to add exchanges to demand side platforms basically came 162 00:09:35,960 --> 00:09:38,760 Speaker 1: from the recognition of this problem that although we have 163 00:09:38,760 --> 00:09:42,120 Speaker 1: the ability to match up advertising publishers, buyers, and sellers, 164 00:09:42,360 --> 00:09:46,080 Speaker 1: neither of them really know how to match up the 165 00:09:46,120 --> 00:09:48,800 Speaker 1: best inventory with each other. They know how to give 166 00:09:48,840 --> 00:09:51,640 Speaker 1: each other, you know, a certain lot size, a certain 167 00:09:51,720 --> 00:09:54,800 Speaker 1: amount of certain quantity in terms of number of ads 168 00:09:54,800 --> 00:09:56,680 Speaker 1: are in terms of revenue spent, but they don't really 169 00:09:56,679 --> 00:09:59,160 Speaker 1: know how to connect to Joe to give him the 170 00:09:59,200 --> 00:10:03,319 Speaker 1: best add that he wants. Tracy, I'm listening to off 171 00:10:03,520 --> 00:10:05,440 Speaker 1: and describe this. I've been thinking back to some of 172 00:10:05,480 --> 00:10:09,319 Speaker 1: our early episodes with Chris White talking about bond market 173 00:10:09,360 --> 00:10:15,320 Speaker 1: structure and just the sheer multitude of potential players in 174 00:10:15,320 --> 00:10:20,480 Speaker 1: the game and varieties of bonds that people could be selling, 175 00:10:20,679 --> 00:10:25,080 Speaker 1: different coupons, different times, different structures, and the sort of 176 00:10:25,320 --> 00:10:31,559 Speaker 1: complicated challenges that that poses in terms of bringing everyone together. Yeah. Absolutely, 177 00:10:31,640 --> 00:10:34,280 Speaker 1: I wonder in that case again, how they're able to 178 00:10:34,320 --> 00:10:37,160 Speaker 1: do it so quickly that we don't even notice that 179 00:10:37,240 --> 00:10:40,200 Speaker 1: it's actually happening in the background. So Offiction, what's the 180 00:10:40,240 --> 00:10:44,320 Speaker 1: technology that's actually allowing people to do this? So the 181 00:10:44,360 --> 00:10:46,920 Speaker 1: technology is referred to as you ready, Joe, another acronym 182 00:10:47,160 --> 00:10:52,000 Speaker 1: RTB Real time didn't and it's meant to happen as 183 00:10:52,080 --> 00:10:56,720 Speaker 1: quickly as possible to connect advertisers and publishers through supplies 184 00:10:56,760 --> 00:11:02,959 Speaker 1: lie platforms SSPs and demands like platforms DSPs. Why it's 185 00:11:03,000 --> 00:11:06,000 Speaker 1: so fast. Well, there's an interesting wrinkle or two there. 186 00:11:06,040 --> 00:11:09,640 Speaker 1: First off, the publisher always has the option of showing 187 00:11:09,679 --> 00:11:12,280 Speaker 1: you a house AD if they don't get a proper 188 00:11:12,360 --> 00:11:14,720 Speaker 1: real time bid that they can match open in time. 189 00:11:15,200 --> 00:11:17,080 Speaker 1: The reason you might want to show a house ad 190 00:11:17,160 --> 00:11:19,599 Speaker 1: versus a targeted AD is because you never want to 191 00:11:19,679 --> 00:11:21,400 Speaker 1: leave a user waiting for an ad load. I'm sure 192 00:11:21,440 --> 00:11:23,720 Speaker 1: you've all had the experience of visiting a page and 193 00:11:23,760 --> 00:11:25,360 Speaker 1: waiting for to load, and waiting for to load, and 194 00:11:25,360 --> 00:11:27,080 Speaker 1: when it does, you get a pop up or something 195 00:11:27,080 --> 00:11:32,319 Speaker 1: more innoxious. Generally you want to avoid that. What URTV 196 00:11:32,400 --> 00:11:37,640 Speaker 1: allows publishers to do is present the best possible targeted AD. 197 00:11:37,640 --> 00:11:39,240 Speaker 1: But if they can't do that in time, they will 198 00:11:39,280 --> 00:11:41,719 Speaker 1: just be happy to show you any ad. Maybe they 199 00:11:41,720 --> 00:11:44,040 Speaker 1: have their own personal brands they'd like to show off 200 00:11:44,080 --> 00:11:46,760 Speaker 1: to you. Maybe they have a preferred private partner who's 201 00:11:46,800 --> 00:11:48,880 Speaker 1: as they have an inventory that they would front and 202 00:11:48,920 --> 00:11:51,600 Speaker 1: run in front of their RTV ads. It really depends 203 00:11:51,600 --> 00:11:56,320 Speaker 1: on the publisher. So explain to us the auction process 204 00:11:56,400 --> 00:12:00,960 Speaker 1: real time bidding. It's essentially describing an auction, So to 205 00:12:01,000 --> 00:12:05,000 Speaker 1: walk us through what is that micro second or whatever 206 00:12:05,120 --> 00:12:09,400 Speaker 1: that or that add is delivered. How the bidding works? Sure? Okay, 207 00:12:09,440 --> 00:12:11,520 Speaker 1: so let's walk through our example again of Joe visiting 208 00:12:11,559 --> 00:12:16,160 Speaker 1: Bloomberg dot Com. Joe visits Bloomberg, Bloomberg sends Joe's data 209 00:12:16,240 --> 00:12:20,120 