• by Daniel Avital , Columnist, June 8, 2018

Imagine if your credit-card provider called to say your account was used for a $10,000 shopping spree, and the money was irretrievable. You’d probably go ballistic. Our expectation today is for our provider to prevent these things from ever happening, not report them after the damage has been done. 

Why should ad verification be any different?

“Hey, Pepsi, you just spent 20% of your budget on fraudulent traffic. Here’s a report, glad to be of service.” That’s not a parody, that’s the reality of ad verification in an industry tasked with protecting brands from fake traffic and unsafe content. Ad verification is focused primarily on providing advertisers with measurement and reporting of damage already done. If we wouldn't accept this from our credit card provider, why should we accept this from a verification vendor, much less pay for that service?

Advertisers: stop measuring the problem and start solving it

Marketers and agencies are drowning in brand safety and fraud reports. Teams of analysts are tasked with analyzing these reports, generating insight and optimizing future campaigns using the measured data. At best, it provides advertisers with incremental improvement, but ultimately, they’re wasting time and energy on monitoring damage rather than preventing it.

Wouldn’t that time be better spent producing stellar creative, a killer media plan, and developing other areas of business growth?

The answer: Fully automated, pre-emptive technology

Ad verification should move on from the “measurement paradigm” and begin developing fully automated and proactive solutions, which actually prevent the damage. This, however, requires changes in the technological approach and vendors’ mindsets.

For example, pre-bid is going to have to get real-time.

Promising to filter out unsafe inventory at the pre-bid stage may sound suitably pre-emptive — but the basis of that decision is usually itself unsafe. Most platforms use simplistic “negative-keyword lists” to assess inventory context using a routine scrape of the web to catalogue safe pages. The choicest inventory lies with publishers and news organizations, where up-to-the-minute articles make vendors’ catalogues look stale.

The industry should apply advanced natural language processing tools to get a far more accurate understanding of content and context. 

The same goes for fraud detection and prevention. Vendors rely primarily on pre-purchased lists of fake IPs and look at surface-level data. What they should do is adopt the best practices of cybersecurity and look at heaps of additional data, examining behavioral anomalies and data discrepancies that are typical of non-human users.

Finally, we need to optimize all this advanced content and user analysis at super-fast speed to enable real-time decision making, which is the key to autonomous prevention. As long as platforms can’t process complex data at high-speeds, we will continue to see after-the-fact reporting lead our industry.

If we want advertisers to continue spending rather than diverting money to offline or native, and if we want to spend more time advertising than analyzing, then our approach must evolve from measuring the problem, to proactively preventing it.

That solution will not come from peering at the scene of the crime by gazing at the numbers. It will come from better tech.

Brands and agencies can emulate other industries, such as cybersecurity, e-commerce, and automotive, all of which apply the most advanced AI in the world to combat fraud. 

In the next couple years, reading the tea leaves of brand safety reports will seem like a ridiculous moment in time, a rest stop on the journey toward true digital brand safety.