An article from Bloomberg Business this September, “How Much of Your Audience is Fake” caught our eye because it highlighted how much money is being wasn’t on robot (bot) traffic. We tracked down our friend and ad fraud expert Dr. Augustine Fou to talk with us about the article and some of his new research on bots and how to avoid them.
Eleven percent of display ads and almost a quarter of video ads were “viewed” by software, not people. According to the ANA study fake traffic will cost advertisers $6.3 billion this year. Click To Tweet
Resources
- Previous podcast on bot traffic trends
- How to set up advanced podcast measurement
- Understanding your web traffic with Goals on Google Analytics
Find the podcast on:
Transcription
Episode 40
George Weiner: This is Using the Whole Whale, a podcast that brings you stories of data and technology in the nonprofit world. My name is George Weiner, your host and the chief whaler of wholewhale.com. Thanks for joining us.
This is episode 40 of using the whole whale thanks for joining us. This week we are doing a second shot at love with bot traffic. That’s not actually the title. It could be though. We’re talking with Dr. Augustin Fou because of the amount of just recent news that we’ve been reading about, you know, the increase in bot traffic, especially around advertising when people are paying for low-cost advertising they’re actually just paying for bots.
As a primer before we jump into the interview, we talk about a particular metric: “bounce rate” which is something you can find in your Google Analytics which measures web traffic but bounce rate means the percent of people that just go to one page of your site and then immediately leave. Super important that you understand that because the indicator there is that they didn’t find what they were looking for. One note about bounce rate is that it doesn’t matter how long that you spend on that one page, based on the way google analytics is setup out of the box. So you could spend one second or one minute, if you only go to that one page then you’re going to count as a bounce. There are ways to change this in your advanced configuration but that’s a podcast for another time. For now let’s jump into the interview with Dr. Augustin Fou.
Welcome back. Dr. Augustin Fou thank you for joining us again because I don’t think we can get enough of you. How’s it going?
Dr. Augustine Fou: Great. Thanks George. It’s great to be back on with you.
George: Yeah so I’ll start this off by just asking, you know, what’s keeping you up at night?
Dr. Augustine Fou: You know, it’s bots. So we’re seeing a lot of coverage, especially this year, where the not only the industry trade associations but there’s more and more people you know actually being quoted in the news. Like for example Kellog, and they were quoted in that recent Bloomberg Business article. So they’re taking steps and they’re proud of it so it’s all good but you know bots and digital ad fraud are still what I research and what keeps me up at night.
George: Yeah and that’s why you’re…that’s why we brought you back on here because I think there’s a lot of you know confusion and obviously fear around this idea of when we do spend money, like, are you kidding me, we’re gonna spend and just get taken advantage of online. Is this every platform? Are we just doomed if we’re spending on Facebook or Twitter or you know Pinterest? What are we talking about when we’re talking about ad fraud in terms of venues?
Dr. Augustine Fou: Yeah, I would say it’s typically the smaller sites that belong to very large programmatic ad networks and basically you know some of those terms may be foreign to folks. So programmatic ad networks are the ones that bring together tens of thousands of small websites if not hundreds of thousands of small websites and they place ads across all of them programmatically. They use algorithms to do that. And when you start to get into these really, quote-unquote, “long tail” websites, which are very tiny in and of themselves, very few people are actually going to visit them or check them out before they’re led into these programmatic ad networks. So when you have hundreds and thousands of these, the small websites that are carrying ads and they’re all being done programmatically that opens the door to fraud. And the fraud that we’re talking about, I mentioned bots before, the fraud is basically committed by the bots. So the bots are browsers that are programmed to go to websites and when they go to the website and cause the web page to load they’re also causing the ads to load. So when you’re an advertiser and you’re paying on simply the number of add impressions served, right like cost per thousand, so you’re buying billions and millions of ad impressions. Then when those ad impressions are being caused by the bots then those are ad impressions that you don’t want to pay for and, you know, if you knew about it. And so that’s what I mean by fraud and it’s being caused by bots and it’s actually being made easier by the rise of programmatic ad networks. So I would say in those cases that’s where you find a greater incidence of fraud vs some of the more mainstream properties like on Google.com or in Facebook or on YouTube and really the main reason for that is the bad guys don’t have a way to get the money out, right. When the ad is shown on Facebook, Facebook makes the money. So you know pretty much in those kind of walled gardens it’s usually a lot cleaner than when you go into these really long tail websites and your placing ads in places that you don’t know about.