Speaker 1: to their SSP, their supply side platform. The SSP in 210 00:12:20,160 --> 00:12:23,800 Speaker 1: turn transforms this data into a bid and passes it 211 00:12:23,840 --> 00:12:27,680 Speaker 1: off to the ad exchanges. The ad exchanges in turn 212 00:12:28,040 --> 00:12:33,000 Speaker 1: pass this off to their d MPs. Are you following 213 00:12:33,000 --> 00:12:37,640 Speaker 1: the microxtra platforms? Sorry? I say d MPs and a 214 00:12:37,720 --> 00:12:40,280 Speaker 1: d MP DSPs Even I get confused. The d MPs 215 00:12:40,280 --> 00:12:42,079 Speaker 1: are the data management platforms. They're in charge of gain 216 00:12:42,160 --> 00:12:45,560 Speaker 1: the data. The DSPs are the demand side platforms, and 217 00:12:45,600 --> 00:12:47,640 Speaker 1: the demand side platforms in turn look for the best 218 00:12:47,679 --> 00:12:53,000 Speaker 1: inventory the bids that match up with the eyeballs that 219 00:12:53,040 --> 00:12:55,800 Speaker 1: they're getting from Bloomberg and try to make the best match. 220 00:12:55,960 --> 00:12:59,319 Speaker 1: It's the DSPs side too, or a job to service 221 00:12:59,360 --> 00:13:02,680 Speaker 1: sort of the match engine here. So this all goes 222 00:13:02,720 --> 00:13:07,800 Speaker 1: through three counterparties and end if you're counting. So I'm 223 00:13:07,840 --> 00:13:11,720 Speaker 1: curious what pricing actually looks like under this system, because 224 00:13:12,000 --> 00:13:15,880 Speaker 1: if I think back to old style advertising, um, and 225 00:13:15,960 --> 00:13:18,319 Speaker 1: you know, please let me engage in some media naval 226 00:13:18,360 --> 00:13:22,720 Speaker 1: gazing here. But if you were selling newspapers, um, for instance, 227 00:13:22,960 --> 00:13:26,360 Speaker 1: you would know roughly what the audience for your newspaper 228 00:13:26,400 --> 00:13:30,120 Speaker 1: circulation was, and you would have a decent idea of 229 00:13:30,240 --> 00:13:32,679 Speaker 1: you know their income levels and where they live, and 230 00:13:32,720 --> 00:13:34,880 Speaker 1: what they like to spend money on and things like that. 231 00:13:36,000 --> 00:13:40,440 Speaker 1: But now these online ads are so targeted, does that 232 00:13:40,640 --> 00:13:43,520 Speaker 1: drive down the pricing because you're not You're not really 233 00:13:43,800 --> 00:13:48,240 Speaker 1: targeting like a big swathe of the population that you're 234 00:13:48,280 --> 00:13:52,240 Speaker 1: hoping might buy your product. You're actually targeting a specific person. 235 00:13:53,120 --> 00:13:55,199 Speaker 1: It actually seems to drive up the pricing because you 236 00:13:55,240 --> 00:13:58,319 Speaker 1: don't really want to flood the market with ad volume 237 00:13:58,360 --> 00:14:00,319 Speaker 1: that most people won't engage, and you want to hit 238 00:14:00,640 --> 00:14:03,480 Speaker 1: the lowest number of people because remember there's a ton 239 00:14:03,520 --> 00:14:06,080 Speaker 1: of technical effort going behind all this. There are servers 240 00:14:06,120 --> 00:14:08,240 Speaker 1: that need to run and engineers that need to be paid, 241 00:14:08,240 --> 00:14:10,199 Speaker 1: and all that. You want to put the least amount 242 00:14:10,200 --> 00:14:12,760 Speaker 1: of effort into pushing your ads to the most the 243 00:14:12,760 --> 00:14:15,280 Speaker 1: people who are most likely to engage with them. You 244 00:14:15,400 --> 00:14:19,120 Speaker 1: never want to have any extra ad target hit that 245 00:14:19,200 --> 00:14:22,880 Speaker 1: hits someone who's unlikely to click it. Ideally, not think 246 00:14:22,920 --> 00:14:25,760 Speaker 1: about it like this. Let's say you have a thousand 247 00:14:25,840 --> 00:14:28,280 Speaker 1: ads that you want to push to people, you have 248 00:14:28,360 --> 00:14:30,280 Speaker 1: the choice. And this is how it used to work 249 00:14:30,280 --> 00:14:32,880 Speaker 1: about ten or fifteen years ago, of just agreeing to 250 00:14:32,920 --> 00:14:35,880 Speaker 1: a rate upfront with a publisher and saying these thousand 251 00:14:35,960 --> 00:14:38,200 Speaker 1: ads will sell for ten dollars and leaving the publisher 252 00:14:38,200 --> 00:14:40,520 Speaker 1: in charge of running those ads as fast as they 253 00:14:40,560 --> 00:14:43,800 Speaker 1: can sell. And some people will be engaging with those 254 00:14:43,840 --> 00:14:46,960 Speaker 1: ads Monday around two pm, some it's Saturday four am 255 00:14:47,000 --> 00:14:49,920 Speaker 1: in the morning. Some from the United States, some from Indonesia. 