WEBVTT - The Internet Is Secretly Powered By Billions Of Tiny Auctions 0:00:10.119 --> 0:00:13.520 Lord, Welcome to another episode of The Odd Laws Podcast. 0:00:13.600 --> 0:00:17.800 I'm Joe Wisenhall and I'm Tracy Alloway. Tracy, do you 0:00:17.960 --> 0:00:26.960 know what powers the internet? Um? Hamsters on wheels? No, Um, electricity. 0:00:27.200 --> 0:00:31.319 I don't know. Data. Is this like a metaphor? I 0:00:31.360 --> 0:00:34.120 guess you could probably say. I have a lot of 0:00:34.159 --> 0:00:38.160 different answers. Electricity is probably one. Someone could say data. 0:00:38.760 --> 0:00:43.720 I was gonna say online advertising. Ah, now this is interesting. Um. 0:00:43.840 --> 0:00:46.840 Do you mean in the sense that online advertising pays 0:00:46.880 --> 0:00:51.199 for websites to actually run? Yeah. Basically, it's like if 0:00:51.240 --> 0:00:56.040 you think about virtually anything we do on the internet, 0:00:56.520 --> 0:00:59.960 if you think about at least some aspect of our 0:01:00.200 --> 0:01:03.720 careers having written on the internet, if you think of 0:01:03.760 --> 0:01:07.679 all the social networking, literally everything in some way, it 0:01:07.760 --> 0:01:12.479 was probably paid for the online advertising. Yeah, I guess 0:01:12.480 --> 0:01:15.080 you're right. I mean, none of these things are really 0:01:15.160 --> 0:01:19.360 provided for for free. And I guess the cliche is 0:01:19.400 --> 0:01:22.240 always if you're not paying for the service, Uh, then 0:01:22.440 --> 0:01:24.920 you know the product they're selling is is probably you 0:01:25.080 --> 0:01:28.200 and your personal data, and they're selling that to advertisers, 0:01:28.280 --> 0:01:31.120 right exactly. And of course, as we know, a lot 0:01:31.160 --> 0:01:34.800 of the practices and online advertising these days are pretty controversial, 0:01:35.560 --> 0:01:38.480 and people are concerned about privacy and as you said, 0:01:38.520 --> 0:01:41.440 your personal data and the reader or the user being 0:01:41.480 --> 0:01:46.120 the product. But have you ever wondered, really how they 0:01:46.319 --> 0:01:49.040 figure out what ad to serve you? In the split 0:01:49.080 --> 0:01:51.120 second you go to a website and there's already an 0:01:51.120 --> 0:01:53.840 AD that's perfectly tailored to your interest. Have you ever 0:01:53.880 --> 0:01:59.280 really explored how that happens? I'm not lying when I 0:01:59.320 --> 0:02:02.760 say I have actually wondered this specifically, I've wondered why 0:02:02.800 --> 0:02:04.960 the pair of shoes that I looked at for like 0:02:05.280 --> 0:02:08.600 five seconds two weeks ago follow me around the internet 0:02:08.680 --> 0:02:12.040 from months later begging to be bought. That really annoys me. 0:02:12.120 --> 0:02:14.200 So if this is what we're going to talk about, 0:02:14.480 --> 0:02:16.480 then I'm on board, Joe. But what does it have 0:02:16.560 --> 0:02:20.160 to do with markets exactly? Well, it's uh, first of all, 0:02:20.320 --> 0:02:22.280 this is exactly what we're gonna be talking about, So 0:02:22.360 --> 0:02:24.760 you're in luck. And of course it's a market thing 0:02:25.080 --> 0:02:29.560 because someone has to buy and sell that ad inventory 0:02:29.600 --> 0:02:32.240 and essentially you go to a website or go to 0:02:32.280 --> 0:02:35.760 Facebook or whatever, and there has to be some sort 0:02:35.760 --> 0:02:40.840 of process for allocating that advertising space and how much 0:02:40.880 --> 0:02:42.720 are people going to pay for it? And how much 0:02:42.800 --> 0:02:45.320 is that worth and how much are uh they is 0:02:45.360 --> 0:02:47.840 the website going to sell it for? And that is 0:02:48.040 --> 0:02:52.120 a marketplace. So the Internet is fundamentally run by market structure, 0:02:52.160 --> 0:02:56.000 is what you're telling me. I love this episode already exactly. Well, 0:02:56.200 --> 0:02:59.240 I'm very excited about it. So without further ado, I 0:02:59.240 --> 0:03:02.239 want to bring in our guest. His name is Option Bdelli. 