Brand safety remains a frustrating game of Whack-A-Mole

“It is probably a pain point for a lot of buyers out there,” said Ari Levenfeld, director of inventory quality at DSP-provider Rocket Fuel.

Although brand ads appearing next to extreme content has always been a quirk of the internet, the current political environment has placed pressure on brands that have had ads appear next to racist content on websites like Breitbart and YouTube. Programmatic companies have responded by blocking fringe websites from their platforms. These blacklists bar inventory that is directly sold through exchanges, which decreases the probability of ads appearing on those sites. But because of rampant reselling and mislabeling, the blocks don’t eliminate websites from slipping through.

Reporting problems
A lot of these snafus stem from poor reporting. Shady websites and the ad networks that they work with will use masked and false URLs to get through blacklists. Masked domains are where a website is listed as “unknown” in a supply-side platform or exchange, while false URLs are where a website is presented as being something it is not. Imagine ihatesports.com being presented as espn.com to buyers.

“Blacklist reporting sucks because website-level reporting on the open exchange sucks,” said an ad tech analyst on the condition of anonymity.

Some exchanges will make concessions with shady publishers because every website they turn away is viewed as lost revenue, the analyst said. If a publisher realizes it has been blacklisted by major buyers, it will request that its domain be masked, and the exchange will comply to keep its business. Advertisers can then buy masked domains, which are usually discounted to encourage buyers to take risks, to reach users they track across the web. With false URLs, things get more complicated as websites engage in reselling and arbitrage aimed at burying shady websites in the transaction.

“Major buying platforms only look as far as their contracts require them to go,” said Adyapper CEO Elliot Hirsch. “For each domain listed by a DSP, you need to think of it almost as a 2007 mortgage crisis ‘mortgage bundle,’ where it may be rated on the surface as a C-plus, but inside it’s composed of many more hidden websites.”

Some bad actors will use URLs for their iframes that are totally unrelated to their actual domain. For example, filesharing.com might list iframes under ilovedogs.com. Like a Russian matryoshka doll, the website will then embedded iframe within iframe to avoid detection.

Banned websites have incentive for trickery, but the programmatic shenanigans are also driven by ad networks who are trying to reach traffic goals by acting on their own behalf.

“There are a lot of second- and third-tier networks that are just scraping by that utilize some shady practices to make money,” said John Murphy, vp of marketplace quality at ad exchange OpenX. “It is on the exchanges to make sure that isn’t happening. … But limiting reselling is a time-consuming and technical process. A lot of exchanges might ask [networks if they are authorized to sell], and if they get a yes, that might be good enough for them.”

Verification limits
Verification services can be useful to buyers trying to prevent their ads from showing up on questionable websites. But they have limits that aren’t always understood by advertisers. The amount of data from each bid that can be scraped from the open exchange is limited. It isn’t until a buyer makes a purchase that they get a full view into the publisher offering the inventory.

Prior to purchase, the only information DSPs and their vendors obtain is a bid request that describes the inventory and the website’s users, and the thoroughness of these reports varies per exchange. Because of this, verification services don’t have access to enough pre-purchase information to make a proper evaluation of whether or not the seller is misrepresenting themselves, said independent ad tech consultant Brad Holcenberg.

It isn’t until the buyer makes the purchase that its vendor’s javascript can run on-page to confirm where the ad showed up. Buyers can use that information to inform upcoming buys or request rebates, but sometimes the damage is already done if a brand’s ad already riled up activists by appearing next to hate speech.

How do we fix this?
As for solutions to these issues, there is no quick fix. Whitelists — which restricts impressions so that ads only appear on websites on the list — can be a useful tool to limit brand-unsafe buys, but they can also fail if a network gets a SSP or exchange to mislabel, say, watchpornvids.com as nytimes.com. Concerned buyers should double-check what they get from the open market and ad tech execs will self-servingly suggest vendors that might be able to help there. Ben Alpren, who heads vendor partnerships at programmatic agency The Exchange Lab, said that buyers should connect directly to publishers if they’re worried about they’re purchasing from exchanges.

So there are several things advertisers can do to reduce their vulnerability to bad websites, which may be enough for some brands since campaigns aren’t judged on where a small amount of impressions land. But totally rooting out bad websites from exchanges, and not just reducing their access in a politically tumultuous era, would require diligence and cooperation from networks, exchanges and programmatic platforms. And these players are rewarded on pushing volume in an environment where the amount of premium inventory is scarcer than its advertised to be.

Rocket Fuel’s Levenfeld believes that most of the large exchanges and buying platforms are doing their best to prevent blocked websites from slipping in.

“But there are those that just don’t care,” he acknowledged. “Some people aren’t paying attention, and it makes it easy to take advantage of them.”