By David Ward
Do brands care enough about fraud?
Ask most brand managers and their agencies and digital publishing partners that question, and the answer will be a qualified yes.
They'll concede that fraud is an ongoing problem, one that isn't going away soon. But they'll also point out that its impact can be minimized by using fraud detection tools, following industry best practices, and adopting open source solutions like IAB Tech Lab's ads.txt to let online buyers check the validity of all sellers of ad inventory.
Others within the industry counter that fraud is the 800-pound gorilla completely distorting huge segments of the digital ad landscape. They say it's getting worse, not better, and many brands are aware of only a fraction of the fraud that is actually taking place in their campaigns.
"They don't know what they don't know — and they don't want to find out more," says New York–based fraud researcher Dr. Augustine Fou. "They don't want to hear about fraud because that will reduce the quantities of cheap impressions they can buy."
The fraud problem only seems to grow as marketers move away from traditional digital and social channels and into emerging platforms.
Grant Simmons, VP of client analytics at Kochava, which provides mobile app attribution and analytics, suggests bad actors have grown increasingly brazen and sophisticated throughout the purchase funnel, especially on harder-to-measure platforms like mobile.
"On a traffic basis in mobile, roughly 70 percent of the ad signal that is reported to us is fraud," he says. "If you look at that on a conversion basis it drops down to around 35 percent fraud, which is still large."
The Varying Estimates of Fraud
Currently, there is some debate over just how much advertising fraud is taking place today and how much more of a problem ad fraud could become in the next few years.
According to Juniper Research and TrafficGuard, by 2023 fraud in the U.S. will account for one in five ad transactions, costing marketers up to $100 million a day.
In its recent report on brand safety, media company GroupM estimates that annual global ad fraud is about $22.4 billion, with most of that taking place outside of North America.
Joe Barone, head of brand safety at GroupM in the Americas, says many of its clients are now well aware of not just the current challenges, but potential new fraud dangers on emerging platforms and are taking steps to address the problem.
"Our more conservative estimate of average IVT (invalid or nonhuman bot traffic) in North America, at 3.3 percent, is because of the progress we've witnessed," he explains. "Those steps include continued adoption of third-party verification services, further supply chain optimization based on quality, pressure on media and tech companies to be certified to industry standards, and advancements with technical solutions to shut down ad fraud services."
There are some grumblings in the industry that some ad networks and content sites are downplaying or ignoring fraud.
Though marketers may not be aware of all the different types of fraud that could reduce the effectiveness of their ad spending, Karim Rayes, chief product officer at digital advertising technology company RhythmOne, stresses that it's a myth to say brands are unconcerned about the issue.
"Ad fraud awareness — and the intent to avoid it — is frequently front-and-center in our interactions with clients," says Rayes, adding most marketers now set up parameters well in advance of a campaign based on how much of a fraud risk they're willing to take. "The lower the risk threshold, the more accurate campaign metrics are — and brands and agencies are willing to take these steps because most often the intention is to weed out as much fraud as possible."
Jenna Umbrianna, chief development officer at programmatic media buying company Digilant, agrees that most brands are aware of the issue, but suggests that different clients may have different ways of detecting and measuring fraud.
That shouldn't impact the agencies working with those brands, which need to have their own rigorous ad-detection tools and standards in place to monitor traffic for fraud in real time. That will ensure brand managers don't have to continually go back and adjust numbers weeks or months after the fact.
"As a general rule, we keep a pulse on the quality of the traffic we're sending to our clients' sites, which helps mitigate concerns around retroactive measurement," Umbrianna says. "We either have logins or get regular reporting around site engagement of the traffic our media channels are contributing and can make assumptions about traffic quality and optimize using that data."
There are some grumblings in the industry that some ad networks and content sites are downplaying or ignoring fraud, because truly accounting for fraudulent traffic means less revenue for them.
Stuart Moncada, VP of product at Ad-Juster, the San Diego–headquartered provider of unified data reporting and analytics for digital advertising, suggests most legitimate publishers view fraud as something that can cost them in the long run if not addressed.
"Fraudulent traffic that is detected is usually revenue that is not paid out to the publisher or is clawed back from them," he explains. "In addition, having premium nonfraudulent traffic is a differentiation that is extremely attractive to advertisers looking to advertise to real users in a brand safe environment."
The New Frontiers in the Fight Against Fraud
The biggest obstacle for brands fighting fraud is the lack of a silver bullet for effectively stopping all types of fraudulent behavior across all platforms.
The detection tools and policies many brands have in place today to battle bots in display advertising or counter domain spoofing (fraudsters masquerading as a popular website but directing payments to their own pockets) are of only limited effectiveness on emerging platforms like in-app mobile, connected TV, podcast, and influencer marketing.