0:03:02.360 --> 0:03:06.919 He is a systems engineer for online ad platforms, and 0:03:06.960 --> 0:03:10.519 we are going to talk about market structure as it 0:03:10.680 --> 0:03:20.200 pertained to the world of online advertising. Thank you very 0:03:20.240 --> 0:03:24.200 much for coming on. Thank you, and hi Tracy Hija. 0:03:25.440 --> 0:03:29.240 So what happens when I go to a website? Wow, 0:03:30.320 --> 0:03:33.040 let's just let's just jump right in. Great, so today 0:03:33.520 --> 0:03:36.280 on average for the average website, let's go through what happens. 0:03:37.080 --> 0:03:40.200 So you load up your favorite web page, whatever it is, 0:03:40.240 --> 0:03:44.800 let's say bloom Bloomberg dot com of course, and chances 0:03:44.840 --> 0:03:46.800 are you visited Bloomberg dot com before. This is not 0:03:46.840 --> 0:03:49.840 the first time you visited this site, so that's important. 0:03:49.880 --> 0:03:52.200 We'll get back to that in a bit. What happens 0:03:52.720 --> 0:03:55.560 the moment you load the page there is a race. 0:03:55.840 --> 0:03:58.320 People want to find out who you are and whether 0:03:58.360 --> 0:04:00.880 it's worth their time to show you an AD, And 0:04:01.000 --> 0:04:03.600 the hope is if they show you an AD, maybe 0:04:03.680 --> 0:04:07.320 you will in some way engage or convert or respond 0:04:07.360 --> 0:04:10.360 to that ad in different ways, and there are different 0:04:10.520 --> 0:04:15.440 methods for measuring exactly how you convert. So the whole 0:04:15.800 --> 0:04:18.320 game at this point is to show you an AD 0:04:18.839 --> 0:04:23.240 as quickly as possible, that is, as revenue impacting as 0:04:23.279 --> 0:04:26.599 possible for both the person purchasing the ad and the 0:04:26.640 --> 0:04:30.040 purchase the person selling you the ad, the publisher and 0:04:30.120 --> 0:04:33.120 the advertiser. So I have a bunch of questions already 0:04:33.200 --> 0:04:36.520 about how the buyers and the sellers kind of come together. 0:04:36.720 --> 0:04:39.440 But I guess before we get to that, can you 0:04:39.560 --> 0:04:43.480 describe what kind of personal data is available if Joe 0:04:43.560 --> 0:04:46.600 or I, you know, actually click on a website, be 0:04:46.640 --> 0:04:51.200 at the Bloomberg website or something else. Sure, so we 0:04:51.240 --> 0:04:55.040 can divide this generally into three buckets of data. Let's 0:04:55.080 --> 0:04:57.520 say there's the data that we know about the user 0:04:57.800 --> 0:05:01.000 as as soon as they visit the page, the data 0:05:01.000 --> 0:05:03.560 that is live that is part of the request. Then 0:05:03.600 --> 0:05:06.520 there's the data the Bloomberg in this example has from 0:05:06.560 --> 0:05:11.600 your past visits. And then there's the data that the 0:05:11.640 --> 0:05:16.200 advertiser has purchased from a third party that has been 0:05:16.200 --> 0:05:19.599 collected from you over the course of months and years, 0:05:19.600 --> 0:05:23.600 from all sorts of different sources. The party that is 0:05:23.640 --> 0:05:25.520 in charge of providing this last piet of data, which 0:05:25.520 --> 0:05:28.080 I'm guessing is the most interesting to you, is called 0:05:28.320 --> 0:05:32.080 a d MP, a data management platform. It's the job 0:05:32.080 --> 0:05:35.720 of these DMPs to find out as much data as 0:05:36.000 --> 0:05:38.080 they can possibly get about you, and aggregate from all 0:05:38.080 --> 0:05:41.000 sorts of different databases and data sources, and correlate it 0:05:41.080 --> 0:05:44.240 to the information that is returned to them by the publisher, 0:05:44.279 --> 0:05:47.560 in this case, Bloomberg. So I have a bunch of 0:05:47.760 --> 0:05:53.359 potentially different interests. I'm interested in financial market, I'm interested 0:05:53.640 --> 0:05:59.839 in boxing, I'm interested in barbecuing. So in theory, there's 0:05:59.880 --> 0:06:03.240 a lot of different types of advertisers, very different from 0:06:03.279 --> 0:06:07.440 each other, that would might want to compete for my attention, 0:06:07.480 --> 0:06:10.000 that might want to put up some product that they're 0:06:10.040 --> 0:06:13.440 selling in front