“Data Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Will Luttrell, Co-founder and CTO at Integral Ad Science.
Fraud impacts the entire online advertising industry and costs us millions of dollars each year. While everyone can play a role in minimizing the problem, the buy-side in particular can help lead us to the solution. It seems counter-intuitive; our natural tendency is to assume the sellers of fraudulent inventory are to blame, and therefore must be responsible for a solution. After numerous discussions with the sellers caught up in this activity, I’m more convinced than ever that buying behavior must be addressed. Buyers must reevaluate how they gauge performance.
Before I go any further, allow me to clarify: When I refer to fraud, I’m referring to ads that are served, but cannot be seen. This is different than ads that aren’t viewed due to poor placement (that’s another topic for another column). I'm specifically talking about the practice of deliberately and intentionally rendering an ad so that it will never have the chance to be seen by users, frequently inside a 1x1 pixel or hidden behind other ads. There can be literally hundreds of advertisements in the background of a browser or a page and never on the webpage itself – and yet someone is claiming credit for those clicks, impressions and subsequent acquisitions.
The conversation closed with a discussion of action items and how the sell-side could address this problem. It was suggested that we create a collaborative blacklist that the sell side could share. It was even recommended that the larger tech companies with more scale and resources create and maintain the blacklist, and all the other players subscribe. Not only is this unrealistic, but I contend that published blacklists would actually make the problem worse. The bad actors would continue to engage in fraudulent behavior - until the instant they showed up on the blacklist, at which point they would abandon ship and register new domains. A blacklist gives them a window to optimize their behavior. It would also let the grey hat guys, who essentially provide channels for the other bad actors to play in, to subscribe to the blacklist and be "compliant" - and continue on with business as usual.
The most challenging conundrum regarding fraudulent activity is that the campaigns that are affected by it appear to perform very well. How can that be? These ads are being gamed in two ways. They're being clicked programmatically, so if a marketer is measuring the effectiveness of a campaign by clicks, these ads will appear to perform well. Frequently, they're placed in such a way that they're going to get a lot of last-touch credit, so if the campaign is being measured by CPA, they're going to perform very well because, although they're not driving sales, they're going to get credit for those sales by being the last ad served before a purchase. It's a bulk-buying, cookie-bombing technique.
Speaking with representatives on the sell side, I’ve tried to understand their perspective. Their responses have ranged from, “Our buyers don't care, so we don't care," to shock and dismay. In other words, those being rewarded for clicks and CPAs don’t automatically care because they are meeting the KPIs they are tasked with. Others have endeavored to figure out how they can address the issue, only to discover that solving the problem hurts their bottom line so much, that they can't do anything about it. Their clicks, conversions, impressions and/or acquisition credit rates drop below those of competitors who choose to turn a blind eye. Rather than risk losing valuable customers based on “performance,” they sit idle.
We've actually seen leading household brand campaigns that are 80% fraudulent. That number probably began lower, with a media buy across several channels and perhaps a 10% fraud rate on a $10 million budget. If the marketers saw that one channel in particular was receiving a lot of clicks or last-touch credit, they may have opted to double down on that channel. And if it continued to work just as well after they doubled down, they may have doubled down again and again, until their entire $10 million budget is going toward a channel that is 80% fraud.
Let’s examine who’s winning and losing in this scenario. The biggest winner is the fraudulent publisher, of course. This can be a household name publisher who is sourcing their traffic on the grey market – or it can be a fraudulent company who has no real traffic at all. The media planner also wins, having shown the ability to generate better performance in their key compensation metrics. In fact, almost everyone in the value chain takes some margin on the increased “supply” of inventory.
Meanwhile, on the losing side of the equation, the CMO is looking at flat sales numbers. They can really only draw one reasonable conclusion: display ads don’t work, regardless of how many fancy KPIs they are shown. The media channel that did the best job scrubbing out fraudulent inventory just lost a major client. Finally, the publishers who offer good exposures and real opportunities to impact revenue are marginalized, and watch as their CPMs are dragged down by the halo effect of this activity. They are also seeing revenue stolen as fraudulent activity is given click credit or attribution credit on sales that they helped create.
So the sell-side is being held hostage when it comes to fraud. They're being held accountable for the negative activity, but at the same time, they have no choice but to let it continue. If they don’t, the victims of this activity, the advertisers, will abandon them for another channel that "performs" better, with a better click-through rate or better CPA attribution. Although advertisers are the ones being hurt, potentially losing millions of advertising dollars on ads that can never be seen, they are simultaneously creating and perpetuating the problem. It is a natural result of hinging all their buying decisions on a few metrics that are being gamed.
The solution here isn’t simple, but it has to start with a change in how we evaluate the success of campaigns. That’s key, and it’s key for everyone in the industry to understand and own. Simple metrics are easy to comprehend, but they are also easy to game. We need to move beyond KPIs (like clicks and soft conversions) that are simply easy to measure and massage, and into KPIs that matter and truly impact advertiser businesses.
So, while it’s an industry wide challenge to drive the changes necessary to address this issue, the ball is now in the buyer’s court – as suppliers will adjust to their demands. Buyers; start telling your suppliers a different story: insist on clean, safe, quality advertising environments that get noticed by real consumers. But go one step further: measure and reward the sellers that actually provide those environments. If the buy side takes the performance emphasis off last click/last touch KPIs and places it back onto quality placements, causal attribution, and more modern means of evaluation which strip out the influence of non-viewed/fraudulent ad purchases, all the brain power locked up in the ad tech space will give them exactly what they’re asking for.
Just like it does today.