Is What You Dont Know about Ad Fraud Costing You Big Money? with Augustine Fou

Marketing Inc., sole author of Do It Wrong Quickly: How the Web Changes the Old Marketing Rules, and a former IBM Distinguished Engineer. Today you'll be hearing from Augustin Fu, who will present Is What You Don't Know About Ad Fraud Costing You Big Money? But before we start, we need to recognize our sponsors. BarnRaisers, a full-service digital and social media solutions company that builds brands using proven relationship principles and ROI. Gagalamp empowers companies to amplify their social media content by leveraging employees, partners, and resellers. Try Gagalamp for free today. Marketing Science Consulting Group, the art of marketing distilled into science. MENG, Marketing Executives Networking Group, New Jersey Chapter. We aren't just about networking to land a great job, we're about staying current to grow in our careers. Unison, the unagency, your secret weapon in the noisy world of 21st century brand communications. As we wait for more attendees to join, we'd like to launch a quick poll. We'd like to find out what you already know about digital ad fraud. This is just a quick question. Were you aware of digital ad fraud before this webinar? You just have to answer yes or no. And while we're waiting for you to answer that question, let's review the format of our webinar. Our Biznology webinars last just 30 minutes. You can easily fit them into your busy schedule. We record each webinar and we'll email you that link later this week. During our speaker's presentation, you can use GoToWebinar to ask a question. That orange arrow opens and closes your controls. If you have a question, simply type it into the box labeled Questions at any time during the event and press the Send button. I'll select a few questions at the end of our webinar and pose them to Augustin. While we're waiting for the last few attendees to join, I'd like to remind you that the Biznology newsletter and blog are available for free at biznology.com. So if you're not already a subscriber, we hope that you'll sign up now. Looks like we have about a 60/40 split on the poll results. It looks like 60% of you knew about digital ad fraud, but for 40% of you, this is a fairly new subject. So, um, I think all of you are going to be in luck because Augustin is an expert in this area. So thanks for spending 30 minutes with us. We know how valuable your time is. So I'd like to introduce today's speaker. Dr. Augustin Fu is founder of Marketing Science Consulting Group. He's also applying advanced technical forensic techniques to help clients detect and mitigate a variety of forms of digital ad fraud. So if you've ever struggled with online ad fraud, this is the webinar for you. Augustin, take it away.