George: So ironically Facebook’s greed in this case is protecting us. Thank.. thank you?
Dr. Augustine Fou: Yeah I mean some of those walled gardens are actually better. Well I mean if your objective is to show your ads to humans, right? So in certain cases when the requirement is simply to get more ad impressions, no matter where you get it, and get it more cheaply. That’s what’s led to a lot of these things. Because these programmatic ad networks are able to sell the traffic and ad impressions extremely inexpensively, right? Again it’s covered by the Bloomberg Businessweek article.
George: Yeah let’s talk about that actually some more here. The title of the article is “Marketers thought the web would allow perfectly targeted ads. Hasn’t worked out that way.” That came out September 2015.
Dr. Augustine Fou: Yeah, very recent.
George: It featured some scary things. You know, they start off by talking about Heineken and the fact that 80% of like, like a hundred and fifty million dollar range type of budget just…poof!
Dr. Augustine Fou: Yeah.
George: How does that happen?
Dr. Augustine Fou: Well, I mean it’s very scary to think that that is even possible but it’s because a lot of these advertisers have brought what I call a “TV mentality”into digital. SO in TV it was absolutely true that when you increase the region frequency you’re gonna actually increase the business results. Right? ‘Cause you’re getting your ad in front of more people more often and so when they see it and they become aware then you are going to drive sales. In digital it’s really much more of a performance medium, right? So it’s not so much about the quantity that you put out there. It’s actually more important that you figure out the quality and getting your ads in front of humans as opposed to these automated browsers called bots.
So in digital you know one of the main kind of theses of the article is that, oh, it’s even calling into question the whole promise of digital. Right, like you said digital was supposed to be this you know, medium that you could use to target very well and see the results immediately. But it hasn’t…it seems to have not lived up to its promise. But I think obviously it still could if people use it in the right way. Right so instead of just asking for region frequency, more ad impressions, how many billions of ad impressions can you by for us? Right? It’s really are those ad impressions, or are those advertising tactics, actually working in digital? So really when I say “is it actually working?” there are things beyond the number of impressions served or the number of clicks you got. There are other KPIs that are actually better for measuring whether it worked or not. So the obvious examples would be: did you make a sale or not? So they’d be translating that into the nonprofit world it’s gonna be did you get a pledge or did the visitor actually make a donation? Right? Those are things that nonprofits care about. Or, did the person sign up some kind of event or some kind of…something like that. If those things are not happening then.
George: You have a couple of examples. Right? Where…the game though really is that right look…it’s really sexy to say we had a hundred thousand people, sessions, come to our website last month.
Dr. Augustine Fou: Yeah.
George: Right? That just looks good.
Dr. Augustine Fou: Yes.
George: But if you actually are getting after things like you know pledges, donations, things like that…can you share some of the you know, things and experiences that you’ve had recently with that?
Dr. Augustine Fou: Yeah, yeah. So not anything specifically but I can just tell you that there were cases where you know literally 60%… I’ll share two cases. One was 60% of the traffic to one particular website was fake and the way we could tell was when we cut that off literally by blacklisting some websites that were showing the ads and sending the traffic we saw absolutely no change in the number of pledges that we got on that website. So you know if you look at the, just the traffic coming to the website without asking the one question more which is “did it actually drive conversions?”or in this case pledges. Then, you’re right. You know like a lot of people have been historically just reporting on traffic. “Oh yeah we got this many people…and when we increased ad spending we got more people!” But unfortunately those are not necessarily people who came to your website. So again it goes back to something I said on the last podcast that we did together. You know, bots don’t make donations. Bots don’t convert. So there are some specific measures or KPIs, “key performance indicators,” that you should be looking at that pertain to what humans would actually do that bots have no incentive to do. So in the second case we saw exactly the same thing: when we cut out a very large proportion of the traffic we didn’t see any change to the number of conversion events that we had on the website.
