Advertising Fraud is a term that is used to describe a wide variety of destructive behaviors in digital advertising, but also in traditional advertising as well. Because digital advertising is acutely measureable, there has commonly been intent to fake the data through non-human ad views, fraudulent clicks to ads, cookie stuffing and non-viewable ads. Unfortunately, it happens in all forms of advertising, including magazines that may embellish subscriber numbers in order to increase ad rates or television and radio advertisers that assume an audience is engaged when they may not be paying attention. And of course there’s John Wanamaker’s now famous quote about advertising, “Half the money I spend on advertising is wasted; the trouble is, I don’t know which half.”
Fortunately, Silverlight Digital structures its defenses to address and combat fraud by utilizing technologies that enable us to stay ahead of the bad guys. Because it comes in so many forms and is constantly evolving, we combat it through a broad and evolving range of both internal and partner technologies, as well as through business protocols and processes, which include the following:
- Strict Inventory pre-vetting – We are very selective about the publishers we work with and require they meet our stringent standards. This acts as a gatekeeper to prevent fraud-vulnerable publishers and media into our purview. This filtering process includes a double pre-screening of inventory lists by partners and again by our internal team, plus blanket exclusion of undesirable publishers, sites and categories. This filters out of any non-directly contracted inventory and prohibits activities such as daisy chaining and other unacceptable forms of impression delivery. That screening process filters suppliers that contain high levels of undesirable or non-human traffic and who often show up in other platforms.
- Automated fraud detection – We utilize cutting edge fraud detection software that uses analytic techniques to scan for unusual, suspicious, or outlier activity, in terms of browsing behavior and click behavior along vectors including cookie, site, IP address, and more. These measures have proved extremely effective at early identification of fraud.
- Proactive filtering – Once identified, we flag fraudulent traffic so that we never bid on it in the first place. This leads to improved performance both from the avoidance of spending budget against non-human traffic and from the redeployment of that budget against high-quality traffic.
- Ongoing monitoring – We have automated internal reporting that monitors impression, click and conversion data across our system to spot new anomalies our existing measures might miss and flag these for review to spot issues and isolate causes.
- Open data sharing with partners – We feed regular reporting of flagged or suspicious activity to our exchange partners and work collaboratively with them to locate and prevent the issue at the source.
Incident resolution process – We have developed real-time internal processes to identify the source of the new issues and quarantine it from live bidding for immediate investigation and resolution.
Marketing technology providers face these challenges every day.
Some cases of fraud are black and white, but others are not. We understand that in addition to the internal measures above, it is important to provide our advertisers with additional layers of filtering and control. We provide clients with a variety of these controls, including:
- Fraud filters from leading third-party vendors, including Integral Ad Science’s pre-bid targeting solution, Bid Expert (which in addition to fraud offers brand safety, viewability, an overall media quality TRAQ Score targeting), quality-targeting settings powered by Peer39, and integrated tag support for a wide range of other leading vendors.
- Powerful inventory management tools, including global, campaign and strategy-level site list management, as well as transparent URL targeting and site transparency reporting.
- Unique and proprietary UI targeting features, such as Minimize Ad Collisions, which effectively selects against low-quality traffic with a higher chance of fraud, as well as new fraud filters are in development.
- Built-in goal-based optimization allowing users to direct the system to algorithmically optimize to results that cannot be generated by bots (i.e., purchases, ROI).
“We know these measures are working, because we see it in the data,” said Brown. “When fraudulent publishers are named in the press, we find little to no delivery against these, having already screened them out.” At Silverlight Digital, our partners tell us we are consistently among the first to report fraud back to them. Our advertisers who measure not just conversion quantity, but conversion quality, tell us that ours is consistently highest. Our click-through rates are in line with realistic levels corresponding to human activity.