News & Views
17/02/17 / news

Stack Q&A…Ad Fraud

Leading expert and commentator on ad fraud, Shailin Dhar from The Dhar Method talks to Ryan Kangisser, Founder of Stack I/O about the recent ANA report, the immediate challenges facing the industry, and most importantly what brands can do now to tackle ad fraud.

RK: The recent ANA report on ad fraud wShailin Dharas especially revealing in terms of the magnitude of the issue ($7.2bn) but at the same time we see conflicting data. Do you think the scale of the issue is being sufficiently reported?

SD: I don’t think it’s being sufficiently reported because of two basic points. One is that fraudulent traffic detection is always an estimate and not disclosed fully: they take one sample of traffic, use one detection method, and then extrapolate their detected rate to the entire market. This is why estimates range from $7.2bn to $18.5bn from different bot detection companies for the same year. 

“Robots will never become your loyal customers, no matter how many ads you show them.”

Secondly, reports perpetuate the misunderstanding that fraud is an external force in advertising. It is actually an issue with the supply chain, beginning with websites and ad-networks that source cheap, overwhelmingly robotic and fraudulent traffic.

I believe exposure to fraud can be significantly reduced by simply cleaning up an advertiser’s supply chain (measurement technology should be a second layer of defence) but reports on ad-fraud generally suggest third-party verification as the first line of defence against fraudulent traffic.

RK: What do you see as the most common cause of ad fraud? 

SD: Websites, both big and small, sourcing cheap traffic to meet page views/impressions targets and ultimately their revenue goals. The root intention is never to buy fraudulent, it is to buy cheap; but sub-penny clicks are more than 99% likely to be robotically generated since humans are not waiting around on the internet to be told where to go. If we can raise the premium for validated, quality ad impressions, we can reduce the pressure many publishers feel to buy traffic.

RK: Who between the buy-side and sell-side do you think is working harder to tackle ad fraud?

SD: I think the buy-side has definitely been working harder. Ultimately, they are the ones responsible for spending the advertisers’ money. Everyone in the ad tech chain generally makes money on impression transactions whether they are valid or not. If we stop the financial benefit of these “extra” impressions to agencies, DSP’s, exchanges, SSP’s and ad-networks, we will see significantly more action being taken by all parties because they will incur technical costs for hosting auctions that they will never make a commission on.

“The sustainable solutions lie in a combination of auditing advertisers’ supply chains and using verification technology as a second layer of defence.”

RK: Do you feel the industry trade bodies are doing enough to both educate and mitigate ad fraud?

SD: I truly believe that trade bodies are making an honest effort. The issue is that there is more misunderstanding and misinformation about what fraud is, where it comes from, and how it happens than there is factual material on the subject. My research collaborators and I have made significant efforts in furthering a more transparent and evidence based understanding of the topic and had some rewarding successes along the way. But as an industry, I think we are still quite far from being well informed on this topic.

RK: Can technology alone solve ad fraud? If not, what other processes can be put in place to complete the ‘loop’ on a comprehensive ad fraud prevention programme?

SD: I absolutely do not believe technology alone can solve ad-fraud as what we are battling on the other side is also technology. It is an arms race in a sense and subject to constant change. Established anti-virus software providers have been around for many years now but we are yet to see the emergence of a ’perfect’ anti-virus software, mainly because computer viruses constantly find ways to get around the blocks.

RK: Many established ad verification technologies offer robust fraud detection software. Do you believe this ‘full service’ approach to fraud detection is fit-for-purpose, or a more specialist approach (i.e. specialist fraud detection provider) will deliver greater efficacy?

SD: I don’t think it’s fair to expect any company to limit their offering to just fraud detection. Companies expand or restrict their offerings based on what their customers want and the advertising industry definitely needs fraud detection, viewability and brand safety services because they work towards a higher advertiser ROI. I do believe that it’s unwise to focus on viewability or brand-safety first without ensuring that the traffic or impressions are valid to begin with.

“What’s the point spending money on impressions that fit your brand-safety requirements and are 100% viewable, but were seen by a bot that will never buy your product or service?”

RK: Most buying technologies claim their pre-bid capability provides the necessary detection. Do you agree with this claim?

SD: I don’t agree with this. I am currently working on initiatives to help make this claim more believable. At this moment in time almost all detection technologies allow their clients, the buying technologies, to scan only a sample of the impressions in order to save on costs.

Nobody is forced to disclose whether they are sampling and if so, what percentage of their campaign they actually scan.

I have evidence of ad-tech vendors scanning between 2%-50% of the impressions they buy and sell. I think this is a disservice to advertisers and results in inaccurate representation of the levels of fraud.

RK: Is it as simple as going back to a world of direct-to-publisher CPM buying? 

SD: I think direct-to-publisher CPM buying is great if the goal of a campaign is to buy the attention of a category of people. This does not however eliminate the risk of fraudulent impressions because publishers, both big and small, are the ones buying traffic in the first place. But we have come so far with targeting capabilities that allow us to target specific users based on interests rather than a general demographic; it would mean we’ve given up on targeting technology and programmatic execution.

RK: For those brands only starting to get to grips with ad fraud, where should they start? 

SD: Education. I cannot stress enough how important the education aspect is.

If a brand’s marketing teams are empowered with the knowledge about ad-fraud it is much easier to hold their agencies and other ad-technology partners accountable when the issue of fraud arises.

Teams armed with proper education and knowledge of ad-fraud will be much more successful in delivering a return on investment of the media spend of their brands. Again, robots will never become your loyal customers, no matter how many ads you show them.

Stack I/O is a specialist consultancy enabling brands to accelerate the performance of their data-driven, technology enabled marketing activities. For support with tackling ad fraud, please contact