AdExchanger.com: How big a problem is impression fraud today?
Impression fraud is a huge issue in online advertising and it happens across thousands of sites a day. There are currently 4 million sites in the U.S. carrying ads on a daily basis and studies show that on average, 31 percent of all online impressions are non-compliant. A significant cause of this waste is due to common fraudulent activities and we have noticed quite a large amount of questionable activity on some sites already. There are potentially hundreds of ways in which a site can engage in advertising fraud. For example, publishers can stack 10 ads on top of each other, where the user only sees the topmost ad but all 10 advertisers pay for an impression. There are numerous other ways to do this as well (Professor Ben Edelman publishes other types of advertising fraud on his blog at http://www.benedelman.org.
Unfortunately, there is no one way of indentifying advertising fraud and that's what makes it complicated. Our researchers look into suspicious activity, analyze it and then come up with a "signature" to identify different types of fraud so they can be tracked. This is very similar to how viruses are found. We then push out signatures to our crawling infrastructure that crawls millions of sites daily to look for fraud, which allows us to identify many other sites that engage in non-compliant practices using the same techniques. Every one of these fraud types that are found are usually in low numbers, but combining them can result in some big numbers. We don't know what they are yet, but we suspect that there are hundreds of different types of fraud that are pervasive across a significant portion of the long tail Internet. Our goal with the Advertising Fraud Detection Lab is to bring light to the dark corners of online advertising so the entire industry can move forward and away from practices that hurt advertisers and the industry at large.
Is DoubleVerify getting into the impression fraud and click fraud business similar to Click Forensics, Anchor Intelligence and others?
Companies like Click Forensics and others focus more on the validity of the user that is viewing or clicking the ad and their technology is based more on mining data about the user than verifying ads or detecting the types of fraud we're looking for. We are addressing the problem of fraudulent sites and looking for patterns of fraudulent activities in the html source of the page as well as looking for adware/malware activity from browser toolbars and other downloadble applications.
This is based on our Virtual Visitor technology that crawls millions of sites a day, renders all the third party code on the page and scans to look for signatures across huge volumes of pages. If anyone is delivering an ad inside an invisible ad space, our crawling technology would recognize that and flag the site as fraudulent.
DoubleVerify will continue to be a leading media verification provider and we will constantly iterate on our technology to cover new forms of non-compliance. The research done by the Advertising Fraud Detection Lab will influence our product offerings in the future, but our core values and mission will stay the same.
How will your fraud research directly impact DoubleVerify's client campaigns? For example, will it report out to a client that a publisher on their campaign is using "Stacked Ads" ("publisher stacks ads on top of each other to give the appearance of multiple ads served on a page, even if only one ad is visible")?
Whenever our researchers find a new type of fraud we will incorporate the ability to identify a web page that uses this type of fraud in our crawling technology. Since we crawl millions of sites a day, this will scale it across the Internet and we are likely to identify all the sites and pages employing this type of fraud within a few days. When our crawlers find a site with this type of fraud, they will flag that site as fraudulent in our site database. Then depending on what type of service our clients are using - if they are using our blocking product (BrandShield), we will block the ad from being delivered to the fraudulent site. If they are using our reporting product (BrandAssure), we will report on the fraudulent site.
What's the revenue model for the research arm?
Currently there is no change in revenue structure as the research arm itself is not a revenue generator. The results of the research will be incorporated into our product suite to improve what we can offer our clients. This is another pioneering effort and unique advantage that only DoubleVerify can offer to its customers.
By John Ebbert