#MediaSnack Meets: Shailin Dhar, Method Media Intelligence

Intro: 0:02

Hello everyone. I'm Tom Denford, cofounder of ID Comms. Welcome to episode 26 of MediaSnack Meets recorded each week in New York. We get to meet the individuals and organizations doing great work to inspire success and drive change within the global media and marketing industry. In each episode we find out what is behind that success, what it takes to make change in the industry, and what the rest of us can learn from that experience. My guest for this episode is Shailin Dhar. He's the cofounder and CEO of Method Media Intelligence. He is one of the world's foremost experts on counterfeit web traffic, which is essentially the raw materials for committing what is commonly known as advertising fraud or ad fraud. Ad Fraud is becoming an epidemic problem for the industry and estimates of the proportion of fraudulent ad impressions and web traffic seems to grow each year. Whilst there is a consistent outcry, the industry has struggled to find consensus on how to deal with it. This is where people like Shailin come in helping educate marketers on how to change behavior to mitigate the risks. Shailin has been called the fraud crusader helping to shine a light on the practices of fraudulent advertising, impressions and clicks, and he knows what he's talking about because he actually used to be on the dark side of that equation for many years, creating the sites that generated fraudulent web traffic. He has since now turn to the bright side and formed a company helping advertisers avoid counterfeit traffic. In this episode, Shailin explains how Ad Fraud Works. How the fraudsters are usually actually just normal people with LinkedIn profiles and business cards and how marketers can use new technologies to protect themselves now. You can check the full show notes for this episode@ mediasnackpodcast.com/26. Without further ado, please enjoy this fascinating interview with Shailin Dhar.

Tom Denford: 2:04

Hi Shailin, Welcome to MediaSnack Meets.

Shailin Dhar: 2:06

Thanks for having me, Tom.

Tom Denford: 2:07

So I've explained a little bit about your background and where you came from and we'll dig definitely into that because I think it's a fascinating story and I love your story, your route into becoming the fraud crusader. For those that don't know, I've used the word ad fraud with you before and you've kind of corrected me to call it a study of counterfeit traffic, but let's just go right back to the basics. Just tell me what is ad fraud and why is it a problem for the industry at the moment?

Shailin Dhar: 2:30

I have a two part answer. One is the industry definition, ad fraud as an industry definition basically covers anything not good coming up in a media buy and that includes nonhuman traffic and different classifications of nonhuman traffic. It also includes purposely non viewable media in the industry definition as well as malicious brand safety violations or domain spoofing or things of those sorts. My definition of ad fraud is far more focused and it doesn't go to just the malicious instances, but it goes to any instance where a marketer is intending to buy ad space or attention of a human and gets something entirely different. We mainly focus on counterfeit web traffic. I prefer a counterfeit web traffic because it's far more specific and everybody can understand what that is; a visit to a webpage and a loading of an ad space that is not done by a human.

Tom Denford: 3:35

How does this happen? Why is this not an easy thing to identify and fix?

Shailin Dhar: 3:41

So technically there's lots of ways to address this problem, but again, it goes back to the industry definition as everything as a whole and when you're trying to address 15-20 different types of fraud or misbehavior, you're never going to have that pinpointed solution. This is because that market is forcing verification vendors to do 15 different things, they have to detect bots, but also have an engineering department that can crawl and contextually analyze the context and content of a page to see whether this fits with an advertiser's intentions or not. When you're asking one company to do all of those things, that's why there's no real productive solution that comes out at the other end.

Tom Denford: 4:27

Is it fair to say that the counterfeiters are smarter than the police or not?

Shailin Dhar: 4:32

No, it's not that they're smarter. It's that more of the supply chains incentives are aligned towards the"fraudster" and I say that kind of tongue in cheek because they don't think of themselves as criminal masterminds. A lot of them are just employees at a business that attends conferences and they're an ad network or they're some type of ad platform. A lot of it happens outside of the boardroom and then outside of the entry level employees, or the execution people. It happens with the misalignment of sales incentives and then performance goals where somebody in that organization is forced to find easy routes to see that number appeared on their dashboard.