of me. So how what is the 0:06:13.480 --> 0:06:17.680 process by which that collection of data that the d 0:06:17.880 --> 0:06:22.839 MP has on me turns into someone putting an atom 0:06:22.880 --> 0:06:26.160 from ah? So that is the job of the DSP, 0:06:26.360 --> 0:06:28.920 the demand side platform. Are you enjoying the outfits? I 0:06:28.920 --> 0:06:30.600 I am already I have a feel like there's more 0:06:30.640 --> 0:06:33.880 to come to there are Okay, So the demand side 0:06:33.880 --> 0:06:37.680 platform partners with several d MPs, and it's their job 0:06:37.760 --> 0:06:41.360 to look at your incoming data, your page visit data 0:06:41.440 --> 0:06:43.560 that happens when you visit bloomber dot com in this 0:06:43.600 --> 0:06:45.920 one moment and correlated to all the different data it 0:06:45.960 --> 0:06:47.920 gets from the d MPs. And how they go about 0:06:47.960 --> 0:06:50.640 that is sort of proprietary. It's the special sauce. But 0:06:50.760 --> 0:06:54.280 imagine they have just about any imaginal data stores that 0:06:54.320 --> 0:06:57.320 they can possibly call on you based on your email 0:06:57.320 --> 0:06:59.239 address and how who you've given to you in the past, 0:06:59.279 --> 0:07:01.760 cookies that you've often websites in the past, even location 0:07:01.800 --> 0:07:04.920 begins from from brick and mortar real life stories that 0:07:04.960 --> 0:07:07.560 you visit in the past, and they're correlating it all 0:07:08.120 --> 0:07:10.720 and that single moment while you're viewing the page in 0:07:10.800 --> 0:07:12.680 order to figure out which ad is best to serve 0:07:12.680 --> 0:07:15.320 to you. It's weird to think there's like a behind 0:07:15.360 --> 0:07:19.200 the scenes battle happening every time you open up a 0:07:19.200 --> 0:07:22.720 new web page. I'm wondering how how did this structure 0:07:22.880 --> 0:07:25.720 actually come into place? Like who made the decision that 0:07:25.760 --> 0:07:27.880 this was going to be the way um that the 0:07:27.920 --> 0:07:32.040 Internet essentially works and that advertising is sold online. So 0:07:32.120 --> 0:07:34.400 let's go through a brief history of how the market 0:07:34.440 --> 0:07:38.400 micro structure of online advertising evolved. Those are words I 0:07:38.400 --> 0:07:40.600 never thought I'd be saying before, but here we are. 0:07:41.120 --> 0:07:43.360 So in the beginning that this is why they welcome 0:07:43.400 --> 0:07:48.600 to the outlaws. So in the beginning there were advertisers 0:07:48.600 --> 0:07:52.440 and publishers. So in this case, there's let's say Bloomberg 0:07:52.520 --> 0:07:56.440 and someone who wants to advertise on Bloomberg, and around 0:07:56.600 --> 0:07:59.240 let's say twenty years ago, was fairly easy for these 0:07:59.240 --> 0:08:01.440 two parties to find each other and match up. There 0:08:01.440 --> 0:08:03.320 were not that many people who were advertising, and they're 0:08:03.360 --> 0:08:06.960 not many people publishing websites. But more and more advertisers 0:08:06.960 --> 0:08:10.360 and more and more publishers makes this market microstructure really inefficient. 0:08:10.400 --> 0:08:13.880 Imagine the equipment of an open outcry pit, but only 0:08:13.920 --> 0:08:17.080 about five people in each pit, scattered all around the world. 0:08:17.440 --> 0:08:19.480 How to buyers and sellers match up with each other 0:08:19.520 --> 0:08:22.800 after they've reached a certain volume. The solution to that 0:08:23.000 --> 0:08:28.679 first became advertising networks. This would be an aggregation of 0:08:29.000 --> 0:08:35.000 advertisers under one umbrella being presented to publishers. That solves 0:08:35.040 --> 0:08:39.080 one problem, how do you get more supply to satisfy demand, 0:08:39.679 --> 0:08:44.000 but it exposes another problem. To give an example where 0:08:44.000 --> 0:08:46.440 that problem is, let's put some color on what ad 0:08:46.440 --> 0:08:48.720 we're serving on Bloomberg dot com. Let's say we're serving 0:08:48.720 --> 0:08:53.240 an ad for men's shoes. So if I present that 0:08:53.280 --> 0:08:56.000 add to Joe