256 00:14:50,000 --> 00:14:52,720 Speaker 1: They'll all get the same rate. That's really not efficient 257 00:14:52,720 --> 00:14:54,520 Speaker 1: from an advertiser's point of view. They're really not getting 258 00:14:54,560 --> 00:14:57,640 Speaker 1: the most bang for their buck. So what they would 259 00:14:57,680 --> 00:14:59,920 Speaker 1: like to do instead is push to people who they 260 00:15:00,080 --> 00:15:03,520 Speaker 1: know as much information about and for whatever reason, they've 261 00:15:03,560 --> 00:15:07,720 Speaker 1: decided have the highest confidence and their ability to convert 262 00:15:08,040 --> 00:15:12,440 Speaker 1: with the add Now I'm curious about sort of you know, 263 00:15:12,480 --> 00:15:15,280 Speaker 1: if we really dive into the sort of financial market 264 00:15:15,680 --> 00:15:19,360 Speaker 1: aspect of this, are there players in the market that 265 00:15:19,440 --> 00:15:24,040 Speaker 1: attempt to for better lack of better warriors engage in 266 00:15:24,160 --> 00:15:28,400 Speaker 1: arbitrage buy up cheap, buy up inventory that they think 267 00:15:28,640 --> 00:15:30,680 Speaker 1: they could get on the cheap, and then sort of 268 00:15:30,720 --> 00:15:34,200 Speaker 1: repackage it in some way and resell up more expensive 269 00:15:34,280 --> 00:15:39,440 Speaker 1: like essentially, you know, essentially become traders. Absolutely, And there 270 00:15:39,440 --> 00:15:41,680 Speaker 1: are several different arbitraged strategies, just as you'll find in 271 00:15:41,680 --> 00:15:45,080 Speaker 1: any other market microstructure. There are plenty of companies who 272 00:15:45,080 --> 00:15:47,480 Speaker 1: are in the business of taking in raw ad supply 273 00:15:47,600 --> 00:15:50,720 Speaker 1: and sort of sharry picking what they think is the 274 00:15:50,720 --> 00:15:54,280 Speaker 1: best possible converting ad. And on the other side, there 275 00:15:54,320 --> 00:15:56,320 Speaker 1: are plenty of people who in the business of aggregating 276 00:15:56,400 --> 00:16:02,560 Speaker 1: eyeballs and only presenting the best possible eyeballs to vertisers. Fortunately, 277 00:16:02,600 --> 00:16:04,440 Speaker 1: I'm not sure I can give any of their names 278 00:16:04,680 --> 00:16:08,440 Speaker 1: on air. How profitable is that business? Then, like can 279 00:16:08,480 --> 00:16:10,680 Speaker 1: you make a ton of money out of it? Given 280 00:16:10,680 --> 00:16:14,200 Speaker 1: that the pricing is quite low, it is extraordinarily profitable 281 00:16:14,800 --> 00:16:18,240 Speaker 1: in the tens of billions of dollars a year. So 282 00:16:18,560 --> 00:16:22,360 Speaker 1: from time to time you mentioned secret sauce, I think 283 00:16:22,400 --> 00:16:24,680 Speaker 1: at one point and you talked about this, and you 284 00:16:24,760 --> 00:16:28,560 Speaker 1: hear about this world of companies called ad tech, and 285 00:16:28,640 --> 00:16:30,960 Speaker 1: I never am totally clear what they do. And a 286 00:16:31,040 --> 00:16:33,720 Speaker 1: lot of them seem to rise really fast and fall 287 00:16:33,880 --> 00:16:37,640 Speaker 1: really fast. So what are some of the competing strategies? 288 00:16:37,720 --> 00:16:40,640 Speaker 1: Some company comes along and says we have something new. 289 00:16:40,720 --> 00:16:42,840 Speaker 1: We have a new way that you can reach reach 290 00:16:42,840 --> 00:16:45,280 Speaker 1: your clients more officially, or we're going to use AI 291 00:16:45,440 --> 00:16:47,840 Speaker 1: or machine learning or big data or whatever. Like. What 292 00:16:47,920 --> 00:16:51,760 Speaker 1: are these ad tech companies? Where are they in the process, 293 00:16:51,800 --> 00:16:54,680 Speaker 1: and how do they attempt to make things more efficient? 294 00:16:55,320 --> 00:16:57,720 Speaker 1: So there are really two basic strategies for these at 295 00:16:57,800 --> 00:17:01,960 Speaker 1: tech companies. One is to make the process faster to 296 00:17:02,480 --> 00:17:06,280 Speaker 1: present adds faster to eyeballs that are willing to view 297 00:17:06,280 --> 00:17:10,600 Speaker 1: them and gain a sort of latency arbitrage edge. The 298 00:17:10,640 --> 00:17:14,080 Speaker 1: second strategy is to gather as much information as possible 299 00:17:14,119 --> 00:17:16,560 Speaker 1: from all these different accurate sources and as quickly as 300 00:17:16,560 --> 00:17:18,960 Speaker 1: possible match it up to the visitor that's trying to 301 00:17:19,040 --> 00:17:21,720 Speaker 1: view the ad. Most add to companies focus on the 302 00:17:21,760 --> 00:17:24,760 Speaker 1: second strategy. There in the business of gathering tremendous amounts 303 00:17:24,760 --> 00:17:28,199 Speaker 1: of data, correling it all very very quickly, and presenting 304 00:17:28,200 --> 00:17:31,680 Speaker 1: it to you quicker than the page will load. I'm 305 00:17:31,720 --> 00:17:34,760 Speaker 1: curious about the data gathering aspect. So let's say you 306 00:17:34,840 --> 00:17:38,439 Speaker 1: and I or the three