All right, thank you, Mark. Let's see, let me start by saying I'm going to give a very brief overview on digital ad fraud, and let me show you a couple of the items we're going to touch upon today. This is not meant to be very in-depth or comprehensive. It's really just to expose the audience to the different kinds of ad fraud that I've seen in the marketplace so far. So we'll start with the background, and then we'll hit upon each of these different kinds of ad fraud, starting with impression or display ad fraud, then search click fraud, affiliate revenue share fraud, lead fraud. And finally, we'll talk about some of the industry initiatives that are going on right now to start to combat this problem. In digital advertising. And then we'll take a quick look ahead. So starting with the background, you know, you might be wondering why all of this is happening now. One of the key reasons is that as more and more of the digital ad inventory is being bought and sold through ad exchanges, a lot of this is being done programmatically or using algorithms with little or no intervention by human media buyers. So as more and more of that inventory runs through ad exchanges, it's giving the bad guys more opportunity to create these fake impressions and sell them into these ad exchanges to big advertisers who are willing to buy it. And it's also making it far easier for them to hide in a larger haystack. So that's one of the key trends that's enabling a greater quantity of fraud. Furthermore, the industry and a lot of the big brand advertisers are becoming aware of this, and finally we're starting to see them take action. Some industry bodies like IAB, ANA, they're starting to actually try to figure out as an industry how to deal with this kind of fraud. It's actually quite sizable. This was relatively eye-opening to me when I started researching this, and various different companies have published their own perspectives on the size of the ad fraud. So if you can look at the numbers on the right-hand side, it's almost so large you wouldn't even believe it. But different companies in different industries are subject to different amounts of this fraud. So as you can imagine, certain industries like financial services or pharmaceuticals, because they have very large ad spend in these areas, they are prime targets for the bad guys. So let me go into the rest of the deck and talk about how some of this kind of fraud is being perpetrated and also try to size the market for you just so you can get a sense of how big of a problem this is. And ultimately, we can talk about some of the strategies and tactics that you can employ to to start mitigating this. So we'll start with impression fraud, and I'll talk about how it works. So it's relatively straightforward. And as you might have seen earlier, this is probably the easiest kind of fraud to commit, because what the bad guys do is they put up these websites, and they stick a whole bunch of ads on them. There's practically no content on these sites. It's pretty much filled with ads. And if you notice on the right-hand side, we saw one instance where There were 85 ads on the same page. And then, so these are just a few more examples of how they load up all the ads on the page. Some of which, as you can imagine, are below the fold. These are so long of a page, I had to turn them sideways. And what they then do is actually use bots. So you've heard of botnets and all that kind of stuff. They're using these bots to just repeatedly load these pages. And as these pages load, the ad impressions get generated. So what that results in is that they now are creating a large quantity of inventory. So these 2 are examples that were cited by an Adweek article, where these sites are selling tens of millions of impressions into ad exchanges per day. And if you actually look at their traffic via Compete or Quantcast, there's hardly any traffic going through these sites. So that's a key red flag that you guys should look for in terms of figuring out which sites are legitimate and which sites are not. So these guys are selling enormous quantities of ad impressions when there's actually virtually no human traffic or users going to the sites. So now given that there's a large amount of this kind of fraud going on, let's take a look at how large in dollar amounts this would be. So if we take the range of 30% to 75% are fraudulent, or not in view, or ad-blocked, and you apply that to the IAB numbers of display ad spend, which is just under $8 billion. Even if you take the low end of this spectrum of 30%, that's still $2.5 billion worth of ad spend that are not seen by humans. So that's quite a sizable amount. And so let me show you some of the background research where we got these numbers from. So Solve Media is a company that provides CAPTCHAs, and basically the theory is that only humans can solve CAPTCHAs. So from their research, they're showing that roughly 24% to 29% are confirmed bot traffic hitting websites. So that's where the 30% comes from. And up to about 46% are suspect or suspicious kind of traffic. And then there's other evidence like this where we're using data from Alexa and other places where we can see the traffic either goes up very quickly or goes down very quickly. So that again is indicative of bot traffic because humans going to a website, the traffic curves are much smoother. They're looking for content, they're reading the content, things like that. In this case, it's basically a traffic fire hose that the bad guys are generating using their botnets. And then it's been corroborated by other companies, so in this case Integral Ad Science. They're showing where the fraudulent impressions are coming from. And as you can see, I put a rectangle around exchanges and networks. Those kinds of ad exchanges are places where we see a greater proportion of these fake impressions. Because again, the bad guys sell large quantities of impressions into the ad exchanges. Now video ads are also similar to the display ads because the revenue model is still on a CPM basis. So for the bad guys, because the video ad impressions are much much more lucrative, they're now pointing their botnets or traffic fire hoses to the video ads. And so it's causing a much greater— it's kind of an explosion in video ad views. And a lot of this is not corroborated by the number of humans actually watching that many more videos online. So a lot of this is already suspicious from the beginning. And there's a corresponding shift of a lot of ad dollars into video ads online. So the big brands are taking some of their TV budgets and putting it into digital video ads, because those are most similar to the TV ads themselves. So you see this eMarketer chart to say that there's going to be a huge or continued increase in digital video ad spending by the big brands. So again, it's making it very, very lucrative for the bad guys to