Is Invalid Traffic Really A Problem In OTT?

Over-the-top (OTT) television is at an inflection point.

While it’s a promising way for media owners to capitalize on changing consumer consumption habits – the number of connected TV users totaled 181.8 million viewers in 2016, a 20% year-over-year increase, according to eMarketer – with reward comes risk.

OTT’s lucrative CPMs could also attract bad actors, including nefarious third-party resellers or traffic sources that inflate viewership numbers, that wish to take advantage of a relatively immature supply chain.

Recent research from ad intelligence provider Pixalate seems to support that case. Its inaugural Video Seller Trust Index, aimed at benchmarking video ad quality across screens, concluded that 44% of the OTT traffic it studied was invalid.

Pixalate claimed its measured OTT impressions had high volumes of traffic coming from data centers, which can indicate bot traffic if that system isn’t affiliated with a consumer or corporate network.

“In additional cases we measured, we saw impossibly high bursts of single user traffic in extremely short time periods,” noted Tony Casson, director of product strategy for Pixalate. “Our data, which allows us to identify sellers responsible for the measured impressions, comes from sources throughout the entire advertising stack.”

One emerging consideration is that a video publisher might buy third-party traffic, making it hard for the OTT platform that hosts the app, to detect where the ad or page load request came from. It could have originated from a real user’s device – or a third-party unaffiliated with the content owner, reasons Pixalate.

The latter scenario introduces the ability for non-human traffic to flow through an OTT platform.

Is Invalid Traffic On OTT Happening?

Other experts claim invalid OTT traffic isn’t rampant today, since the channel’s finite inventory tends to sell out. Also, that inventory mostly sells in private marketplaces (PMPs) or through direct deals, where there’s more control and brand safety.

“Right now, it’s a very premium, high-quality environment with high CPMs and exceedingly high sell-through rates,” said Allen Klosowski, VP of mobile and connected devices for video SSP and ad server SpotX. 

That’s one reason why Scott Rosenberg, VP of advertising and audience strategy at Roku, was thrown off by Pixalate’s stats around OTT fraud – specifically one result which found that 59% of Roku traffic was allegedly invalid or fraudulent.

Most Roku deals are private deals, Rosenberg argued, which reduces the chance of third parties unknowingly entering the mix. This factor suggests that the volume of invalid ad impressions cited occurred in open exchanges.

“The vast majority of audience, and therefore, ad inventory, on Roku is locked up against major apps like a CBS, Fox, ESPN or Hulu, and I’m not aware of them selling media in a public exchange setting,” Rosenberg said. “What are you measuring if you’re not partnered with us or our direct publishers? You can’t be a third-party measurement entity in a private transaction unless you have the buyer or the seller putting you into the mix.”

Pixalate claims it detects traffic across open exchanges, as well as programmatic PMPs and direct placements. Pixalate also claims it measures all impressions directly from the client device – be it the app, browser or video player.

OTT Is A Very Different Animal

Although high percentages of data center traffic might signal fraud in the desktop world, OTT is different.

TV publishers use a plethora of third-party platforms, including content delivery systems and SSPs to support OTT streams, so there are naturally more intermediaries in the mix.

A server might call web content before delivering it to an OTT device to improve video performance or overall playback, noted Tamer Hassan, chief technology officer of fraud detection solution White Ops.

For instance, in server-side stitching, a publisher might use a solution from FreeWheel, mDialog (now owned by Google) or Brightcove, to facilitate a smooth and bufferless video stream.

In the process, a bunch of ads get “stitched” into content.

Combining ad insertion with video quality can make it tough to trace a specific URL’s location and the original traffic source.

“You have to know enough about the ecosystem to whitelist those [server side] IP addresses you know to be valid,” Rosenberg said.

SpotX’s Klosowski agreed.

SpotX works with many devices and platforms, such as game consoles or Blu-ray players, which initiate streams from a data center because of the heavy bandwidth required for video.

“Just because traffic was data center driven, it doesn’t mean it’s fraud,” Klosowski said. A lot of players initiate a video stream, but then call up a data center to do ad insertion, trigger creative from a DSP or manage the campaign delivery.

What Can Be Done To Prevent OTT Fraud?

Today, connected TV measurement is difficult because there aren’t quality or viewability standards.

“The MRC doesn’t really have a position yet on OTT and all of the early movement around fraud and viewability has to do with the browser environment,” Rosenberg said.

Many buyers measured interactive video using the VPAID standard, which isn’t compatible with most connected TVs. (Some video/OTT players like Brightcove and JW Player support VPAID 2.0).

“We don’t support javascript in the Roku platform, so you wouldn’t query the player to find out what’s above below the fold or behind the window the way you would in desktop,” Rosenberg said. “None of that methodology applies in OTT.”

And many early OTT platforms and game consoles weren’t built as open platforms; AppleTV, for instance, just released its tvOS last year.

So there is opportunity for fraudsters to escape client-side, device-level detection common in standard VPAID environments.

But industry bodies such as the Interactive Advertising Bureau want flexible standards, like VAST 4.0, to provide more metadata between the video ad player and server, and help publishers standardize ad delivery, measurement and creative across platforms.

Even with emerging standards, content producers and OTT platforms must maintain quality as their PMPs scale. Roku says it has tested Moat and Integral Ad Science in campaigns. Roku’s ad framework also integrates with Nielsen and comScore.

The industry should monitor the emerging programmatic ecosystem in OTT, experts say. Roku’s Rosenberg predicts that over time, more OTT audiences could be bought in an exchange capacity.

If more OTT inventory is sold programmatically, more ad networks could enter the picture – lengthening the distance between the advertiser and end publisher or platform.

“Who can help ensure a high signal-to-noise ratio is the question,” Rosenberg said. “Conceptually, as more inventory flows to programmatic, certainly the exchange is obligated to play an enforcement role to ensure the ecosystem remains pristine.”

This post was syndicated from Ad Exchanger.