of us here, we're like, okay, 307 00:17:38,480 --> 00:17:41,800 Speaker 1: we want to start a new ad tech company, and 308 00:17:42,600 --> 00:17:44,640 Speaker 1: where are we how do we start gathering that data, 309 00:17:44,720 --> 00:17:47,760 Speaker 1: like where are there wholesale brokers of it? Or that 310 00:17:47,920 --> 00:17:52,000 Speaker 1: can we collected ourselves? Like what's the process there? So 311 00:17:52,160 --> 00:17:55,560 Speaker 1: first we were probably partnered with several publishers like Bloomberg 312 00:17:55,640 --> 00:17:58,320 Speaker 1: for example, to go back to the beginning, we would 313 00:17:58,359 --> 00:18:00,720 Speaker 1: also partner with third party data providers. We would want 314 00:18:00,720 --> 00:18:03,840 Speaker 1: a combination of raw fresh data coming in from new 315 00:18:03,920 --> 00:18:07,080 Speaker 1: visits so we can build our own database, and we 316 00:18:07,200 --> 00:18:09,080 Speaker 1: want to have the ability to correlate it with as 317 00:18:09,160 --> 00:18:11,879 Speaker 1: many pre existing data sources. You probably don't want to 318 00:18:11,880 --> 00:18:13,960 Speaker 1: reinvent the wheel plenty. There are plenty of public data 319 00:18:14,000 --> 00:18:16,480 Speaker 1: basines and four paid databases out there where you can 320 00:18:16,480 --> 00:18:20,119 Speaker 1: find out, for example, from your email address or from 321 00:18:20,160 --> 00:18:23,080 Speaker 1: your IP address, what country you're in, what city you're in, 322 00:18:24,119 --> 00:18:26,800 Speaker 1: what sites you've signed up for the past, what your 323 00:18:26,920 --> 00:18:29,240 Speaker 1: LS you visited in the past, what time zone you're in, 324 00:18:29,280 --> 00:18:32,400 Speaker 1: what language you speak, all that sort of stuff. So 325 00:18:32,440 --> 00:18:35,159 Speaker 1: what's the most lucrative area when it comes to this 326 00:18:35,240 --> 00:18:38,760 Speaker 1: ecosystem of online ad selling is that the data collection? 327 00:18:39,280 --> 00:18:43,200 Speaker 1: Is it? You know, the underlying technology of the bidding system? 328 00:18:43,320 --> 00:18:46,720 Speaker 1: Is it if you're you know, arbitraging big blocks of 329 00:18:47,160 --> 00:18:52,680 Speaker 1: potential eyeballs to sell to people. What makes the most money. So, 330 00:18:53,119 --> 00:18:56,040 Speaker 1: judging from the market cap of the at two companies 331 00:18:56,080 --> 00:18:58,359 Speaker 1: out there today, I'm going to guess it's a combination 332 00:18:58,560 --> 00:19:03,879 Speaker 1: of selling the ad and also taking a bit of 333 00:19:04,200 --> 00:19:07,120 Speaker 1: the cut of every ad that's sold. So good example 334 00:19:07,320 --> 00:19:11,600 Speaker 1: here might be Google, which most publishers use as a 335 00:19:11,800 --> 00:19:16,359 Speaker 1: demand side partner. So Google gets their money two ways. 336 00:19:16,480 --> 00:19:19,960 Speaker 1: If you're using their product DFP double clip for publishers, 337 00:19:20,000 --> 00:19:23,800 Speaker 1: if you're connecting to Google through DFP, they get a 338 00:19:23,840 --> 00:19:27,520 Speaker 1: cut of every ad that is served through DFP, but 339 00:19:27,600 --> 00:19:29,800 Speaker 1: also they have the option of serving ads from their 340 00:19:29,840 --> 00:19:34,200 Speaker 1: own inventory through their exchange addicts. So for an analogy, 341 00:19:34,320 --> 00:19:38,800 Speaker 1: think if, for example, on the COMEX, you both ran 342 00:19:38,880 --> 00:19:41,919 Speaker 1: the exchange and also ran a gold mining company and 343 00:19:42,000 --> 00:19:44,440 Speaker 1: offered a gold etf on the exchange, you not only 344 00:19:44,480 --> 00:19:47,080 Speaker 1: get a cut of every trade, but you're sourcing inventory 345 00:19:47,119 --> 00:19:50,280 Speaker 1: that seems to be the most profitable. So Google not 346 00:19:50,480 --> 00:19:55,040 Speaker 1: only runs the exchange, they also are participated on the 347 00:19:55,080 --> 00:19:58,400 Speaker 1: exchange and are selling a raw commodity on that same exchange, 348 00:19:58,400 --> 00:20:00,800 Speaker 1: and so they make money multip both ways from that 349 00:20:00,880 --> 00:20:04,480 Speaker 1: trade from that environment exactly. In fact, they even it 350 00:20:04,520 --> 00:20:08,639 Speaker 1: sounds like Google's pretty good business, yeah, which I thought of? 351 00:20:08,680 --> 00:20:12,399 Speaker 1: That is Facebook similar in that respect, they have a 352 00:20:12,440 --> 00:20:15,240 Speaker 1: similar business. It's slightly different for social media, but similar. 