commit fraud in terms of video ads. So kind of to put this in perspective and really try to size the dollar amounts, in terms of display ads, there's hundreds of billions of display ads shown each month. And if you take 30% of them to be confirmed bot traffic, we're not even talking about the 46% that are suspicious. If we just take the low end of that, and then you multiply it by the typical or average CPM of $1 to $3, then that's where you get the $2 to $6 billion of annualized wasted ad spend. And similarly, on the video side, even though there are far fewer impressions, the fact is they're almost 10 times more expensive on a cost-per-thousand basis. So again, that still leads to quite a sizable amount of wasted ad spend. So again, like I said, this is just a very quick overview on the different kinds of fraud. And if you have any questions, please do type them in, and Mike will share them with me later. And now let me get into the search fraud side of things. So let me show you how the bad guys commit this and what they do. So they first start by choosing the most expensive keywords. So these are just a couple of examples of CPG companies that have relatively expensive keywords like cosmetic facelift. That's about an $11 CPC. Residential home cleaning, that's about a $10 CPC, and the list goes on. And once they choose those expensive keywords, they create these sites and they type in those keywords into their own sites in order to generate the search ads. Okay, so once they type that in, you'll see that in this case, an olay.com ad shows up in the first position. And you can see they're called sponsored results, right? And then furthermore, when you click through on that, you'll see that they actually go through a ton of redirect links. So some of these are used to commit other kinds of fraud, but nonetheless, it redirects through all these different links and then ultimately ends up on olay.com. So for the advertiser, it seems like they're getting traffic from search ads. But what you'll notice if you look a little bit closely on the right-hand side, these bad guys are also passing fake URLs. So when it says there, UTM source equals MSN and medium equals CPC, to the advertiser it comes through as, oh, that click was sent from MSN, when clearly it was not sent from MSN. So now the bad guys are even hiding themselves by passing fake URL parameters. So that being said, even though search ad fraud or click fraud is a little bit harder to commit because it's not just loading a page with ads and just continuously hitting the page with bots, they actually have to pick the keyword, load, type in the keyword to load up the search ads, and then click the ads. Now because of botnets, even that entire process can be automated. So that being said, let's try to size the problem. So in this case, the range you'll notice is a little bit lower than the 30% to 75%. It's roughly 20% to 40% from all the different studies that we've seen and our own data. And a lot of this is corroborated by other companies' research. So if you look at roughly 20% of the clicks from this company called Click Forensics, they actually showed that roughly 20% of the clicks could be considered fraudulent. And then here's other data from Spider.io that shows where the clicks occur. And so you'll see what would normally— what would look like normal human clicking because they would tend to click on buttons and place the cursor in text boxes for searching and things like that. Whereas the bots really don't even disguise hiding where their clicks are. Okay, so some of these are are great visuals of the difference between humans and bots. And then in this case, this was for one of our pharma clients. They allowed us to publish some anonymized data. And you'll see that before and after. In the before picture, you'll see that the top 2, quote unquote, good guys represent only 76% of the ad spend. So in this case, they wanted clicks to come from MSN and Bing, and they also wanted clicks that come from WebMD. But when you look at some of these other sites in here, like HealthyWeb, MedicineNet, and so on and so forth, some of those are the potential fraudulent sites. And so after we were able to kind of blacklist those sites and take them out of circulation so that they're not siphoning off the search ad dollars, we see that the top 5 good guys now represent 94% of the total ad spend. So in fact, the client did want their their clicks to be coming from WebMD.com, EverydayHealth.com, because those are legitimate mainstream sites. So by blacklisting some of these bad guy sites, we're able to shift about 18%, almost 20% of their ad spend from fraudulent websites to the good guys where they actually wanted their traffic to come from. And so if you think about a user clicking through on a WebMD ad, those are the kinds of users that you actually want. Obviously not the bots. Now if you look at some of these other categories, these industries, insurance companies spend a ton of money on search ads. And in some cases, their CPCs could go up to $60 to $78 per click. So again, like I was mentioning earlier, certain industries are particularly prone to this kind of fraud because the bad guys are targeting them. So let me just show you a few more examples of these bad guy sites. In our research, we've now cataloged thousands upon thousands of these fake websites that are put up strictly to load ads and click on them. So there's actually no content. So if you're a human and you actually got to this site, there's actually no content for you to read. And then you're kind of wondering, how is it possible that some of these sites are sending through thousands upon thousands of clicks per month? So again, these are examples of the bad guy sites. And here's a screenshot of one of them. So they build in certain keywords right into the site itself. And you'll notice the domain name also has certain keywords like cancer. So again, these guys are targeting pharmaceutical companies. But if you look across Alexa, Compete, and Quantcast, there's absolutely no data, which means that there's no detectable traffic to these websites. And furthermore, there's hundreds of them. So this is a screenshot where we took just 6 different websites. And they all look the same, because these were generated by algorithms or bots. And so they didn't even bother to make them look different. And then here's just more examples, hundreds upon hundreds of sites that are put up by the bad guys. So in these last 2 sections, so we just talked about display ad fraud and video ad fraud, which are on a CPM basis. And then we talked about search ad fraud, which is based on CPCs, cost per click. And then now for the last 2 sections, I'm just going to mention briefly affiliate sales and lead gen and how those things are committed. I think— let's see, here's where we do a second poll. Let's have Eileen put that up.