when he's being a boomer dot com, 0:08:56.040 --> 0:08:59.240 there's probably a higher percentage of a chance that he's 0:08:59.240 --> 0:09:01.200 going to convert actually engage with that ad. Then if 0:09:01.240 --> 0:09:03.240 I were to show those same men shoes to Tracy, 0:09:03.480 --> 0:09:05.840 who I'm guessing is not in the market for Adidas 0:09:06.000 --> 0:09:11.640 right now, so we need to wait now. So we 0:09:11.679 --> 0:09:13.200 need a way to solve that problem. We need a 0:09:13.200 --> 0:09:15.720 way to actually show Joe the ads that he cares 0:09:15.760 --> 0:09:19.720 about and Tracy the ads that she cares about. So 0:09:20.320 --> 0:09:24.960 that led to add exchanges, and that in turn became 0:09:25.040 --> 0:09:32.360 demand side platforms. The evolution how we went from ad 0:09:32.480 --> 0:09:35.920 networks to add exchanges to demand side platforms basically came 0:09:35.960 --> 0:09:38.760 from the recognition of this problem that although we have 0:09:38.760 --> 0:09:42.120 the ability to match up advertising publishers, buyers, and sellers, 0:09:42.360 --> 0:09:46.080 neither of them really know how to match up the 0:09:46.120 --> 0:09:48.800 best inventory with each other. They know how to give 0:09:48.840 --> 0:09:51.640 each other, you know, a certain lot size, a certain 0:09:51.720 --> 0:09:54.800 amount of certain quantity in terms of number of ads 0:09:54.800 --> 0:09:56.680 are in terms of revenue spent, but they don't really 0:09:56.679 --> 0:09:59.160 know how to connect to Joe to give him the 0:09:59.200 --> 0:10:03.319 best add that he wants. Tracy, I'm listening to off 0:10:03.520 --> 0:10:05.440 and describe this. I've been thinking back to some of 0:10:05.480 --> 0:10:09.319 our early episodes with Chris White talking about bond market 0:10:09.360 --> 0:10:15.320 structure and just the sheer multitude of potential players in 0:10:15.320 --> 0:10:20.480 the game and varieties of bonds that people could be selling, 0:10:20.679 --> 0:10:25.080 different coupons, different times, different structures, and the sort of 0:10:25.320 --> 0:10:31.559 complicated challenges that that poses in terms of bringing everyone together. Yeah. Absolutely, 0:10:31.640 --> 0:10:34.280 I wonder in that case again, how they're able to 0:10:34.320 --> 0:10:37.160 do it so quickly that we don't even notice that 0:10:37.240 --> 0:10:40.200 it's actually happening in the background. So Offiction, what's the 0:10:40.240 --> 0:10:44.320 technology that's actually allowing people to do this? So the 0:10:44.360 --> 0:10:46.920 technology is referred to as you ready, Joe, another acronym 0:10:47.160 --> 0:10:52.000 RTB Real time didn't and it's meant to happen as 0:10:52.080 --> 0:10:56.720 quickly as possible to connect advertisers and publishers through supplies 0:10:56.760 --> 0:11:02.959 lie platforms SSPs and demands like platforms DSPs. Why it's 0:11:03.000 --> 0:11:06.000 so fast. Well, there's an interesting wrinkle or two there. 0:11:06.040 --> 0:11:09.640 First off, the publisher always has the option of showing 0:11:09.679 --> 0:11:12.280 you a house AD if they don't get a proper 0:11:12.360 --> 0:11:14.720 real time bid that they can match open in time. 0:11:15.200 --> 0:11:17.080 The reason you might want to show a house ad 0:11:17.160 --> 0:11:19.599 versus a targeted AD is because you never want to 0:11:19.679 --> 0:11:21.400 leave a user waiting for an ad load. I'm sure 0:11:21.440 --> 0:11:23.720 you've all had the experience of visiting a page and 0:11:23.760 --> 0:11:25.360 waiting for to load, and waiting for to load, and 0:11:25.360 --> 0:11:27.080 when it does, you get a pop up or something 0:11:27.080 --> 0:11:32.319 more innoxious. Generally you want to avoid that. What URTV 0:11:32.400 --> 0:11:37.640 allows publishers to do is present the best possible targeted AD. 0:11:37.640 --> 0:11:39.240 But if they can't do that in time, they will 0:11:39.280 --> 0:11:41.719 just be happy to show you any ad. Maybe they 0:11:41.720 --> 0:11:44.040 have their own personal brands they'd like to show off 0:11:44.080 --> 0:11:46.760 to you. Maybe they have a