Online ad fraud is on the rise, and costing companies billions of dollars each year. Display advertising and programmatic ad buys get most of the press, but clever fraudsters are finding new ways to fake video views and mobile ad impressions, as well. Smart marketers need to stay vigilant for signs of ad fraud and take steps to protect themselves.
Amit Joshi is director of product and data science at Forensiq, a leading advertising fraud detection company. I invited Amit to Marketing Smarts to discuss the ad fraud ecosystem, explain the different types of ad fraud, and share strategies for detecting and preventing ad fraud from siphoning off your marketing dollars.
Here are just a few highlights from my conversation with Amit:
Ad fraud isn't just about bots; there are real people actively defrauding marketers (01:24): "Ad fraud is really varied in its definition.... Basically, you have your whole conversion path (if you're thinking programmatic) all the way down to a conversion or an acquisition lead purchase up to a bid request in the programmatic or [Real Time Bidding] environment. Fraud can occur at any point in that funnel. But what we do at Forensiq is chop that funnel in half and have ad tech or programmatic fraud and performance or conversion-level fraud.
"I'll start with the bottom of the funnel.... Think if you're an insurance company and you're trying to purchase leads for people to sign up for you insurance program, you might use an affiliate network to help drive those leads for your program. And what we saw was that a lot of fake leads were being submitted to these companies.... Fake in two senses: one is the more obvious case where all of the information is fake. Fake person, fake email, things like that.
"Then the trickier case where what we were seeing was that these fraudsters were grabbing real people's information and then submitting [it] to multiple insurance programs.... There's two sides to the fraud equation where there's all this botnet-driven, programmatic, automated type of fraud, and there's all this human driven, real people interacting, but interacting in a way that there's no [genuine]intent behind these interactions."
Ad fraud can occur at any point, but the jackpot for fraudsters tends to be at the bottom of the funnel (03:30): "At the bottom of the funnel, payouts tend to be higher. We were working with some customers who could be paying between $10 and $200 for a lead, so human-generated fraud was a lot more profitable. If you submit five of those fake leads in a day, you might get a thousand dollars worth of payout, so there wasn't as much need of automation, versus at the top of the funnel where you're trying to...steal impression dollars where someone might be paying one dollar for a thousand views of an ad and you have to hit that ad way more times...to generate enough fraud to be profitable, which requires some sort of automation or nonhuman component."
Switching from display advertising to video won't insulate you from ad fraud (07:11): "In this programmatic ecosystem, whether you're thinking about display or video views, one of the key ways that advertisers have historically protected themselves against fraud is using viewable percentages—the percentage of people that actually viewed my ad. They're using some sort of third party to measure whether the ad was in view on a page for a certain amount of time.
"But these bots are becoming increasingly clever. What they do now is take an advertiser KPI. So they figure out how the advertiser is measuring the effectiveness of a campaign and then specifically figure out how to fake those signals or measures. They've actually figured out a way to fake views. A bot could go to a page and make it seem to a third-party vendor that they've viewed an ad when it's really some sort of nonhuman customer browsing in an invisible browser."
To protect yourself from ad fraud, choose publishers wisely (14:58): "There are a few trends that marketers can look for. What I would do as a marketer is always make sure that you're vetting the partners that your working with, making sure they're high-quality. For example, if you're running on The New York Times, your fear of ad fraud is obviously a lot less.
"But even with a reputable publisher, some fraud can occur because let's say New York Times is loading an ad from someone that's committing fraud, that fraudster might inject some malicious code via their ad unit even if they don't own that domain, but working with trusted partners is a really good first step to protect yourself."
Amit and I talked about much more including how ad fraud impacts mobile, which metrics to monitor for telltales signs of fraud, and what to do if your ad impressions are marked as fraud, so be sure to listen to the entire show, which you can do above, or download the mp3 and listen at your convenience. Of course, you can also subscribe to the Marketing Smarts podcast in iTunes or via RSS and never miss an episode!
Music credit: Noam Weinstein.