Tom Denford: 5:14

You and I met like two or three years ago now and we were originally introduced by a mutual friend that many listeners may know or be aware of. A guy called Mikko Kotila, who was one of the first very early researchers into ad fraud. I'd gotten to know Mikko and I had a request of him, which was how can you make this understandable by humans? Because this is very technical and very complex. It's a high tech, counterfeit, fraudulent industry. I really need someone who can help me translate that into human language for the benefit of marketers. That was really what I thought the opportunity was. If marketers are going to address this, they need to be able to hear from people who are not explaining it in binary code issues, but in human terms. As you know I said to Mikko"when you come across someone who can explain it in human language, like let me know" and I didn't hear from him for a long time and then eventually he said,"there's this guy, Shailin Dhar in San Francisco, you need to meet him" and that's how we got introduced. Then a couple of years subsequent to that we actually met face to face at an event in London and we've been able to do some work together for a couple of advertisers. When we met in London, you'd explained to me the journey that you personally had been on to get into ad fraud because it's not something when you're at school that you aspire to become an ad fraud researcher. So what was that journey? Because it's fascinating where you started.

Shailin Dhar: 6:32

I started by working for a company as a salesperson that offered marketing services to dental clinics and veterinary clinics. They did that by basically going into the affiliate market space and buying leads for specific geographies. So whether it's a zip code or a state or province, they would do that type of targeting. Along that process, I went outside of sales to also managing the actual sourcing of these leads. Attending an affiliate summit was my first peek into the reality that there are thousands of people selling traffic with no discrimination of where it's coming from and no seeming limit on the quantity that's available. The first time I realized that this type of stuff can be faked is when- this company had a call center and where we would process these leads and then find the appropriate dental and veterinary clinic depending on their needs- Once we expanded our budget and basically said,"okay, we want these three provinces in Canada and the entire United States," we started getting all types of fake leads. That's when I realized that they just wrote a program to scrape the director of pages of the geographies that we outlined and then wrote another program to visit our landing page and fill out the lead form 200 times a day, 500 times a day, whatever it was. So that was the first time I actually had that epiphany moment of, these people can fake whatever we ask them to deliver. From there, I actually left Canada and started working at an ad network here in New York and my job was to grow the operation from two sites that had traffic running through it. I didn't know what that meant at the time, but it was basically buy traffic to the site. We have ad partners that have their ad codes on the site, make sure this generates a profit, make sure that the revenue from the ads is always higher than the cost of the traffic. We grew that from two sites to 50 sites and from 50 sites to 300 sites. After my first instance of getting a hundred sites blacklisted at the same time, my immediate job role response was, I need to always have a hundred sites as backup and then when you're doing all these things and kind of always trying to circumvent the policies of the system, you realize, okay, this is not that hard. But also you realize that the incentives at all of these big ad tech companies- the account directors, the account management, everybody- their incentives are to make you successful as their customer and as their customer I was asking to pre-approve a hundred sites at once that had no traffic. Why any exchange would do this doesn't make any sense. I don't think anybody would be able to give a public answer as to why this is a legitimate practice, but they did it. That's when I realized that this system that's in place is not designed to forbid this type of activity. A lot of the ad exchanges actually would court us, take us out to lavish dinners and whatnot because they looked at the impression volume that we were doing. It didn't matter about the CPMS, it didn't matter that we were selling ad space that sold for 15-25 cents CPM, which is by no standards premium ad space. We were doing hundreds of millions of impressions and if you're paying a 1 cent CPM ad serving rate, you're paying quite a bit of money every day and ad serving fees. That's a lot of revenue for these mid stage venture funded companies. Not only were we being successful at circumventing the system, we were actually being rewarded by operators in the system for what we were doing and at no point did we lie and have to say, this is all human traffic or any type of substantiation of the operation. That kind of sparked my interest in,"Oh, I think there's a lot of education to be done", so I went from there to just realizing that I can at least enlighten everybody to this world of everything can be faked if you make the expectations at a certain level, which is just technical deliverables, not customers. Can you give me impressions? Can you give me web visits? Can you give me lead forms? All of these types of things can be faked and I think that's still something that's missing as general knowledge in the industry.