353 00:20:15,920 --> 00:20:19,880 Speaker 1: So I'm curious about the matching of the right add 354 00:20:19,920 --> 00:20:23,520 Speaker 1: to my interest because maybe I'm interested in buying a 355 00:20:23,640 --> 00:20:26,399 Speaker 1: yacht or a new car, some very lucrative thing, or 356 00:20:26,440 --> 00:20:28,280 Speaker 1: if I clicked on the ad someone could make a 357 00:20:28,280 --> 00:20:30,919 Speaker 1: ton of money. But I might also be interested in 358 00:20:31,040 --> 00:20:33,840 Speaker 1: buying a book from Amazon that you know, it's probably 359 00:20:33,880 --> 00:20:37,640 Speaker 1: the margins are small. So how does it balance those 360 00:20:38,040 --> 00:20:41,760 Speaker 1: disparate potential outcomes to figure out which one is the 361 00:20:41,800 --> 00:20:44,040 Speaker 1: best to serve? Well, the best way to describe it 362 00:20:44,080 --> 00:20:47,199 Speaker 1: would be the free market. The DSP is connected to 363 00:20:47,200 --> 00:20:49,560 Speaker 1: several advertisers. Each one of them is willing to bid 364 00:20:49,600 --> 00:20:52,080 Speaker 1: on your impression. So it really comes down to is 365 00:20:52,240 --> 00:20:55,560 Speaker 1: the yacht seller advertiser in the bookseller advertiser? How much 366 00:20:55,560 --> 00:20:58,960 Speaker 1: do they both value your eyeballs? Maybe the yacht seller 367 00:20:59,080 --> 00:21:01,920 Speaker 1: thinks you're not really convert on an ad because he's 368 00:21:01,920 --> 00:21:03,840 Speaker 1: looked up your date and says, oh, well, you know 369 00:21:04,359 --> 00:21:06,159 Speaker 1: you live here in this time. I've never bought a 370 00:21:06,200 --> 00:21:08,920 Speaker 1: yacht before. Yeah, so it's say that you probably would 371 00:21:08,960 --> 00:21:10,720 Speaker 1: in the future, but you've probably bought plenty of books, 372 00:21:10,760 --> 00:21:13,280 Speaker 1: so maybe the yacht seller is not going to fight. 373 00:21:13,359 --> 00:21:15,879 Speaker 1: It's harder for you using the price mechanism than the 374 00:21:15,920 --> 00:21:19,680 Speaker 1: bookseller would. So if in one case, let's say the 375 00:21:19,720 --> 00:21:23,600 Speaker 1: yacht seller bids fifty cents for your eyeballs and the 376 00:21:23,600 --> 00:21:26,680 Speaker 1: bookseller bids a dollar the books, that would win and 377 00:21:26,680 --> 00:21:28,879 Speaker 1: they'll end up paying fifty one cents, just enough to 378 00:21:28,920 --> 00:21:32,399 Speaker 1: beat the yacht sellers bid. Got it, So the bookseller 379 00:21:32,520 --> 00:21:36,439 Speaker 1: might end up bidding fifty one cents and on a 380 00:21:36,560 --> 00:21:40,040 Speaker 1: transaction that you know, maybe we'll make them fifty five 381 00:21:40,080 --> 00:21:42,720 Speaker 1: cents or a dollar or something like that. Right, And 382 00:21:42,760 --> 00:21:45,400 Speaker 1: the idea is you will probably lose money on most 383 00:21:45,440 --> 00:21:48,600 Speaker 1: of these in aggregate, but the winners are real winners. 384 00:21:49,080 --> 00:21:51,840 Speaker 1: And the odds seller occasionally is going to get that 385 00:21:51,960 --> 00:21:56,320 Speaker 1: ad and it's probably going to serve thousands and thousands 386 00:21:56,320 --> 00:21:59,159 Speaker 1: of ads that don't turn to anything, maybe millions, but 387 00:21:59,240 --> 00:22:02,720 Speaker 1: then when they do, those are huge jackpots. Yes, and 388 00:22:02,880 --> 00:22:05,240 Speaker 1: I'll add one more wrinkle. We're talking about ads that 389 00:22:06,080 --> 00:22:08,760 Speaker 1: ads come in many different forms. This isn't just a 390 00:22:08,840 --> 00:22:12,440 Speaker 1: race for eyeballs, could be raising for video ad views. 391 00:22:12,600 --> 00:22:16,320 Speaker 1: Or application installs. Even so I have a sort of 392 00:22:16,320 --> 00:22:20,600 Speaker 1: related question, um sort of maybe even a philosophical question. 393 00:22:20,920 --> 00:22:24,399 Speaker 1: But so much of the online advertising world seems to 394 00:22:24,400 --> 00:22:29,119 Speaker 1: be about targeted advertising. So does anyone just you know, 395 00:22:29,600 --> 00:22:32,600 Speaker 1: throw stuff out there nowadays just on the off chance 396 00:22:32,680 --> 00:22:37,040 Speaker 1: that someone might see a good product and want to 397 00:22:37,040 --> 00:22:39,439 Speaker 1: buy it to Joe's y odd example, you know what 398 00:22:39,520 --> 00:22:42,280 Speaker 1: if someone has a really really nice yacht and they 399 00:22:42,320 --> 00:22:44,639 Speaker 1: think people will buy it if only they were aware 400 00:22:44,680 --> 00:22:48,040 Speaker 1: of it. Sure that does happen. It's also pretty rare 401 00:22:48,119 --> 00:22:50,840 Speaker 1: because in order to actually get a high enough conversion 402 00:22:50,920 --> 00:22:53,399 Speaker 1: ratio for this to be worth the advertiser's time, you 403 00:22:53,640 --> 00:22:55,800 Speaker 1: really have to buy a tremendous amount of volume if 404 00:22:55,800 --> 00:22:59,120 Speaker 1: you're not doing any targeting on who your customer is, 405 00:22:59,520 --> 00:23:02,119 Speaker 1: and now she might be putting up a billboard versus 406 00:23:02,160 --> 00:23:04,480 Speaker 1: sending a direct mail flyer. If you don't know much 407 00:23:04,520 --> 00:23:06,760 Speaker 1: about the person who's engaging on the other side of 408 00:23:06,800 --> 00:23:10,440 Speaker 1: the bye, who knows if they're going to bite? Is there? Though? 