preferred private partner who's 0:11:46.800 --> 0:11:48.880 as they have an inventory that they would front and 0:11:48.920 --> 0:11:51.600 run in front of their RTV ads. It really depends 0:11:51.600 --> 0:11:56.320 on the publisher. So explain to us the auction process 0:11:56.400 --> 0:12:00.960 real time bidding. It's essentially describing an auction, So to 0:12:01.000 --> 0:12:05.000 walk us through what is that micro second or whatever 0:12:05.120 --> 0:12:09.400 that or that add is delivered. How the bidding works? Sure? Okay, 0:12:09.440 --> 0:12:11.520 so let's walk through our example again of Joe visiting 0:12:11.559 --> 0:12:16.160 Bloomberg dot Com. Joe visits Bloomberg, Bloomberg sends Joe's data 0:12:16.240 --> 0:12:20.120 to their SSP, their supply side platform. The SSP in 0:12:20.160 --> 0:12:23.800 turn transforms this data into a bid and passes it 0:12:23.840 --> 0:12:27.680 off to the ad exchanges. The ad exchanges in turn 0:12:28.040 --> 0:12:33.000 pass this off to their d MPs. Are you following 0:12:33.000 --> 0:12:37.640 the microxtra platforms? Sorry? I say d MPs and a 0:12:37.720 --> 0:12:40.280 d MP DSPs Even I get confused. The d MPs 0:12:40.280 --> 0:12:42.079 are the data management platforms. They're in charge of gain 0:12:42.160 --> 0:12:45.560 the data. The DSPs are the demand side platforms, and 0:12:45.600 --> 0:12:47.640 the demand side platforms in turn look for the best 0:12:47.679 --> 0:12:53.000 inventory the bids that match up with the eyeballs that 0:12:53.040 --> 0:12:55.800 they're getting from Bloomberg and try to make the best match. 0:12:55.960 --> 0:12:59.319 It's the DSPs side too, or a job to service 0:12:59.360 --> 0:13:02.680 sort of the match engine here. So this all goes 0:13:02.720 --> 0:13:07.800 through three counterparties and end if you're counting. So I'm 0:13:07.840 --> 0:13:11.720 curious what pricing actually looks like under this system, because 0:13:12.000 --> 0:13:15.880 if I think back to old style advertising, um, and 0:13:15.960 --> 0:13:18.319 you know, please let me engage in some media naval 0:13:18.360 --> 0:13:22.720 gazing here. But if you were selling newspapers, um, for instance, 0:13:22.960 --> 0:13:26.360 you would know roughly what the audience for your newspaper 0:13:26.400 --> 0:13:30.120 circulation was, and you would have a decent idea of 0:13:30.240 --> 0:13:32.679 you know their income levels and where they live, and 0:13:32.720 --> 0:13:34.880 what they like to spend money on and things like that. 0:13:36.000 --> 0:13:40.440 But now these online ads are so targeted, does that 0:13:40.640 --> 0:13:43.520 drive down the pricing because you're not You're not really 0:13:43.800 --> 0:13:48.240 targeting like a big swathe of the population that you're 0:13:48.280 --> 0:13:52.240 hoping might buy your product. You're actually targeting a specific person. 0:13:53.120 --> 0:13:55.199 It actually seems to drive up the pricing because you 0:13:55.240 --> 0:13:58.319 don't really want to flood the market with ad volume 0:13:58.360 --> 0:14:00.319 that most people won't engage, and you want to hit 0:14:00.640 --> 0:14:03.480 the lowest number of people because remember there's a ton 0:14:03.520 --> 0:14:06.080 of technical effort going behind all this. There are servers 0:14:06.120 --> 0:14:08.240 that need to run and engineers that need to be paid, 0:14:08.240 --> 0:14:10.199 and all that. You want to put the least amount 0:14:10.200 --> 0:14:12.760 of effort into pushing your ads to the most the 0:14:12.760 --> 0:14:15.280 people who are most likely to engage with them. You 0:14:15.400 --> 0:14:19.120 never want to have any extra ad target hit that 0:14:19.200 --> 0:14:22.880 hits someone who's unlikely to click it. Ideally, not think 0:14:22.920 --> 0:14:25.760 about it like this. Let's say you have a thousand 0:14:25.840 --> 0:14:28.280 ads that you want to push to people, you have 0:14:28.360 --> 0:14:30.280 the choice. And this is how it used to work 0:14:30.280 --> 0:14:32.880 about ten or fifteen years ago, of just agreeing to 0:14:32.920 --> 0:14:35.880 a rate upfront with a publisher and saying these thousand 0:14:35.960 --> 0:14:38.200 ads will sell for ten dollars and leaving the publisher 0:14:38.200 --> 0:14:40.520 in charge of running those ads as fast as they 0:14:40.560 --> 0:14:43.800 can sell. And some people will be engaging with those 0:14:43.840 --> 0:14:46.960 ads Monday around two pm, some it's Saturday four am 0:14:47.000 --> 0:14:49.920 in the morning. Some from the United States, some from Indonesia. 