So we started by looking at the analytics. So again it’s not something that would cost people a lot of money. So a lot of these non-profits, literally if they had Omniture, Webtrends, or Google Analytics installed they can look there and start to look at the traffic sources that have 100% bounce rates or zero time on site. Right? Something’s wrong with that. And when you’re seeing those domains that are sending that kind of very low quality traffic then you can tell your ad serving partners or your ad networks to blacklist them i.e. don’t serve ads on them. So you know those are things you can actually see in your own analytics and those are steps nonprofits can actually take right away. And that’s kind of the low hanging fruit that they can do first before they get into more sophisticated measurement techniques.
George: So as traffic become cheaper and cheaper let’s say…you know, the board members are like “We want a website that’s getting a quarter million people.” How much would that cost me if I just didn’t care at all?
Dr. Augustine Fou: So…you know these days again you know citing a lot of different sources, the most recent one being the Bloomberg article, you can buy traffic for five cents a thousand. A thousand hits. It’s so inexpensive because it’s practically free to generate for the bad guys. So they’re more than happy to sell it for extremely low cost. So it literally goes back to the idiom “you get what you pay for.”And so if you don’t care if it’s human or not and you just need the hits to a website bad guys can generate as much of it as you want. That includes visits to websites, clicks on whatever you need clicked on. Right? So for some of these clients that I’ve seen even if they make the metrics, things like visiting a particular page, executing a zip code search or watching a video, all of those things, because it’s involving just a click, any of these automated browsers i.e. bots can do all of those things. So imagine a scenario: What if the visits were fake? And what if the clicks were fake? The bigger issue there is that all the analytics that you have been looking at are also completely messed up. Because if all the clicks are fake and created by these bots then you’re optimizing for the wrong things. You might be thinking that “oh yeah I’m optimizing for the source that gives us the most clicks” but unfortunately because those clicks are also fake then you’re optimizing towards the wrong things. So a lot of that kind of stuff. If the people realize that those could be faked then they’ll be a little bit more vigilant in looking for other metrics like the donation or the things that bots have no incentive to fake.
George: Yeah. Tracking actual user behavior.
Dr. Augustine Fou: Correct.
George: The report that is cited by the Association of National Advertisers, “The Bot Baseline: Fraud in Digital Advertising, they claim that it’s gonna cost advertisers 6.3 billion in this year.
Dr. Augustine Fou: Yeah!
George: I don’t know I guess at this point why point three and not point seven? That’s a lot of money, right?
Dr. Augustine Fou: That’s a big huge number, right? That’s very scary and…it is very scary. But…those are kind of industry average numbers and if you look more closely at that report it’ll say that display ad impressions i.e. “banner ads”have about an 11% bot rate, or fraud rate. And video ads have double that at about 23%. So about a quarter of the video ad impressions are fake. Now given that is an industry report you can probably imagine that it is conservative. But bottom line is you know the numbers could be all over the place. In my research I’ve seen it as high as 99-100% bots. And I’ve seen other cases where it’s 0-1% bots. So all of these kinds of industry numbers are generalizations and they should be treated as such. What really matters is the bots and the fraud rate that is impacting your media buy or that’s coming to your website. So really the key take away from that is that, yeah those numbers are big and huge and scary but what matters is if, you know, what’s impacting you. So the key takeaway is to actually go measure it. So if you don’t know, taking those generalized numbers is really not gonna help you. It doesn’t tell you what to go do about it. But what you should do is actually measure the rate of bots that are coming to your websites. Measure the rate of fraud that’s impacting your media buys, right? So some of these non-profits have relatively large media budgets to generate traffic and things like that. So if again if a large percentage of that is fake, i.e. shown to bots and not to humans, all of that’s going down the drain. So they could do a lot better with those dollars if they are able to measure the bots and start cutting off those websites that are sending the bots. So, you know, those are industry numbers, it gives a good sort of benchmark to look at, but again you know big scary numbers don’t help if you’re not actually measuring it for yourself and then being able to take action on it.