Okay.

So let me read that out. If you knew about ad fraud at the beginning of this webinar, what proportion of your spend do you think was impacted by fraud? So I'll let the audience have just about 30 seconds to just click one of those answers. And then we'll show the results.

OK.

Okay, hopefully you all had a chance to click vote. Perfect. Okay, that's great. So from the results, I think most people think about a quarter. So that's about right. You know, it corroborates, you know, what we've seen before, and, you know, it's kind of on the low end of the spectrum. You know, sometimes we see it's much more egregious than others, but that's a great indicator or feedback from the audience. Thank you. And so let me just quickly show you 2 more categories of fraud. So this is affiliate sales. So for example, when a site sends through a click to eBay or to Amazon, then Amazon gives them a rev share for their efforts in promoting. But what we've seen is— these are some examples. This was relatively recent. In terms of eBay, this person was found guilty of creating affiliate sales fraud. And he was able to siphon off roughly $2 million of revenue share from eBay. And finally, he was caught and convicted. And examples of how this is perpetrated would be web pages that are generated, again, by algorithm where they assemble the content. And in fact, if you look closely, The content is actually not human readable. So it is really just keywords meant to trick the search engines to deliver traffic to these pages. And once a person gets to the page, they're going to have these affiliate links embedded throughout the page. So when that person clicks through on that link, then the cookie is planted. And so that particular site gets credit and therefore gets a revenue share from Amazon or eBay or any other site. But what the bad guys are starting to do now, and again using their botnets, they don't even rely on a human visitor to get to the site. They're in fact starting to plant cookies without the user's knowledge. So when you visit certain sites or you accidentally click through to a site, they may actually do what is known as cookie stuffing. So they stuff the cookie that gives them credit for having sent that person through to Amazon even though they never actually sent that user through to Amazon. And when they do that, they're basically stealing revenue share dollars from the retailer like Amazon. And then these could actually get to be quite expensive. So if you look at some of these revenue shares on bigger ticket electronic items, so some of these revenue shares could range from $25 to $78. So imagine doing that for hundreds of thousands of websites. So that's starting to amount to a sizable amount of dollars listed. And then finally, lead fraud. Again, we're estimating this in the range of 20% to 40%. And this is a case where when someone fills out a form, like they are pretending to be interested in a mortgage, they will be asked to fill out the form. And then because the banks pay out on a cost-per-lead basis, so sometimes they would pay $100 for a completed lead form. In other cases, they might pay $300 for a completed lead form. Guys first employing a lot of low-wage workers in foreign countries to fill out these lead forms for a dollar apiece. And basically, they would sell those completed lead forms to those lead aggregators and ultimately to the big banks who are trying to get mortgage leads. Okay, so some of this stuff again was perpetrated by the bad guys using humans to do the work. But increasingly, again, using bots, they're able to fill out those forms and even have data that passes the rudimentary checks that the advertisers do, like was it a legitimate address, postal address in the US somewhere. So a lot of this kind of fraud is happening as well. So let me kind of wrap up by getting into the industry initiatives. And so there's a bunch of different players. Some are technology-oriented. Some are industry bodies. That are helping to both quantify and also put in place some mitigation mechanisms. So let me show you some of these countermeasures. So we talked about blacklisting of websites. So in these cases, you would want to exclude these bad guy sites from your search ads, your display ads, so your ads won't show up on these sites. The second would be enforcing viewability. So for example, if your ad shows up below the fold, You don't want to be paying for that. But again, the bad guys now stack all the ads above the fold, and they're able to defeat something like viewability. Detecting bot traffic. So companies like Spider.io have algorithms to help detect the bot traffic. But again, if we have good guy algorithms detecting bad guy algorithms, it's still a continuous arms race. And finally, use CAPTCHAs. So again, the theory is that Most CAPTCHAs can only be solved by humans. So that's a great way of distinguishing bot traffic from human traffic. So finally, let's just talk about a couple of ways you can think about reducing ad waste or ad fraud in your own ad spend. So in the bottom part of this pyramid, we see that these are traffic and visits that are created by bots. So I call that really— it's mainly about waste reduction. And then at the top of this pyramid, if you can take care of the waste reduction, then you can improve your optimization. So let me be a little bit more specific. So remember the data from Solve Media, 24% to 29% are confirmed bot. So if you can start to blacklist those sites and exclude them from your search ad placements, then that would be a good thing. And then furthermore, as you move up the pyramid, you can improve your targeting. Improve your viewability and things like that. And then finally, this is just a quick slide on how we do our process. We typically go in and do a preliminary scan for clients to figure out the size of the problem. And then we dig deeper with our forensic analysis techniques using some of our technology tools. And then ultimately help them come up with, you know, starting with a blacklist and then a whitelist and other things that they can do to mitigate the ad fraud that's impacting their digital ad spend. So with that said, some more related articles you can look at on your own time. And then let me turn it back to Mike.