0:14:50.000 --> 0:14:52.720 They'll all get the same rate. That's really not efficient 0:14:52.720 --> 0:14:54.520 from an advertiser's point of view. They're really not getting 0:14:54.560 --> 0:14:57.640 the most bang for their buck. So what they would 0:14:57.680 --> 0:14:59.920 like to do instead is push to people who they 0:15:00.080 --> 0:15:03.520 know as much information about and for whatever reason, they've 0:15:03.560 --> 0:15:07.720 decided have the highest confidence and their ability to convert 0:15:08.040 --> 0:15:12.440 with the add Now I'm curious about sort of you know, 0:15:12.480 --> 0:15:15.280 if we really dive into the sort of financial market 0:15:15.680 --> 0:15:19.360 aspect of this, are there players in the market that 0:15:19.440 --> 0:15:24.040 attempt to for better lack of better warriors engage in 0:15:24.160 --> 0:15:28.400 arbitrage buy up cheap, buy up inventory that they think 0:15:28.640 --> 0:15:30.680 they could get on the cheap, and then sort of 0:15:30.720 --> 0:15:34.200 repackage it in some way and resell up more expensive 0:15:34.280 --> 0:15:39.440 like essentially, you know, essentially become traders. Absolutely, And there 0:15:39.440 --> 0:15:41.680 are several different arbitraged strategies, just as you'll find in 0:15:41.680 --> 0:15:45.080 any other market microstructure. There are plenty of companies who 0:15:45.080 --> 0:15:47.480 are in the business of taking in raw ad supply 0:15:47.600 --> 0:15:50.720 and sort of sharry picking what they think is the 0:15:50.720 --> 0:15:54.280 best possible converting ad. And on the other side, there 0:15:54.320 --> 0:15:56.320 are plenty of people who in the business of aggregating 0:15:56.400 --> 0:16:02.560 eyeballs and only presenting the best possible eyeballs to vertisers. Fortunately, 0:16:02.600 --> 0:16:04.440 I'm not sure I can give any of their names 0:16:04.680 --> 0:16:08.440 on air. How profitable is that business? Then, like can 0:16:08.480 --> 0:16:10.680 you make a ton of money out of it? Given 0:16:10.680 --> 0:16:14.200 that the pricing is quite low, it is extraordinarily profitable 0:16:14.800 --> 0:16:18.240 in the tens of billions of dollars a year. So 0:16:18.560 --> 0:16:22.360 from time to time you mentioned secret sauce, I think 0:16:22.400 --> 0:16:24.680 at one point and you talked about this, and you 0:16:24.760 --> 0:16:28.560 hear about this world of companies called ad tech, and 0:16:28.640 --> 0:16:30.960 I never am totally clear what they do. And a 0:16:31.040 --> 0:16:33.720 lot of them seem to rise really fast and fall 0:16:33.880 --> 0:16:37.640 really fast. So what are some of the competing strategies? 0:16:37.720 --> 0:16:40.640 Some company comes along and says we have something new. 0:16:40.720 --> 0:16:42.840 We have a new way that you can reach reach 0:16:42.840 --> 0:16:45.280 your clients more officially, or we're going to use AI 0:16:45.440 --> 0:16:47.840 or machine learning or big data or whatever. Like. What 0:16:47.920 --> 0:16:51.760 are these ad tech companies? Where are they in the process, 0:16:51.800 --> 0:16:54.680 and how do they attempt to make things more efficient? 