How AOL Works to Prevent Ad Fraud - eMarketer

Newsletters Sign-Up

Plans & Pricing

Does My Company Subscribe?

How AOL Works to Prevent Ad Fraud

May 1, 2015

Sean Crawford
Senior Vice President, Publisher Client Services and Operations for ONE by AOL: Video (formerly Marketplace)

Sean Crawford oversees all publisher partnerships for AOL’s programmatic video platform ONE by AOL: Video Marketplace (formerly Marketplace) and has a direct hand in crafting AOL’s anti-fraud practices and policies. Crawford spoke with eMarketer’s Lauren Fisher about how AOL works with buyers to minimize ad fraud—a growing problem across desktop, mobile and video.

eMarketer: How much of an influence is fraud having on the digital advertising space today?

Sean Crawford: Very. You can try and solve for viewability, for URLs and all that other stuff, but if an impression isn’t real and if the traffic to a website isn’t real or if it’s bot traffic, then everything else doesn’t matter.

The problem is there isn’t really a comprehensive definition of what fraud is for everyone. At the basic level, you talk about impression fraud or bot traffic or suspicious traffic, but there’s no agreement on what needs to be rooted out. For us, that’s the starting point of the fraud conversation and for everything else.

“We tell buyers that we want our threshold [for ad fraud] to be zero, but in reality, that’s very difficult and might not be achievable.”

eMarketer: What is AOL doing to address fraud today?

Crawford: For bot and suspicious traffic, we use a mixture of three things to root out the problems: in-house technology, third-party technology and humans.

For the human portion, no site or publisher is allowed to work with [now called ONE by AOL: Video Marketplace] or any other AOL properties without actually talking to a person. You just can’t show up and sell. It has to be a vetted site. We have a whole subset of rules for how they get vetted and what we’re looking for before they’re allowed to partner with us.

Next, we use internal and external technology to monitor all the traffic for bot traffic, and we have thresholds that we manage to. We’re constantly trying to push network traffic down below a certain percentage of suspicious traffic, constantly trying to push that closer to zero.

eMarketer: What kind of expectations are advertisers coming to you with in regard to traffic quality concerns such as fraud and viewability?

Crawford: The expectations differ by buyer, but everyone wants to know what we are doing for suspicious traffic. They also want to know about viewability. Everyone is looking for more viewability, but everyone has a different percentage. Some demand 100%, and some are OK with the industry standard definition. Others want to make their own. Everyone has a different take on it.

“The expectations differ by buyer, but everyone wants to know what we are doing for suspicious traffic.”

In programmatic video specifically, we also talk about buyers’ ability to use their own measurement tools. If they have a favorite vendor, they can bring that vendor to the platform to measure and get their own reports back so they understand what they’re seeing. We have our data, they have theirs, and we can have a conversation. The number of buyers using their own verification tools has picked up a lot in the last year.

eMarketer: Right now, with viewability, we have industry standards for what is considered an acceptable threshold. But when it comes to suspicious traffic and ad fraud, what’s the expectation? Is it zero?

Crawford: The expectation is zero until you walk them through the last time their credit card number got stolen. Companies like American Express, Visa and MasterCard have thresholds they manage to. Grocery stores manage to a certain level of things getting stolen. Of course they prefer it to be zero, but they know people are going to come in and steal. So it becomes about trying to manage to an acceptable threshold.

It’s the same with ad fraud. We tell buyers that we want our threshold to be zero, but in reality, that’s very difficult and might not be achievable. Of course we’re always trying to push that threshold toward zero, and we work with them to pick a model they want to apply to that particular campaign or that contract to manage to that threshold accordingly. For example, we might overdeliver on a particular campaign to get that number down to zero.

eMarketer: What about fraud in mobile?

Crawford: For us, we look a lot at mobile video, which is important to note because there is a difference between video and display. There’s also a difference in mobile app and mobile web. On the app side of it, for video, focusing on things like the app store ID, app store URL, correct app name and all that type of stuff can solve a huge part of any problem that might arise. And I’d say that’s true across display and video.

With mobile web, you have the same problems you have on the desktop. So whether I look at a webpage from my PC or my iPhone, it’s the same webpage. The issue in mobile is making sure we have good detection and making sure we can understand the problem and solve it quickly. There is some technical challenge there, but if a website is on my blocklist, it should be on my blocklist for mobile web as well because it’s the same site. If the webpage is bad, the webpage is bad. So even just the basic stuff like making sure websites on blocked lists for desktop are also blocked on mobile devices can solve a ton of problems for the buyer.

Go beyond the articles

Corporate subscribers get quick and easy access to the exact data and analysis they need to make critical business decisions:


  • Go beyond the articles:

    eMarketer Products

    You've never experienced research like this.

  • Hear from our clients:

    Customer Stories

    Nearly all Fortune 500 companies rely on us.

  • Want to learn more?

    Contact Us

    Inquire about corporate subscriptions today.