409 00:23:10,520 --> 00:23:15,080 Speaker 1: Like sometimes you see like a huge site takeover, like 410 00:23:15,119 --> 00:23:16,840 Speaker 1: on the New York Times or something like that. You'll 411 00:23:16,880 --> 00:23:21,200 Speaker 1: see some massive they'll buy every ad on the website 412 00:23:21,320 --> 00:23:23,200 Speaker 1: or on the front page or something like that. It's 413 00:23:23,240 --> 00:23:27,119 Speaker 1: clearly not particularly targeted. There is still some sort of 414 00:23:27,119 --> 00:23:29,920 Speaker 1: like brand advertising on the internet. Right, Yes, those are 415 00:23:29,960 --> 00:23:33,960 Speaker 1: private auctions, and those are great, but they're really only 416 00:23:34,000 --> 00:23:38,200 Speaker 1: great for publishers who can demand that sort of leverage. 417 00:23:38,920 --> 00:23:41,240 Speaker 1: For bloomber dot com in this example, that would be 418 00:23:41,280 --> 00:23:43,400 Speaker 1: a great example of the site that would have enough 419 00:23:43,920 --> 00:23:46,920 Speaker 1: cashier to pull a private auction. I can imagine plenty 420 00:23:46,920 --> 00:23:49,000 Speaker 1: of brands would want to have exclusive access to the 421 00:23:49,080 --> 00:23:52,679 Speaker 1: eyeballs at the basing dot com. My personal blog probably 422 00:23:52,720 --> 00:23:54,560 Speaker 1: not so much. I'm probably not gonna be able to 423 00:23:54,640 --> 00:23:56,720 Speaker 1: land that big deal with Pepsi or Coca Cola to 424 00:23:57,119 --> 00:23:59,280 Speaker 1: have them take over my site for the next thirty days. 425 00:23:59,800 --> 00:24:03,520 Speaker 1: And that's something that would be negotiated among humans, and 426 00:24:03,560 --> 00:24:06,080 Speaker 1: there would be a deal maybe like there was just 427 00:24:06,160 --> 00:24:11,159 Speaker 1: the big ad conference and canned or people like partied 428 00:24:11,200 --> 00:24:13,280 Speaker 1: on yachts and stuff like that, and those are the 429 00:24:13,359 --> 00:24:16,280 Speaker 1: kinds of deals that get negotiated there rather than through 430 00:24:16,400 --> 00:24:19,880 Speaker 1: some sort of like algorithmic matching engine. Right, and think 431 00:24:19,920 --> 00:24:23,560 Speaker 1: about it if you will. There's the difference between buying 432 00:24:23,680 --> 00:24:26,280 Speaker 1: at the market price on an exchange versus setting up 433 00:24:26,280 --> 00:24:29,480 Speaker 1: an OTC Trade. We like to talk about changes to 434 00:24:29,560 --> 00:24:33,000 Speaker 1: market structure quite a lot on the Pots podcast. Are 435 00:24:33,040 --> 00:24:37,040 Speaker 1: there any big changes or big pressures or disruptions on 436 00:24:37,080 --> 00:24:40,480 Speaker 1: the horizon for the current structure of the way these 437 00:24:40,480 --> 00:24:43,679 Speaker 1: online ads are sold. I'm glad you asked. Let's talk 438 00:24:43,720 --> 00:24:49,439 Speaker 1: about header bidding. Header bidding. Header bidding is what seems 439 00:24:49,440 --> 00:24:51,800 Speaker 1: to be what's going to come next after real time bidding, 440 00:24:51,960 --> 00:24:54,440 Speaker 1: and let's explain what it is and what the motivations 441 00:24:54,440 --> 00:24:57,399 Speaker 1: are behind it. I mentioned Google's DFP double cliff for 442 00:24:57,440 --> 00:25:01,480 Speaker 1: publishers before as an interesting feature. First off, most people 443 00:25:01,480 --> 00:25:03,639 Speaker 1: most of the time use Google for advertising, just in 444 00:25:03,680 --> 00:25:05,120 Speaker 1: the same way that most people most of the time 445 00:25:05,200 --> 00:25:09,480 Speaker 1: use Google for search, and most publishers are content to 446 00:25:09,520 --> 00:25:12,199 Speaker 1: just trust Google to sort of drive the process and 447 00:25:12,240 --> 00:25:14,280 Speaker 1: they don't really dig too much into it because Google 448 00:25:14,280 --> 00:25:17,359 Speaker 1: wouldn't always serve an ad that seems to satisfy both 449 00:25:17,400 --> 00:25:22,639 Speaker 1: the advertisers and the publishers. The reason, in part that 450 00:25:22,680 --> 00:25:24,240 Speaker 1: Google is able to do this, besides the fact that 451 00:25:24,240 --> 00:25:26,880 Speaker 1: they're so great at aggregating data, is what's known as 452 00:25:26,920 --> 00:25:31,119 Speaker 1: the waterfall. Here's how