Tom Denford: 10:38

Did you get a sense of what the scale was, like how many other people were doing this or did you at the time think that you were one special person, like filling the system?

Shailin Dhar: 10:47

When I started growing that operation from two sites to 50 sites and onwards, I thought I was special, but then I realized that there are thousands of people just like me who wear suits, sweaters and polish shoes to a conference and come there to do sales and whatnot, and none of this is done as a crime, but instead a malpractice by the people that do it. They're not hiding in the shadows, they're not afraid. Go to any Ad Tech Conference and ask 20 people if they have traffic. You'll find one that is willing to sell you whatever you want. Go to an affiliate summit and go to any booth and ask them where to buy traffic that passes whatever filter you want and you will find somebody there that's willing to sell it to you. They're not hiding, you don't go into some backroom. Everything happens out on the conference floor, so it's that moment and that exposure that really ingrained in me that we need to stop approaching this as super complex criminal enterprise and actually address it as a supply chain problem.

Tom Denford: 11:49

This is not regulated, right? This is not legislated against. To a marketer they would say, I mean you're selling me things that don't exist. This is fraudulent, you're stealing my money. How can this be happening in the open and an accepted public forums?

Shailin Dhar: 12:03

It's because nobody, marketers and then down the chain, has explicitly set those standards of,"I do not want nonhuman traffic" up until recently. Even now it's I want traffic that complies with this one filter vendor that I have integrated into my company and I want it to pass their standard. You're not saying I want no nonhuman traffic, you're saying I want it to pass vendor A's verification filter. What everybody does down the chain is create traffic that passes that one vendor's verification filter. If there is a market leader that's adopted by the marketers, like a market leader verification vendor that all advertisers or the majority of advertisers are using, there will be a ton of traffic available that will pass those filters.

Tom Denford: 12:55

You've explained to me before, I found this really helpful, this story about art forgery. Can you share that again, because that for me was when I really made sense what you've just explained.

Shailin Dhar: 13:05

Yeah, so there was an aspiring artist who, what he started doing was recreating lost masterpieces of art. Picasso paintings that were documented and described, but nobody knew where these were and so he would create those and basically follow the signature points that some art critic in the past had described what made this unique; that there was a red thumb print in the top left, smudged as a mistake on this piece and he would intentionally do all of those things because all the art critics and gallery owners kind of ascribe to this common standard of these are what these look like. So when this guy came in with this lost Picasso painting and it fit all of those descriptions he was making ton and tons of money.

Tom Denford: 13:55

They had a checklist, simple checklist of things that in their mind would tell you that it was a genuine painting.

Shailin Dhar: 14:02

Right, and he recreated all these and sold dozens of these paintings and eventually got caught.

Tom Denford: 14:08

I think it's a really good example. It just helps marketers understand how this stuff gets through. Whatever verification technologies we have, what tools to help us identify fraud. Again, they really just following a check list of indicators, and so sources of fraudulent traffic or counterfeit traffic are just going to address hitting those 12 indicators to get around those and then it isn't actually human, but it's judged to be valid or judged to be sales worthy.

Shailin Dhar: 14:38

Most vendors are relying on behavioral signals that indicate whether it's programmed automated behavior or whether it's more random human-like behavior and they're constantly updating those standards based what they see in the actual live environments. It's not only marketers and media buyers that have access to that. All of these traffic vendors that create this traffic can sign up for any of these services as a customer as well, so they have direct access to what works and what doesn't.

Tom Denford: 15:08

To bring that then up to date, so you develop a method the Dhar Method, which was kind of a method of identifying or mitigating the impact of counterfeit.

