TubeMogul To Issue Automatic Refunds For ‘Nonhuman’ Video Traffic

Brett-Wilson-TubeMogul-2015-2Video platform TubeMogul sees itself as a bastion for brand advertisers and, beginning in April, it’s putting its money where its mouth is.

The company will roll out a Non-Human Traffic Credit Program, an automatic make good on all video impressions it buys on the open exchange that have been identified as fake or fraudulent.

In order to do so, it will license bot-detection technology from White Ops and layer its FraudSensor system into Tube's OpenRTB bid streams.

“We’ve already integrated several third-party measurement providers into our software, so we’ve tried to significantly reduce different types of ad fraud for some time,” said TubeMogul’s CEO Brett Wilson. “This initiative is about giving clients total certainty that, when they buy through us, botnet fraud is no longer a problem for them as a matter of policy.”

TubeMogul is no newbie to policing bad traffic. In 2011, it outed competitors (and sometimes itself) for running “fake pre-roll” or so-called video ads stuffed in banner display. Then, in 2013, it formed an open-source video viewability consortium dubbed OpenVV with a select handful of other providers, which the IAB has since incubated.

Despite these efforts, video traffic quality and viewability issues abound.

A recent study by the ANA and White Ops found video CPMs in excess of $15 faced 173% more botnets than lower-CPM spots. Other findings indicate half or more of all video ads bought programmatically today are still served in-banner or in inferior player sizes.

Wilson spoke with AdExchanger about TubeMogul’s new traffic quality drive and its latest platform developments.

AdExchanger: Why are you doing this now?

BRETT WILSON: It was the right thing to do for advertisers. When we’re running inventory, we will credit clients’ invoices automatically for any fraudulent botnet activity for any given account at the end of the month. They don’t have to do anything. We’re picking up the tab to pay for third-party measurement on 100% of impressions and we’re making sure clients never pay for fraudulent botnet impressions again.

You’ve partnered with Integral Ad Science for viewability measurement. Did White Ops offer more in terms of nonhuman traffic detection?

There are multiple, really solid verification companies in the space. Our engineering team had a preference for White Ops on this initiative because they specialize specifically in botnet fraud. We remain partners of Integral, and will continue to be.

What material impact will this have on you?

As a company, we're happy to take on the cost for anything that makes our product better. You see other big announcements from Johnny-come-latelies who act like this is a new issue. This will be operationally burdensome. It’s difficult to do. If you take the number of platform clients we have, you have that more many transactions to deal with every month so we wanted to make sure our team and the White Ops team was ready and could walk clients through all the implications.

How will you handle mobile, if measurement there isn’t fully developed?

It certainly is the fastest-growing media type through our offering. We’re actually testing mobile in-app video reporting right now with White Ops and expect to roll it out soon. So we fully intend for this to be a desktop and mobile solution in short order.

Is there this Catch-22 where hitting high viewability might really mean you’re hitting a lot of bots?

It’s not just about one rate or viewability percentage. The worst sites in terms of botnet activity sometimes have the highest viewability rates. That doesn’t mean there’s not plenty of great sites out there with actual humans who are watching the video and those are the sites and apps and networks we try to highlight through our platform.

By being transparent, we think we’ve been very aligned with publishers who are more premium. Because even if you spend a little more in raw CPM, your vCPM goes down. That’s consistent with our brand – being this software specifically built for brand advertisers. Not for trading desks, not for ad networks. We’re not looking to help you buy low and sell high.

Have supply partners been receptive?

For the most part, they’re very much in line with what we’re trying to accomplish. This is actually a very good thing for [suppliers’] business because you end up directing more spend to premium publishers and the people working hard to create great content and that’s the inventory and deals they’re trying to aggregate. Some of them have a lot of transition to do from midtail inventory to more premium inventory and these types of initiatives will help.

Has a reduction in impressions due to ongoing supply-side cleanups (e.g., LiveRail) affected you?

There’s always scarcity. That’s a fact. But these recent announcements aren’t really changing anything for us. It will make it harder for other platforms that have been more direct response-centric. If you’re designed to optimize to a click or a CPA, in some ways blocking ad fraud or bad sites hasn’t been your main priority. This move we're making will really push the rest of the industry to follow suit and I think there are a lot of platforms that are ill-prepared for that.

How much inventory does this represent, given the program primarily affects OpenRTB buys?

Advertisers certainly use our platform to buy inventory in the ad exchanges and through SSPs, but an increasing percentage of the ads run through our software are private deals advertisers are cutting directly with publishers.

What percentage of your overall business comes from direct orders?

Platform revenue from private deals has been about 15%-20% of overall platform revenue, and our best clients look at us as software. Some use us for all RTB and some use us for private inventory and some clients use us for a mix of both. It’s important to us that we’re working with brands directly. Quite often that’s in conjunction with an agency partner. That’s exactly where the industry is heading. The age of the nontransparent, performance-pricing trade desk model is waning.

How is your value proposition different from other DSPs?

Our reason for being isn’t solely to help agencies make money and some of the platforms that only target trading desks – when you look at their features – that’s what their strengths are. They’re building features to help agencies create internal ad networks and we want to help agencies be awesome engines to their clients.

How is decentralization at the agency level impacting your agency relations, given you’re regularly working with brand clients’ agencies?

We’re more aligned with agencies than ever as they move toward transparent service models built on top of technology. We called it a private trading desk model years ago. It’s a tough transition for agencies to make because you have these revenue streams that come from their trading desks. And they know that the market demands a transparent product and that they need to grow to evolve and accommodate that without shrinking their businesses.

What’s the ratio of your managed to self-serve or platform business now?

Seventy to 80% of our spend in a given quarter is from self-serve platform direct clients. We’re happy to offer a managed service if that’s what advertisers want, but that’s not the type of company we want to build. In some of the larger brand deals we’ve signed in the last few years – L’Oréal, Mondelez, Lenovo and Heineken – sometimes advertisers have complicated challenges and needs and there’s certainly more of a service demand.

This interview has been edited for clarity and length.


  1. Why is this news? Search engines have been crediting bot traffic for a decade...?

  2. Is this a fancy marketing twist on TubeMogul not investing in pre-bid filtering technologies like most of their competitors are doing? Instead passing the buck down the chain to supply partners?


Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>