The last few years have been encouraging from an ad fraud perspective. Not only are we seeing that the overall projected loss to ad fraud may be starting to decrease (according to numbers published by us and the Association of National Advertisers), but partners up and down the advertising ecosystem are clearly taking the fraud problem seriously and implementing measures to combat it.
The challenge, though, is that this metrics-driven business needs to learn not to rely on metrics to measure success in this particular instance. The irony isn’t lost on us, but using numerical goals for acceptable levels of Sophisticated Invalid Traffic (SIVT) won’t result in a truly improved and cleaned environment. Any number you could choose (other than zero, of course) for an “acceptable” level of bot fraud, by definition, still permits bot fraud to occur. And while getting from 11% to 3% may feel like a big win, particularly in a short period of time, that 3% still represents $x spent and lost.
There are numerous reasons the SIVT percentage can fluctuate over the course of a campaign:
- Did your breakdown of mobile vs. desktop placements change?
- How about the size/style of the ads themselves?
- From a programmatic perspective, did your win rate decrease or increase during this time?
There are many causes that can change the SIVT percentage naturally without any change in campaign strategy. And living or dying based on a single SIVT metric doesn’t account for those adjustments in campaign tactics.
Our recommendation is to take stock of that SIVT percentage, but couch it in the context of the campaigns it’s measuring. Look more closely at SIVT events than at the overall percentage.
Tie the number back to overall ad spend and break it out by display vs. video, mobile vs. desktop, linear video vs. connected TV, etc.
By breaking it down like this, suppliers and buyers can understand where the majority of SIVT events are coming from instead of just focusing on the SIVT percentage. This level of specificity is important in figuring out which partners are delivering more SIVT events than others. The entire premise of exploring invalid traffic is to reduce it, and that’s not just through blocking it, it’s through choosing partners and placements that don’t result in it in the first place. Having the information about where SIVT comes from allows for more educated decisions about safe and effective media buying.
Consider: a SSP could show only 1% SIVT over the course of one month. One percent sounds like a tolerable rate, but that could translate to one billion SIVT events. One billion events isn’t tolerable, and would be worth examining further to uncover what tactics the SSP is using that creates such a bot-ridden environment.
Continuing to be vigilant across all partners, even those who have low SIVT rates, will continue the fight towards reducing fraud. All partners should be empowered to reduce ad fraud as much as possible, instead of trying to live below an arbitrary 3% benchmark.