• by Augustine Fou , May 7, 2015

By now, everyone is aware of the bot problem, also known as NHT (non-human traffic). Most also understand that bots can go to webpages to cause ad impressions to load and then click on the ads too. But probably fewer people are aware of the true extent to which bots are impacting the bottom lines of all the players involved in the digital advertising supply chain.

Outright waste due to fraudulent impressions

Let’s start with the obvious. When bots cause ad impressions to load (display ads, video ads, mobile display, etc), then those impressions are all fraudulent; they are 100% wasted ad spend for the advertiser that will never yield conversions. Furthermore, it’s the bad guys that pocket the money, rather than the premium publishers who would have legitimately earned those ad dollars. According to Ben Trenda, CEO of AreYouAHuman.com "The sophistication of bots on the Internet has advanced to the point where a static list of known bots, while helpful, is not an effective standalone method of solving the myriad problems caused by non-human traffic."

Additional spending on anti-fraud technologies and services

As advertisers and publishers realizethe magnitude of these forms of digital ad fraud, they are allocating more budget to fraud detection and prevention platforms. This kind of spending further cuts into their bottom lines, in addition to the negative impact already seen on top-line revenues. Even the anti-fraud vendors are taking a risk when they make “fraud free” guarantees -- i.e. if there is bot fraud detected, they have to “eat” the difference or pay out “make goods” and “refunds.”

Completely messing up metrics and KPIs

Bot activity also messes up virtually all the metrics currently being used to measure the performance of digital ad campaigns. For example, bots coming to a website artificially inflate traffic numbers. Bot clicks on ads alter click-through rates (CTRs) that are used to measure the engagement of users. In fact, by tuning CTRs with bots, the bad guys can make their sites appear to have higher average click-through rates than legitimate sites. So algorithms that optimize for any one of these single measures could actually be sending more ad dollars to the bad guys.

In addition to bots messing up ad metrics, Roberto Genaro, Digital Marketing Officer at TravelBrands also saw bots impact website metrics -- like traffic, time-on-site, and bounce rates. “Bots were coming in through anonymous proxies and cycling through IP addresses so fast, they were creating bottlenecks and brownouts.” Using technology from Distil Networks, Genaro deflected this bot activity and reduced downtime by 75%, reduced bounce rates, and eliminated duplicative content (stolen via scraping and then used elsewhere). 

Depressing industry-wide CPMs

Because there is virtually zero incremental cost to generating millions more ad impressions, bad guys can sell those impressions for cents per thousand -- and still make nearly 100% pure profit. But this means that when ten cent CPMs are being sold next to ten dollar CPMs from premium publishers, less and less budget will flow to the good guys (decrease in revenues) and they are likely to be forced to lower their CPMs to compete (decrease in profits). And, obviously for advertisers, even if they continuously optimize for higher CTRs, greater viewability, or other measures, they will still see no lift in business impact -- because ALL of the activity and KPIs were created and tuned by bot activity.

Solving the bot problem can have large impact

There is mounting, concrete evidence being shared by various parties in the supply chain that speaks to how dramatic the positive impact on business outcomes could be if the bot problem was even partially solved. A medical device client can cite the reallocation of nearly 20% of their paid search budget away from clearly fraudulent websites (which used bots to send CPC/click traffic) and to legitimate publishers. A pharmaceutical client saved $1 million by removing a single ad network from the overall media buy. 

Data from a webinar by DataXu and DoubleVerify showed that a 20% reduction in fraudulent (ad impression inventory) led to a 75% improvement in cost-per-action (CPA) metrics (i.e. it lowered the cost per action). In fact, their recommendation was for advertisers to evolve beyond click-based metrics (which can easily be manipulated by bots) to action- or conversion-based metrics, which provide a more accurate picture of the business performance of the campaign. And comScore recommends a “holistic” approach that correlates audience, viewability, and NHT measurement. 

So, this is a call-to-arms for all players in the digital ad ecosystem to step up the efforts to solve the bot problem, in order to achieve a clear-and-present impact on their bottom lines.