Shailin Dhar: 15:17

It was more focused- there wasn't any technology behind it at the time- it was more just focused on kind of looking at a media buying process and saying, do you actually have insight into what you're buying? Most people in digital advertising and the Ad Tech space do not interact with log level files, impression level files just because of the volume and the size of these things. Most people look at PDF summaries of a media buy. Rarely do people look at,"does my DSP number, what I have totaled up as auctions that I've won in my demand side platform, does that match up with my ad server numbers?" Typically larger advertisers use two different technologies for those things. Looking at those types of things is very important and so it started out with more process focused things and just doing education on how easy it is for this traffic counterfeiting to be committed, not just talking about,"hey, this is how I solve..." You have to know what the problem is before you create a solution to address it.

Tom Denford: 16:23

What led you to set up Method Media Intelligence?

Shailin Dhar: 16:27

We got reconnected on a professional level, he's a web architect by background. We've known each other since we were eight, our families do Thanksgiving together. We always had casual conversations, but at the time he was working for a company that did web monitoring for basically pages, so performance monitoring for large, large websites. What he started to think about was how much robotic traffic can be coming into these sites and also thinking about,"oh wait, they have ads everywhere, are these robotic visits generating ad revenue?" He called me one day, I remember I was in the car in San Francisco, I just started laughing because he said,"you wouldn't believe what I'm seeing" and I started laughing and saying,"yes I would. I would believe everything you're about to tell me." We teamed up, we put our heads together for one client audit that I had coming up, it was to audit a re-targeting company. I wanted his technical help on that project because I didn't know how to build a Bot. What I wanted to show the client was that we can get re-targeted by their ads if we direct a Bot to their own page.

Tom Denford: 17:40

So what is a typical client method then? It was a consulting business originally right, when you started it? Do you think of yourselves as a technology company?

Shailin Dhar: 17:50

Yeah, for now we're primarily a technology company. We started in February of 2017 and by June we had a patent application filed for our own Bot detection methodology. That's focused just on detecting can this device, can this machine be used by a human? It was very advanced at the time, just compared to what other vendors were doing. We found that we could do this check in five milliseconds compared to some of the much higher times we saw with other vendors in the space. We used the Bot detection for different clients just as a kind of proof of concept. We didn't know what we were going to do with it. After doing lots of historical audits for advertisers, agencies, and different ad tech companies, we found that we could actually create a monitoring product where we could use our bot detection, but also combine the accounting principles from our media audits and actually implement ad monitoring on digital advertising campaigns. It was kind of merging Bot detection and verification with the accounting principles. So rather than saying you have 5% bots, we can tell you exactly how much money,$472,816.14 was spent this month on robotic activity.

Tom Denford: 19:11

What's an extreme example? Are there moments where you've gone,"oh my God, look at the state of this."

Shailin Dhar: 19:18

I think there's fewer moments where we're not still shocked.

Tom Denford: 19:22

At what? Just the scale of it?

Shailin Dhar: 20:22

It's not always malicious, but this high level of dysfunction in marketing organizations where it's a combination of politics. It's a combination of egos clashing, all mixed in with a complete lack of knowledge of how ads actually get loaded on a web-page. You know, we've done many education seminars where it's a room full of media buyers that on a monthly basis collectively probably spend$100 million. For the first time we can see it in their eyes that nobody has ever explained to them how an ad auction works, how real-time bidding auction works. So everybody just seemed to be winging it for months, if not years, because most companies in digital advertising don't have very deliberate and diligent on-boarding education seminars. It's just kind of,"hey, you came from this type of company, this is what we do. Throw you on the desk, get started." When you actually go in to do an audit for an advertiser and say,"hey, where are you spending your money? What platforms are you using and where's the data?" There's just a lot of shrugging and looking around the room.

Tom Denford: 20:38

You can understand then sometimes the reluctance of marketers to even open that conversation. I would highly sympathize if it was me. How do you deal with that? How do you deal with a marketer that's coming to you to ask this question for the first time and avoid making them feel very exposed, maybe feeling silly, asking some kind of silly questions. What's the first comfortable step that they should be taking?