Thanks, Augustin. I'm sure our attendees have a much stronger idea of how to prevent ad fraud in digital advertising, but you didn't answer every question. I've got several good questions teed up for you. And I'd like to remind our audience that it's not too late to ask your own question by typing it into the questions Before we get to the questions, I'd like to remind you that the Biznology team provides JumpStart workshops for corporations who want to apply what they learn directly to their own business and have their teams create a plan for executive approval right in the workshop. Our instructors will guide you step by step and help you to persuade that executive to say yes. Rob Peterson, Tim Peter, and I teach the JumpStart workshop for search marketing, where within 1 week we can work with your team to create a plan for search marketing. Great. With your company to give hands-on training and knowledge in digital marketing strategy and an SEO plan for a full year. Go to biznology.com/training for more information on all of our offerings. We're about to start firing questions at Augustin, but we need to thank our sponsors once again. BarnRaisers, a full-service digital and social media solutions company that builds brands using proven relationship principles and ROI. GagaLamp empowers companies to amplify their social media content by leveraging employees, partners, and resellers. Try GagaLamp for free today. Marketing Science Consulting Group, the art of marketing distilled into science. MENG, Marketing Executive Networking Group, New Jersey chapter. We aren't just about networking to land a great job. We're about staying current to grow in our careers. Unisent, the unagency, your secret weapon in the noisy world of 21st century brand communications. Now, on to your question. Our first question for you, Augustin, is, should I just stop advertising, just spend my budget on other things?

Obviously, the answer is no. You can't stop your advertising, but obviously there's ways to improve the— what I would call the productivity of your ads. So again, if you remember the chart above, let me go back to this. Pyramid, if the bottom third is generated by bots or otherwise considered fraudulent, if you start mitigating that and removing that, that can show immediate impact and immediate uptick in your ROI. So the waste, kind of reducing waste, would be what I would call the lowest-hanging fruit. And companies can't stop advertising. They really You don't need to do that, but by being aware of these kinds of fraud, then you can actually take steps to make your advertising much more productive.

Okay. Next question is, if I just use only Google search ads, am I okay?

Yeah, that's a great point. So, with Google search ads, and specifically the ones that show up on Google.com, Those actually are fine. And that's because the bad guys don't have a way to get money out. So where we see the greatest proportion of ad fraud is in the greater display ad network where the display ads are shown on quote-unquote third-party sites. And also on the Bing side, in the Bing ecosystem, they have something called the Search Partner Network, which are millions of sites that carry Bing ads. And that's where we see the largest amount of ad fraud. Because again, those are the sites that are put up by the bad guys. And so whenever they generate a cost per click, they get a revenue share of that. So that's how they can earn the money or get money out of the system. So if you really just wanted to focus on the main sites like Google.com or Bing.com, that would be a great way to do it. Now, in some of my clients' cases, like the pharma companies, they do have to advertise on WebMD.com. And WebMD would be an example of a third-party, or a search partner in the Bing Search Partner Network. So they can't just turn off Search Partner Network. They have to more carefully select where their ads get placed. And what we talked about before was blacklisting these known bad guys. And another kind of tip along these lines are that if you're noticing an egregiously high click-through rate, Something's wrong. That should be a big, huge red flag for you guys. Sometimes we've seen 49% click-through rates, and clearly that's a bad guy. So again, if you look through some of your own data, there's some obvious red flags that you can start cutting out right away.

Well, that's all the time we have for today. Thanks to you, Augustin, for these great ideas, and thanks especially to our audience for your participation and your questions. If any of you had questions we did not have time to answer, you can email your questions to eileen@mikemorangroup.com, and she'll be sure to get them to Augustin for the answer. Later this week, we'll send you all a link to the recording of this webinar to listen to again and to share with others. We invite you to mark your calendars for our next Biznology webinar, Jumpstart Your Content Marketing Through Better Market Segmentation, with Monique DeMaio, scheduled for 11:00 AM US Eastern time on March 11th. We hope to see everybody back here then. Bye, everybody.

Thanks, Mike. Bye, everybody.