George: Yeah this obviously comes back down to measuring what matters and making sure that the people at the management and senior levels are not just pushing for total impressions because be careful what you asked for. You know? Twelve dollars later you’ve got hundreds of thousands of bots coming to your website.
Dr. Augustine Fou: Right, or fifty cents later times a lot. Right, so…yeah it can be very very cheap. So it’s really important for them to measure it for themselves.
George: But you know we’re talking about, I mean I guess the two of us are pretty familiar with the types of different advertising out there but who on earth is buying banner ads like junky video ads? I mean like safaris is gonna come out with a like a built in pop up blocker and banner ad blocker. I mean it’s archaic.
Dr. Augustine Fou: Well, you know again because they come to the digital media they think of the digital tactics with a TV like mentality of region frequency that more is better. And they think oh well if I’m able to serve up 10 billion banner ads somebody’s got to see them. So they still have hope that it might have some kind of branding effect. But again you probably know there’s a lot of pop up blockers and the estimate of the number of people using them range between 9% and 60%. So again those are big huge numbers but again if you’re measuring for whether the visitor that came to your website did anything then you can actually very rationally approach all forms of digital media. Right so some banner ads are still useful, right? If you have something that needs awareness banner ads are great. Or video ads, again those are for awareness. Or mobile display, mobile banner ads are also good if you need awareness. But there’s other forms of media in the digital space like search marketing where it’s much more targeted in the sense that that ad shows up when a person is searching for something. Right? And the ad is based on the key word that the person is typing in. So if the companies are not using that as well then they’re probably missing out. So each company should look at what their needs are. If their need is awareness then some of these display ads, video ads, you know those forms of media, are good at getting the word out. But if they need to harvest leads then they should be doing things like search optimization or paid search because those ads are shown when the user is looking for something. So again the right mix is really the key for any advertiser.
George: This kind of reminds me of like out housing bubble of when you had overrated junk bonds being thrown together of mortgages. You have, basically, junk in the mix. And they talk about bulk buyers and I guess if you don’t actually see sort of that one spot you’re ending up in you’re buying sort of aggregate…one of the tips may be, you know, go to the source and don’t trust this sort of aggregate.
Dr. Augustine Fou: Yeah. Actually look at it. Sometimes if you sort of delegate that responsibility to the media agency and the media agency delegates that responsibility to some anti-fraud vendor pretty much nobody’s looking. And that’s how all this fraud is still occurring. If someone actually bothered to look. So I get back to really the free thing that a lot of these non-profits can do is go look at your analytics. Right? In the analytics if you go into the sources of traffic and you see sources that a100% bounce rates or zero time on site and those are sending tens of thousands or hundreds of thousands of visits then those are the ones you should cut off right away. You add them to your blacklist, you tell your media vendor don’t place ads on those sites anymore.
George: Yeah and so being consistent, looking at…going to the source of your ads is a practical tip. Look at your analytics and just understanding, having Google set up and making sure the conversion rates are actually there. It’s kind of like the adage of like “follow the money.”Follow the user…or bot”
Dr. Augustine Fou: Yes!
George: Or buyer…so the interesting things here are that it seems like it’s easy enough if you’re paying attention to stay away from this type of ad fraud. There are action steps to do it. And you know as you look towards other types of things and preventive measures, who are the big actors in the field right now that are staying on top of this?