Shailin Dhar: 21:02

We start with the leadership. What are the questions that you have? Every question you've been afraid to ask, every question you've avoided asking. Let's get those out of the way so that you are in line with what we're doing and you can know why we're doing what we're doing. Not just pass it down to junior management or accounts people. So we start there in a closed room with as small of a group as possible. We try to hammer down, do you have the information of: where the money is going, how it's being spent, where it's being spent, what safety measures you have in place and then go out to the rest of the team from there. This is obviously a learned process. We've done it kind of half hazard at the beginning where we went on a giant conference call and said,"who has the, DSP log files? Has anybody here ever seen something but PDF of where your money was spent?" That does not go well.

Tom Denford: 21:59

What's the reason to do this? Does it have a dollar impact? Are you finding that there's a consistent percentage where you're getting rid of waste in the system?

Shailin Dhar: 22:08

Immediately in the first month after adoption and initial education, I'd say 90% of our clients have been able to identify 20% waste that they can cut out immediately. That doesn't mean spend less on media. You can divert that 20% from something that was never giving you a return to the channels that were giving you some return or great returns in the past.

Tom Denford: 22:31

How do you keep on top of the bad guys? Presumably, every time some of them- to go back to your forgery example- as soon as another measurement happens, how do you try and keep one step ahead of them? How do you know what to be looking for?

Shailin Dhar: 22:49

I don't agree with this whole narrative that this is a constant cat and mouse game. I think you can do broad strokes but pinpoint it, like strategic identification of something that is clearly invalid, like what we do is just identifying the machines. We're not going to catch Bot net traffic if you have some malicious software on your computer that is running ads behind your current live browser, we're not going to catch that because it's replaying your behavior. We're not looking at the behavior, it's on a real computer, but there is a giant portion of web traffic that's coming from automated cloud servers that don't have any ability to be used by human. They don't have any graphics capabilities and that's what we patented. For the next five, ten, however long years, this is going to work, unless cloud servers start plugging in screens to every one of their CPUs. So if the cloud hosting industry wants to spend billions of dollars updating all of that then yes, this has become a cat and mouse game but we focused on identifying broader things that we can pinpoint and then just sticking on them repeatedly. That's the only way to create systematic change even within an organization.

Tom Denford: 24:04

For those marketers that are, I guess becoming increasingly aware of ad fraud as an issue being more talked about and I think more publicly, more frequently, which is really good. Ad fraud seems to have risen to the top. Rightly so, because it's the probably the biggest risk so it has the most potential value opportunity. But for those that maybe are now saying,"right, okay, this is something which we want to tackle as a business," what's your advice for them? What do you need to do before they reach out to people like you to start this journey?

Shailin Dhar: 24:34

A reflective taking of inventory within the organization to see how much of my spend is being monitored right now. I think one of the big surprises we see marketers have is we start looking into where is all their current monitoring and they think it's on 100% of their campaigns, 100% of impressions and it's actually only covering 15%. Somebody in the organization didn't want to rack up a big bill because all the verification services are on a CPM basis. So somebody is avoiding racking up too much of a bill because they're under other budget constraints and then the CMO finds out that you are only putting Bot detection on 10% of our programmatic buying. Those types of things, just looking at what is being measured, how much of it is being measured, what are you paying for it and can you quantify what your current process is identifying as waste? Can you quantify that into a monetary amount? The other thing is viewability has been standardized by the browsers. You don't need a third party viewability measurement company except for the reporting of those viewability numbers. Every major browser is now reporting directly which containers that they're loading are within the view-port because the browser is the one that best knows what parts of a web-page are loaded because it's responsible for loading them as you scroll up and down, as you open new tabs, or whatnot. So that's been standardized by the browsers. Safari was the last one to do that update. Google chrome has already been doing this for two years. Checking on how much are you actually paying for something that is standardized by the browsers and then brand safety. We see clients that have very strict guidelines on brand safety and they're not always fully aware of the technical limitations of whatever technology they're using of being able to prevent an airline company from showing an ad on a news article or a blog post about a plane crash. Being aware of those things is becoming increasingly important.

Tom Denford: 26:40

So for those those marketers that then want to start dipping their toe into this. Do you have stuff that you would recommend that listeners can start with to understand this area?

