This is the first in a series of interviews with vendors combating the problem of ad fraud. Other companies participating in this series include White Ops, DoubleVerify, Forensiq, Moat, PubChecker, Telemetry and Integral Ad Science.
Videology might not be specialists in ad fraud like White Ops or Integral Ad Science, but preventing and detecting it remains a top concern. At its core, Videology’s ad tech is designed to serve ads on behalf of its advertiser clients in display, online video and TV.
The company also has a product called Viewpoint, which tackles issues around viewability, brand safety and fraud. The company applies Viewpoint as a free add-on to clients running Videology’s media. It can also apply the technology as an additional security measure for clients doing prenegotiated direct buys or private marketplace deals.
But because fraud prevention/detection is not – in Videology product management director Quinn Sanders’ words – the company's “bread and butter,” it's in the process of evaluating third-party anti-fraud vendors to partner with.
“We’re looking to bring one vendor into the technology stack where we can offer a third-party pre-bid fraud- and bot-prevention solution at no cost to our demand-side clients,” Sanders said. He spoke carefully and declined to name exactly which vendors Videology was vetting.
“We want a third-party partner with a very good name in the space, where we would ask them whether every ad request we get has the characteristics of a bot or not,” he said. “And then we would decide whether or not to serve the ad based on the results. The additional peace of mind [for clients] is it’s not just Videology saying this is a safe impression and not a suspicious impression.”
Sanders spoke with AdExchanger about why fraud prevention is so much more difficult than detection, and the additional complexities video inventory brings to the table.
AdExchanger: Why do you want to partner with a fraud-prevention vendor now?
QUINN SANDERS: While our clients find a lot of value in running campaigns on the media we procure, a lot would prefer to have an independent third party. And there are a lot of high-quality vendors in that space.
We also offer solutions for brand safety and fraud and bot prevention, but we do not consider those our bread and butter. White Ops, Double Verify and Integral Ad Science do it all day every day.
Tell me about Videology’s in-house fraud prevention and detection solution.
All the optimization we do – whether we’re meeting campaign objectives like serving all the committed impressions, meeting the impression amount, hitting completed view requests – is based on our clustering solution. All of the optimization we do, fraud and bot prevention included, is based on that clustering solution, where we build clusters of users.
Are prevention and reporting solutions often segmented?
They’re definitely segmented and you can achieve those goals in different ways. Prevention is not necessarily achieved in the same implementation as detection. Some clients are comfortable with just detection and reporting. And others are very keen on a real-time blocking and prevention solution.
Why are they separate?
Detection is much easier than prevention. It’s much easier to report on fraud after the fact than to detect it at the time of the ad request, before the impression serves.
You can accomplish detection with a much lighter integration. Either via a wrapper or a tag when the campaign is being trafficked or via an SDK the publisher implements. Whereas prevention and blocking requires a much deeper integration either via an API or some server-based technology.
As you evaluate fraud prevention vendors, have you noticed differences in technology and approach?
There are absolute differences. Some of the vendors in this space, and without getting into too many specifics, have history in the display world and are trying to take their proficiency in display media verification and apply it to the video space.
That can have some pitfalls. Video is different than display. That adjustment can take some time.
What are the technical challenges around video fraud?
There are a couple of different aspects. One is that when a display ad is presented, that’s it. The full impact of that display ad can be taken in very briefly, in a second or two.
Video content is at least 15 seconds and can be up to 60 to 90 seconds. All the things we look at from performance to brand safety and fraud prevention are very different.
In the display world, a lot of what you’re looking for is direct response and some call to action. Clicks are very critical in that context.
In video, clicks are almost an indicator of fraud. When you look at a video ad, it’s rare that a user will click on that ad. Video is typically about brand awareness and raising the profile of that brand. It’s not about direct response or an immediate call to action. When you’re looking at tracking the effectiveness and the potential for fraud, all of those things come into play as well.
What is Videology looking for in a partner?
When we’re partnering, we’re looking for guys who do video well. We do display, we realize it’s not going away. But many clients want their entire spend running through us, so we can do the TV spend, the display spend and the video spend through our platforms. But our key is digital video.
We’re also a global company. We have very strong partnerships in Asia, Europe and EMEA, so that’s another aspect we look for when we vet third-party vendors in this space. Can they provide a solution effective on a global scale? Some of the vendors are very US- and North America-centric.
What are the differences in how vendors apply fraud-prevention technology?
Some solutions are cookie-based where they’ll try to drop a cookie on what they believe is a bot. Anybody serving the ad can look for the presence of that cookie and deny the ad request.
There are other solutions that do not rely on cookies, that are a deeper integration and look for characteristics of that request that would denote or show a tendency of being a bot.
What are the pros and cons of each?
To me, the deepest integration is the best. Anything where we can do real time is the optimal scenario. There are benefits to a cookie-based solution, but there are limitations as well. The bots turn over between five to 10 days onto a new IP, onto a new machine and some are getting very good at clearing their cookies and hiding their trail.
What are the detriments of having a non-cookie-based fraud solution?
When you look at a deep indicators of fraud on a real-time basis, the difficulty there is it’s a big commitment to have that integration with the vendor. That takes a lot of work on both the vendor and the ad server.
But in general, we want to be able to partner with any of those vendors.
What’s the benefit of cookie-based prevention for advertisers?
That’s tough for me to answer. I’m not in an agency and I’m not an advertiser. I’d be speaking on their behalf. We try to present all the options to our clients and identify the pros and cons and let them make the decision that’s best for them.