Dr. Augustine Fou: Well I would say the industry trade associations are really doing a great job educating the market about it and also taking some steps. So, the IAB (Interactive Advertising Bureau) has been very vocal about it and they’re taking steps right now to help both the publishers and the advertisers not only become aware but start taking action against bots and fraud. But I think you know again the message is don’t leave it up to an industry trade association or somebody else to do it. You really have to do it yourself. Because I think some of these topics and issues are so new, because you know programmatic really isn’t that old, and these kinds of fraud that are kind of made easy by programmatic advertising, again, is brand new. So everyone is learning as they do it. So if you pay attention and you’re looking at it and you’re reading about it then you’re probably as knowledgeable as the next guy. So I think it’s really take matters into your own hands. Don’t delegate or don’t think that someone else is taking care of it for you.
George: Yeah, solid advice. So, as we move to close here any final bits of advice or things that you foretell for 2016?
Dr. Augustine Fou: I think we’re gonna see good things. I think the amount of fraud will continue to be mitigated so I think finally we’re seeing what I’m gonna call “countervailing forces” which is the good guys actually fighting back, harder. It’s been so easy for the bad guys to just make off with all the money. Now at least the good guys are putting up a concerted effort. So if you’re fighting individually you’re not going to be able to do too much but it’s like, let’s take care of the low hanging fruit first. I mean, what we said earlier: look at your own analytics, there’s probably some obvious things in there that if it doesn’t look right it’s probably not right. But I also see the industry getting together and making a stronger, coordinated push against fraud. So as these new forms media continue to grow by leaps and bounds I think we’re gonna see good things next year in terms of the amount of fraud being kept under control and hopefully sooner than later start to go back down. So it’s really all about countervailing forces and I’m glad to see the good guys fighting back.
George: Yeah, let’s go good guys!
Dr. Augustine Fou: Absolutely.
George: Alright, so as we move to close: how do people find you? How do people help you?
Dr. Augustine Fou: I’m available to help out. As I said this my area of research so I focus on not only digital marketing but more recently on the studying of the fraud and how it’s committed. So, I’ve helped a lot of clients including nonprofits, first measure, that’s really the key thing. Let’s see how big of a problem it is and then decide what to go do. So it’s just my first name dot my last name at Gmail. Anyone can reach me or if you google my name you’ll find a lot of the articles and slideshows that I’ve written in the past that talks a little about how these kinds of fraud are committed and gives examples.
George: Awesome. Well, as always thanks for joining and sharing your knowledge.
Dr. Augustine Fou: Great. Thank you very much George.
George: I love talking with guests that have real world examples and data that help us navigate these waters. The data Dr. Augustin Fou shared with us from you know like real non-profits out that are experiencing this ad-fraud traffic. The moral of the story here is be careful what you pay for. If it sounds too good to be true it absolutely is. You always have to be asking “what do the data say?” And you’re really just looking for human behavior, right? What’s acting like a human vs not and kind of just be honest with yourself. I had a moment early on in the days of this podcast, literally the first month of launching it, where the way I built the feed actually for how we distributed podcasts was being hit by bot traffic. And so my numbers were skyrocketing. I was doing thirty thousand downloads every couple weeks and you know I looked at it and I was like there’s just no way my mom is listening to it that many times. Kidding. But maybe not. What actually happened was I had to rebuild the feed it wasn’t going to be triggered by bots. So we got a true look of how many people were in fact listening to the podcast. I believe in looking at the real numbers and not sort of surrounding yourself with a false vanity metric because you now realize how easy it is to just fake that. Look, if you want an extra hundred thousand users you know how to it but…you’re not achieving your mission with it.
We’re gonna have resources for the things that we noted in this podcast. Some of the presentations that Dr. Augustin mentioned as well as some helpful tips and episode 40 and a write-up that you can find on our site. As always, thanks for joining us.
this has been using the whole whale, for more resources on today’s show please visit wholewhale.com/podcast and consider following us on twitter @wholewhale and thanks for joining us.
If you’re still listening it probably means you didn’t to press stop on your podcast, so since you’ve gone this far, why not go all the way to itunes and leave us a review, it would really help us reach a larger audience of nonprofits, we currently have about 2000 folks listening to us every month and I love hearing from you. This week’s music was brought to you by Greg Thomas band and the interlude artist there was Andy G. Cohen, with the song, Humming and Strumming. Hope you enjoyed it.