"Bad guys follow the money," explains Fou. "When the money was in display ads, they went there; when it went into video ads, they went there; when it went into mobile, they went there; when the money went into mobile apps, they went there. Now the next thing is OTT/CTV streaming."
These aren't fractions of pennies being, in essence, stolen by bad actors. Kochava's Simmons says payment for a single in-app conversion can be anywhere from 50 cents to $15.
That money can add up quickly, incentivizing unscrupulous networks, mobile websites, and others in the advertising supply chain to falsely claim credit for as many of those conversions as they can. "In some cases, there was never an ad unit shown before the user opted to download the app, but the network still wants credit," Simmons says.
Influencer marketing fraud is now costing brands an estimated $1.3 billion annually.
Kochava works with brands such as Nike, Electronic Arts, and McDonald's looking to get consumers to download branded apps to their smartphones. "These companies want an understanding of how their digital ads are working in terms of driving app downloads," Simmons explains. "CMOs and CFOs realize that if they're ingesting and attributing to fraudulent signals, they could be placing their bets on the wrong horses."
Streaming/OTT services face a similar challenge, exacerbated by the fact that no one is exactly sure just how much fraud there is in the connected TV space, though that hasn't stopped brands from spending billions to engage the cord-cutting viewer.
According to eMarketer, a Q3 2018 study from Pixalate estimated that 19 percent of worldwide OTT impressions were invalid, due to a combination of fraud and ad serving and measurement errors caused by technical malfunctions.
"There are legitimate ads that run on Roku, but bad guys can set up fake Roku apps, fake iOS, and fake Android apps," Fou says.
The lack of standards and tools is also making it harder for brands to stop influencer fraud, which primarily involves influencers using bots and other tools to artificially boost their follower numbers. Influencer marketing fraud is now costing brands an estimated $1.3 billion annually.
Instascreener (formally Points North Group), which provides third-party screening and measurement for influencer marketing, released a report last year noting that even large brands such as Pampers and Ritz-Carlton have been victims of influencers using fake followers.
"Some brands and agencies are well ahead of the curve and know all the different ways that likes, comments, impressions, and follower count metrics can be gamed," says Instascreener co-founder and CEO Sean Spielberg. "Others are surprised by both the level and variety of strategies for inflating numbers."
Spielberg suggests that part of the problem could be brands relying on influencer specialty agencies to both execute the campaign and measure the results.
"The better agencies use authenticated or third-party numbers to measure their performance, rather than 'grading their own homework' with self-reported estimates," he says. "Less rigorous agencies — especially those that use raw follower counts or engagement rates to measure campaigns — are hesitant to do so, as removing fake impressions, engagements, and followers lowers the overall numbers they can report to their clients."
Combining Human Oversight with Technology to Combat Fraud
Once a marketer realizes that fraud is a major problem, the solution is not simply hiring multiple fraud detection services and outsourcing the problem to them.
Instead it requires augmenting those detection tools with the development of a marketing department mindset so that everyone, from the assistant account executive on up, is on the lookout for anything suspicious, even if that dampens raw performance metrics.
"I think that anybody at the brand or agency has the potential to spot fraud, but using software is indispensable," Spielberg says. "Influencer fraud is like an arms race between advertisers and bot farms. Once marketers start checking for particular anomalies, bot farms wisen up and cover their tracks."
For that reason, Fou stresses that effectively combating fraud requires ongoing human vigilance, since bad actors will eventually figure out ways to bypass some fraud detection software.
"They can A/B test their bots so they know the ones that can get through," he points out. "Brands should understand that there are technical limitations to what they'll be able to detect."
Barone of GroupM says his company recommends that clients manually check domains and exclude any that are suspicious.
"They should look for dramatic spikes in web traffic, URL masking, traffic referrals, and domains with low performance to determine if sites need to be excluded due to quality," he adds. "How the client staffs for this will vary; some will rely on their agencies to perform these functions based on their expertise."
Barone also recommends that brands create an internal fraudulent traffic process to monitor and investigate suspected fraud, contractually require payment to be blocked or refunds to be required when fraudulent activity is proven, require use of open-source solutions such as IAB Tech Lab's ads.txt, and only work with certified partners who agree to work within TAG Certified Against Fraud Guidelines.
Umbrianna of Digilant says brands also need to keep their expectations in check. "Brands should be wary of introducing conditions that could encourage game-play, and in turn, fraudulent activity — things like head-to-head tests, unrealistic KPI benchmarks, and the like," she says.
Simmons of Kochava agrees that brands need to temper their performance expectations as they expand campaigns, pointing out that the basic rules of advertising haven't changed just because these are new platforms and potentially new types of consumer engagements.
"If it looks too good to be true, it's probably too good to be true," he stresses. "Acquisition is hard, and brands need to be wary of someone who says it's easily expandable. If you're an advertiser and you're leaning on your platforms to provide $1 CPM conversions, and you keep pushing harder and harder on that, you're inviting fraud."