it works. You call out to Google, 453 00:25:31,960 --> 00:25:34,399 Speaker 1: you asked them if they can give an ad to 454 00:25:34,480 --> 00:25:37,600 Speaker 1: you so you can serve it to your visitors. Google 455 00:25:37,640 --> 00:25:40,040 Speaker 1: goes through their inventory to see who the best ad 456 00:25:40,119 --> 00:25:43,280 Speaker 1: is and where it's coming from. They have the option 457 00:25:43,400 --> 00:25:47,640 Speaker 1: of serving it from another exchange or their own exchange addicts, 458 00:25:48,240 --> 00:25:50,920 Speaker 1: so they have the ability to front run basically any 459 00:25:50,960 --> 00:25:54,159 Speaker 1: other advertising ex change if you allow them to. And 460 00:25:54,480 --> 00:25:57,000 Speaker 1: because Google is where all of the publishers show up to, 461 00:25:57,760 --> 00:25:59,600 Speaker 1: all the other exchanges sort of put up with it 462 00:25:59,880 --> 00:26:03,080 Speaker 1: or forced you. Header bidding is meant to change that. 463 00:26:03,200 --> 00:26:07,240 Speaker 1: Header bidding is different from real time bidding, and that 464 00:26:07,400 --> 00:26:10,920 Speaker 1: instead of where in real time bidding, where a publisher 465 00:26:11,000 --> 00:26:14,440 Speaker 1: calls out to a DSP and waits for a response, 466 00:26:15,440 --> 00:26:19,160 Speaker 1: header bidding allows the publisher to call out to multiple 467 00:26:19,200 --> 00:26:21,520 Speaker 1: DSPs all at the same time and select the best 468 00:26:21,520 --> 00:26:25,600 Speaker 1: bid incoming themselves. So you lose the convenience of just 469 00:26:25,680 --> 00:26:29,320 Speaker 1: letting Google drive, but you gain the price competition between 470 00:26:29,320 --> 00:26:32,600 Speaker 1: all of these different bidders. It sounds obviously better for 471 00:26:32,640 --> 00:26:36,119 Speaker 1: the publisher. So why is that not already the norm? Well, 472 00:26:37,000 --> 00:26:40,920 Speaker 1: because Google doesn't like it that way. Most people most 473 00:26:40,920 --> 00:26:43,600 Speaker 1: of the time are actually not as enamored with his 474 00:26:43,720 --> 00:26:45,920 Speaker 1: market mixed structure as you and I are a publisher 475 00:26:46,000 --> 00:26:50,680 Speaker 1: is probably the business of publishing content and doing whatever 476 00:26:50,720 --> 00:26:55,560 Speaker 1: their core businesses. Actually managing and driving this process yourself 477 00:26:55,960 --> 00:26:59,760 Speaker 1: is really tricky. In order to for example, for head 478 00:26:59,760 --> 00:27:01,280 Speaker 1: to be to be really profitable for you, you'd have 479 00:27:01,320 --> 00:27:03,640 Speaker 1: to actually, you know, care about those bids and put 480 00:27:03,640 --> 00:27:07,199 Speaker 1: some thought into whether you might for one partner over another. 481 00:27:07,840 --> 00:27:12,520 Speaker 1: Would you need to build your own technology to determine 482 00:27:12,680 --> 00:27:17,520 Speaker 1: If you're getting multiple bids from multiple exchanges, then it's 483 00:27:17,560 --> 00:27:19,680 Speaker 1: I guess it's on you to actually determine which one 484 00:27:19,720 --> 00:27:21,679 Speaker 1: is the best. Whereas if you're just on one exchange, 485 00:27:22,200 --> 00:27:24,600 Speaker 1: then that determines what's best. And so I imagine to 486 00:27:24,640 --> 00:27:28,720 Speaker 1: take some more supply side infrastructure to build that up. Exactly. 487 00:27:28,720 --> 00:27:31,520 Speaker 1: It's the difference between you know, setting up effectually your 488 00:27:31,520 --> 00:27:34,720 Speaker 1: own prop shop for ads versus just trusting your broker 489 00:27:34,760 --> 00:27:36,960 Speaker 1: to disup why you with what you're asking for. Well, 490 00:27:37,000 --> 00:27:41,240 Speaker 1: that has been absolutely fascinating discussion, and I just love 491 00:27:41,320 --> 00:27:45,440 Speaker 1: the fact that the kind of market structure that topics 492 00:27:45,520 --> 00:27:47,920 Speaker 1: that Tracy and I talked about frequently on this show 493 00:27:48,200 --> 00:27:50,840 Speaker 1: are actually at the root of how the entire Internet works. 494 00:27:50,840 --> 00:27:53,199 Speaker 1: And I didn't even realize that so I'll shooting Bedelli. 495 00:27:53,320 --> 00:27:55,119 Speaker 1: Thank you very much for joining us. Thank you for 496 00:27:55,160 --> 00:28:13,520 Speaker 1: having me, Joe Yea. So, Tracy, you said that you've 497 00:28:13,520 --> 00:28:17,040 Speaker 1: been thinking about online ads and you've been wondering about 498 00:28:17,040 --> 00:28:19,640 Speaker 1: how they get served up to you. Do you feel 499 00:28:19,680 --> 00:28:22,400 Speaker 1: that your questions have been answered? I mean, I definitely 500 00:28:22,400 --> 00:28:25,400 Speaker 1: have a better sense of it now, but I will 501 00:28:25,440 --> 00:28:28,679 Speaker 1: never look at a web page loading the same ever again. 