Shailin Dhar: 26:51

I can go in order of how nerded out you're going to get by these things. I would definitely start off with Bad Men by Bob Hoffman. Bob's been a great, not an official mentor in a sense, but great person to catch up with and just bounce ideas off of with his experience and also his skepticism of everything that's come to fruition in the last 15 years. Bad Men is a great piece that follows his thought process from beginning to end of what he thought was going to happen, what he watched happen, and then what eventually has happened. He goes over the invalid traffic. He goes over the surveillance aspect of marketing and how a lot of what we look at as very invasive tracking, originated for the purposes of targeting for different advertising campaigns. I think that's a great place to start, I never don't enjoy reading something that Bob writes. Second would be Attention Merchants by Tim Wu, who's a professor at Columbia. This was eye-opening to me just when I became aware of the book because I forgot and I think a lot of us forget that that's what this industry is focused on, it's merchants of attention. We talk about how we want these KPIs and everything is measuring"how can we take human attention and create into the purchase of a product or service?" We get too caught up sometimes in the technicalities and the specifics of our jobs that we forget that this is all centered around monetizing attention. He goes into intentions of different organizations. He goes into the specifics of corporate corruption practices that he documents along the way. It's not completely focused on digital advertising, but it's a great way to just kind of zoom out of our specific job roles and look at the industry as a whole. Then, and this is very dense material all centered around the work of a professor who's passed now, but her name was Donella Meadows and her focus was sustainability. How do you intervene in a system where consumer products have switched to all plastic packaging? Where do you intervene there? Do you go to the manufacturers who have no financial interest to change that? Do you go to the consumers who are very unlikely to add inconvenience to their lives for some marginal benefit in the environment to themselves? Where do you intervene in these types of systems? There's just great papers, great books, Thinking and Complex Systems is kind of the summary of a lot of the work, but there's also a three page paper that will include the link for that kind of summary Document of where do you intervene in a system and it ranks the optimal places of where you can have the most effect, if you can access this change point, and where are you going to have the least effect but easiest access.

Tom Denford: 30:15

I can see the parallels to what we're talking about, there's a number of different cohorts of people involved in it. Whilst everybody might agree that this is not a great thing, nobody, no one particular part of the chain is motivated to rise up and change that individually. So all of those links, very good suggestions by the way, I liked the idea of starting accessible to nerdy. I think that might be a pattern that we follow in the future. I've got a feeling that Bob's books are going to be frequently quoted as the first part of that step because he's been a phenomenal force for education and simplifying, which is a real skill. We actually have him coming on the show in a few weeks' time, so I'm sure we'll be talking more deeply about that. Go to mediasnackpodcast.com/26 to get links to some of the resources that Shailin recommended. Give me some hope then, where do we go from here? Is this going to get better? Is it getting worse? Where do you want us to be in 12 months time from here?

Shailin Dhar: 31:16

My kind of dream scenario is that the top 50 marketers in the world do a full diligent audit of where their digital media budget is going, who's monitoring it, who's spending it, who's verifying it, and what are the actual results? I mean do a full review. It can just be four people in your organization sitting around and looking at each other and saying, do I have the answers to these questions? Do I have something except a two page PDF to justify the$50 million that we spent on this platform? That's the starting point.

Tom Denford: 31:51

Even today, we may have inspired a handful of other marketers to start that process.

Shailin Dhar: 31:56

I hope so, it's not very difficult to get started.

Tom Denford: 31:58

Shailin Dhar, thank you very much.

Shailin Dhar: 31:59

Of course. Thank you.

Outro: 32:02

Who would you like to meet on future episodes? Please let us know at mediasnackpodcast.com, where you will also find previous guests, including leading media executives from companies like P&G, L'Oreal, Mars, and many more. Plus some of the industry's most provocative thought leaders, people like Professor Mark Ritson and Gary Vaynerchuk. You can subscribe to get new episodes each week, and if you liked this episode and you think somebody else would, then please do share it. Thank you so much for listening.