0:16:55.320 --> 0:16:57.720 So there are really two basic strategies for these at 0:16:57.800 --> 0:17:01.960 tech companies. One is to make the process faster to 0:17:02.480 --> 0:17:06.280 present adds faster to eyeballs that are willing to view 0:17:06.280 --> 0:17:10.600 them and gain a sort of latency arbitrage edge. The 0:17:10.640 --> 0:17:14.080 second strategy is to gather as much information as possible 0:17:14.119 --> 0:17:16.560 from all these different accurate sources and as quickly as 0:17:16.560 --> 0:17:18.960 possible match it up to the visitor that's trying to 0:17:19.040 --> 0:17:21.720 view the ad. Most add to companies focus on the 0:17:21.760 --> 0:17:24.760 second strategy. There in the business of gathering tremendous amounts 0:17:24.760 --> 0:17:28.199 of data, correling it all very very quickly, and presenting 0:17:28.200 --> 0:17:31.680 it to you quicker than the page will load. I'm 0:17:31.720 --> 0:17:34.760 curious about the data gathering aspect. So let's say you 0:17:34.840 --> 0:17:38.439 and I or the three of us here, we're like, okay, 0:17:38.480 --> 0:17:41.800 we want to start a new ad tech company, and 0:17:42.600 --> 0:17:44.640 where are we how do we start gathering that data, 0:17:44.720 --> 0:17:47.760 like where are there wholesale brokers of it? Or that 0:17:47.920 --> 0:17:52.000 can we collected ourselves? Like what's the process there? So 0:17:52.160 --> 0:17:55.560 first we were probably partnered with several publishers like Bloomberg 0:17:55.640 --> 0:17:58.320 for example, to go back to the beginning, we would 0:17:58.359 --> 0:18:00.720 also partner with third party data providers. We would want 0:18:00.720 --> 0:18:03.840 a combination of raw fresh data coming in from new 0:18:03.920 --> 0:18:07.080 visits so we can build our own database, and we 0:18:07.200 --> 0:18:09.080 want to have the ability to correlate it with as 0:18:09.160 --> 0:18:11.879 many pre existing data sources. You probably don't want to 0:18:11.880 --> 0:18:13.960 reinvent the wheel plenty. There are plenty of public data 0:18:14.000 --> 0:18:16.480 basines and four paid databases out there where you can 0:18:16.480 --> 0:18:20.119 find out, for example, from your email address or from 0:18:20.160 --> 0:18:23.080 your IP address, what country you're in, what city you're in, 0:18:24.119 --> 0:18:26.800 what sites you've signed up for the past, what your 0:18:26.920 --> 0:18:29.240 LS you visited in the past, what time zone you're in, 0:18:29.280 --> 0:18:32.400 what language you speak, all that sort of stuff. So 0:18:32.440 --> 0:18:35.159 what's the most lucrative area when it comes to this 0:18:35.240 --> 0:18:38.760 ecosystem of online ad selling is that the data collection? 0:18:39.280 --> 0:18:43.200 Is it? You know, the underlying technology of the bidding system? 0:18:43.320 --> 0:18:46.720 Is it if you're you know, arbitraging big blocks of 0:18:47.160 --> 0:18:52.680 potential eyeballs to sell to people. What makes the most money. So, 0:18:53.119 --> 0:18:56.040 judging from the market cap of the at two companies 0:18:56.080 --> 0:18:58.359 out there today, I'm going to guess it's a combination 0:18:58.560 --> 0:19:03.879 of selling the ad and also taking a bit of 0:19:04.200 --> 0:19:07.120 the cut of every ad that's sold. So good example 0:19:07.320 --> 0:19:11.600 here might be Google, which most publishers use as a 0:19:11.800 --> 0:19:16.359 demand side partner. So Google gets their money two ways. 0:19:16.480 --> 0:19:19.960 If you're using their product DFP double clip for publishers, 0:19:20.000 --> 0:19:23.800 if you're connecting to Google through DFP, they get a 0:19:23.840 --> 0:19:27.520 cut of every ad that is served through DFP, but 0:19:27.600 --> 0:19:29.800 also they have the option of serving ads from their 0:19:29.840 --> 0:19:34.200 own inventory through their exchange addicts. So for an analogy, 0:19:34.320 --> 0:19:38.800 think if, for example, on the COMEX, you both ran 0:19:38.880 --> 0:19:41.919 the exchange and also ran a gold mining company and 0:19:42.000 --> 0:19:44.440 offered a gold etf on the exchange, you not only 0:19:44.480 --> 0:19:47.080 get a cut of every trade, but you're sourcing inventory 0:19:47.119 --> 0:19:50.280 that seems to be the most profitable. So Google not 0:19:50.480 --> 0:19:55.040 only runs the exchange, they also are participated on the 0:19:55.080 --> 0:19:58.400 exchange and are selling a raw commodity on that same exchange, 0:19:58.400 --> 0:20:00.800 and so they make money multip both ways from that 0:20:00.880 --> 0:20:04.480 trade from that environment exactly. In fact, they even it 0:20:04.520 --> 0:20:08.639 sounds like Google's pretty good business, yeah, which I thought of? 0:20:08.680 --> 0:20:12.399 That is Facebook similar in that respect, they have a 0:20:12.440 --> 0:20:15.240 similar business. It's slightly different for social media, but similar. 