502 00:28:29,160 --> 00:28:32,000 Speaker 1: You know, every time I go someplace, I'm going to 503 00:28:32,119 --> 00:28:36,360 Speaker 1: be thinking about the intense auction process that's currently happening 504 00:28:36,359 --> 00:28:38,840 Speaker 1: in the background with a bunch of people trying to 505 00:28:38,920 --> 00:28:44,040 Speaker 1: bid on my specific profile. It is pretty amazing. I mean, 506 00:28:44,200 --> 00:28:48,400 Speaker 1: sometimes websites or absolute load a little bit slower than 507 00:28:48,440 --> 00:28:51,600 Speaker 1: you would like, but even with that, it is pretty amazing. 508 00:28:51,960 --> 00:28:56,120 Speaker 1: How much has to happen in that short period of time. 509 00:28:56,560 --> 00:28:59,840 Speaker 1: You don't even notice it, but all this information goes out, 510 00:29:00,000 --> 00:29:04,280 Speaker 1: it's analyzed, it's put into some profile. There's a bitting habit, 511 00:29:04,400 --> 00:29:06,960 Speaker 1: a bitty war happens in an ad is served to 512 00:29:07,200 --> 00:29:10,080 Speaker 1: basically in an instant. I don't know if it's exactly 513 00:29:10,160 --> 00:29:13,360 Speaker 1: a miracle, because it's kind of creepy. It's it's kind 514 00:29:13,360 --> 00:29:16,800 Speaker 1: of creepy, right, but it is kind of amazing, isn't it. Yeah, 515 00:29:17,240 --> 00:29:19,400 Speaker 1: I guess it gets to a point that we've talked 516 00:29:19,440 --> 00:29:22,800 Speaker 1: about before on the show, which is also about inequality 517 00:29:22,880 --> 00:29:27,680 Speaker 1: and the ability of algorithms to sort of reinforce a 518 00:29:27,760 --> 00:29:31,320 Speaker 1: certain position that a person is already in. So, for instance, 519 00:29:31,880 --> 00:29:34,440 Speaker 1: you know, if when you're twenty four years old, you 520 00:29:34,480 --> 00:29:38,240 Speaker 1: need to pay day loan to survive until your next 521 00:29:38,320 --> 00:29:42,080 Speaker 1: month's paycheck comes in, and you search online for payday loans, 522 00:29:42,800 --> 00:29:46,360 Speaker 1: those ads might follow you around for years and years 523 00:29:46,400 --> 00:29:49,800 Speaker 1: to come. Um, when other people who have never needed 524 00:29:49,800 --> 00:29:53,400 Speaker 1: to pay day loan might see, you know, advertisements for 525 00:29:53,400 --> 00:29:56,560 Speaker 1: four one case or things like that. I've never really 526 00:29:56,600 --> 00:29:59,240 Speaker 1: thought of it like that before, but I've been thinking. 527 00:29:59,640 --> 00:30:02,680 Speaker 1: We had a whole episode on this, Joe. No, no, no, absolutely, 528 00:30:02,760 --> 00:30:05,960 Speaker 1: I never thought about that before with regards to online 529 00:30:06,000 --> 00:30:09,959 Speaker 1: ads specifically, But I have been thinking about how like 530 00:30:10,040 --> 00:30:12,120 Speaker 1: I went through a phase where I needed to buy 531 00:30:12,160 --> 00:30:14,880 Speaker 1: some clothes. I need to buy some shoes, and then 532 00:30:14,920 --> 00:30:17,400 Speaker 1: I got served tons of ads for weeks and months 533 00:30:17,400 --> 00:30:21,720 Speaker 1: on and on similar stuff. And then I started worrying that, wait, 534 00:30:21,800 --> 00:30:24,840 Speaker 1: am I buying more than I need to now on 535 00:30:25,000 --> 00:30:29,040 Speaker 1: these things? Because I went through this period where I 536 00:30:29,080 --> 00:30:32,120 Speaker 1: made these purchases, and so thinking about the sort of 537 00:30:32,200 --> 00:30:35,440 Speaker 1: lasting impacts of a certain behavior on the type of 538 00:30:35,440 --> 00:30:37,920 Speaker 1: ads were inclined to see for a long time is 539 00:30:37,960 --> 00:30:41,080 Speaker 1: really fascinating and absolutely does speak to that discussion we 540 00:30:41,120 --> 00:30:44,840 Speaker 1: had earlier about the influence of algorithms on our lives. Yeah, 541 00:30:44,960 --> 00:30:48,440 Speaker 1: you cannot escape your Internet history, no matter how hard 542 00:30:48,560 --> 00:30:51,720 Speaker 1: you try. All right, Uh, well, this has been another 543 00:30:51,760 --> 00:30:55,200 Speaker 1: episode of the Odd Thoughts podcast. I'm Tracy Alloway. You 544 00:30:55,200 --> 00:30:58,560 Speaker 1: can follow me on Twitter at Tracy Alloway, and I'm 545 00:30:58,640 --> 00:31:01,280 Speaker 1: Joe Wise though you could allow me on Twitter at 546 00:31:01,280 --> 00:31:05,200 Speaker 1: the Stalwart, and you should follow our producer tofur Foreheads 547 00:31:05,280 --> 00:31:08,360 Speaker 1: on Twitter at Foreheast, as well as the Bloomberg head 548 00:31:08,400 --> 00:31:13,240 Speaker 1: of podcasts, Francesca Levie at Francesca Today. Thanks for listening.