Ad Fraud Ruins Analytics

Ad fraud. The gift that keeps on giving.

Marketers are beginning to understand that ad fraud affects their actual spend (they get less for their ad dollar), but they haven’t yet caught on to the fact that it affects their metrics as well.  Fake traffic and clicks generated by bots throw off analytics and can cause a marketer to optimize a campaign for the wrong things and waste even bigger dollars.

One way to make ad fraud less lucrative for the fraudsters is for marketers to stop paying for “each,” or volume, and start paying only for quality. That would mean coming to terms with the fact that low quality traffic is even worse than no traffic at all. At the recent IAB Annual Leadership Summit, P&G Chief Brand Officer Mark Pritchard announced that his company, one of the largest advertisers in the world, will only pay for traffic if they can see where it is coming from and whether humans actually viewed ads. Since P&G spent $7.2 billion last year, his announcement caused more than just a slight ripple in the industry.

A casualty of the big wave that rolled over the media supply chain happened to be Facebook, which has always provided its own metrics, and has announced this week that it will allow its ad platform to be verified by the Media Rating Council’s standards.

This will be fun.

And you can’t blame this all on the rise of programmatic. It’s not how the ads are sold or bought, but the kinds of sites selling inventory on the exchanges. Most of the bots happen to be in programmatic ad exchanges, because, as fraud researcher Augustine Fou says, “when those bots cause an ad impression to load on a “long tail” website, they earn ad revenue for that site. Bots don’t make money by causing ad impressions on mainstream publisher sites.”

This is the drum we have been  banging for what seems like forever: buy smaller, higher quality audiences. Again, from Dr. Fou,

one key thing for [marketers] to realize is that there are only a finite number of real human beings. So when they target real human identities, the volumes of ad impressions will probably be far less than they are buying now—but that is a good thing, because they want their ads to be shown to real humans, not bots.

Marketers know who has purchased from them before, who has signed up for email newsletters, or who has visited their website.

By onboarding their own data, they can more easily achieve the identity targeting we just talked about—and thus make their media much more effective because they are actually getting their ads in front of real humans.

We are not sure why this is such a difficult lesson for marketers to learn.