0:20:15.920 --> 0:20:19.880 So I'm curious about the matching of the right add 0:20:19.920 --> 0:20:23.520 to my interest because maybe I'm interested in buying a 0:20:23.640 --> 0:20:26.399 yacht or a new car, some very lucrative thing, or 0:20:26.440 --> 0:20:28.280 if I clicked on the ad someone could make a 0:20:28.280 --> 0:20:30.919 ton of money. But I might also be interested in 0:20:31.040 --> 0:20:33.840 buying a book from Amazon that you know, it's probably 0:20:33.880 --> 0:20:37.640 the margins are small. So how does it balance those 0:20:38.040 --> 0:20:41.760 disparate potential outcomes to figure out which one is the 0:20:41.800 --> 0:20:44.040 best to serve? Well, the best way to describe it 0:20:44.080 --> 0:20:47.199 would be the free market. The DSP is connected to 0:20:47.200 --> 0:20:49.560 several advertisers. Each one of them is willing to bid 0:20:49.600 --> 0:20:52.080 on your impression. So it really comes down to is 0:20:52.240 --> 0:20:55.560 the yacht seller advertiser in the bookseller advertiser? How much 0:20:55.560 --> 0:20:58.960 do they both value your eyeballs? Maybe the yacht seller 0:20:59.080 --> 0:21:01.920 thinks you're not really convert on an ad because he's 0:21:01.920 --> 0:21:03.840 looked up your date and says, oh, well, you know 0:21:04.359 --> 0:21:06.159 you live here in this time. I've never bought a 0:21:06.200 --> 0:21:08.920 yacht before. Yeah, so it's say that you probably would 0:21:08.960 --> 0:21:10.720 in the future, but you've probably bought plenty of books, 0:21:10.760 --> 0:21:13.280 so maybe the yacht seller is not going to fight. 0:21:13.359 --> 0:21:15.879 It's harder for you using the price mechanism than the 0:21:15.920 --> 0:21:19.680 bookseller would. So if in one case, let's say the 0:21:19.720 --> 0:21:23.600 yacht seller bids fifty cents for your eyeballs and the 0:21:23.600 --> 0:21:26.680 bookseller bids a dollar the books, that would win and 0:21:26.680 --> 0:21:28.879 they'll end up paying fifty one cents, just enough to 0:21:28.920 --> 0:21:32.399 beat the yacht sellers bid. Got it, So the bookseller 0:21:32.520 --> 0:21:36.439 might end up bidding fifty one cents and on a 0:21:36.560 --> 0:21:40.040 transaction that you know, maybe we'll make them fifty five 0:21:40.080 --> 0:21:42.720 cents or a dollar or something like that. Right, And 0:21:42.760 --> 0:21:45.400 the idea is you will probably lose money on most 0:21:45.440 --> 0:21:48.600 of these in aggregate, but the winners are real winners. 0:21:49.080 --> 0:21:51.840 And the odds seller occasionally is going to get that 0:21:51.960 --> 0:21:56.320 ad and it's probably going to serve thousands and thousands 0:21:56.320 --> 0:21:59.159 of ads that don't turn to anything, maybe millions, but 0:21:59.240 --> 0:22:02.720 then when they do, those are huge jackpots. Yes, and 0:22:02.880 --> 0:22:05.240 I'll add one more wrinkle. We're talking about ads that 0:22:06.080 --> 0:22:08.760 ads come in many different forms. This isn't just a 0:22:08.840 --> 0:22:12.440 race for eyeballs, could be raising for video ad views. 0:22:12.600 --> 0:22:16.320 Or application installs. Even so I have a sort of 0:22:16.320 --> 0:22:20.600 related question, um sort of maybe even a philosophical question. 0:22:20.920 --> 0:22:24.399 But so much of the online advertising world seems to 0:22:24.400 --> 0:22:29.119 be about targeted advertising. So does anyone just you know, 0:22:29.600 --> 0:22:32.600 throw stuff out there nowadays just on the off chance 0:22:32.680 --> 0:22:37.040 that someone might see a good product and want to 0:22:37.040 --> 0:22:39.439 buy it to Joe's y odd example, you know what 0:22:39.520 --> 0:22:42.280 if someone has a really really nice yacht and they 0:22:42.320 --> 0:22:44.639 think people will buy it if only they were aware 0:22:44.680 --> 0:22:48.040 of it. Sure that does happen. It's also pretty rare 0:22:48.119 --> 0:22:50.840