Remove fraudulent ad inventory with Admetrics Fraud Prevention
Admetrics Fraud Prevention can counter the problem by providing the necessary analytical insights to optimize display and video inventory towards non-fraudulent traffic.
What types of fraud techniques does Admetrics Fraud Prevention address?
Botnets (or other automated agents) posing as humans that view or click on display or video ads.
Stuffing ads or content into a tiny pixel, so that they count as served, but never have a chance to be actually viewed by a human consumer.
Also called ghost sites or pseudo sites, Ad Farms are fraudulent web properties that only exist to host a massive amount of ads without providing content targeted to a human audience.
An in-browser attack in which ads are forced onto a website, displacing the initial web page content or overlaying on top of existing content or ads.
Multiple ads are placed on top of each other, with only the top ad actually in view.
Algorithmic fraud risk assessment
ARPA is algorithmic fraud risk assessment via machine learning based on a wide range of probabilistic and deterministic signals.
The ARPA metric can then be used to optimize spend, for example to limit the exposure of a campaign to placements with an assigned risk under, say 65%.
Admetrics analyzes and qualifies referrer data of domains and iFrames to assign the fraud risk on domain level.
Admetrics identifies fraudulent deliveries through techniques like ad stacking, iFrame stuffing, pixel stuffing and ad farming.
Admetrics leverages viewability tracking as an indicator for fraudulent traffic, for example identifying clicks from impressions that are not viewable or finding other common patterns of fraudulent impressions.
Admetrics maintains its own blacklists of compromised IPs or devices. Additionally, we use industry standard, real-time databases of compromised clients, IPs and notorious proxies.
Admetrics analyzes and qualifies incoming traffic for signals that indicate fraudulent intent, including spoofed user agents and browsers, abnormal behaviour or traffic patterns.
Admetrics differentiates between ad fraud botnets and benign automatic traffic (regular spiders and crawlers that are not employed to manipulate ad media).
Putting the data to use
The basic set comprises ARPA metrics like Share of fraudulent automated traffic, Share of non-fraudulent automated traffic, Share of high risk traffic and Spend at risk (total spend considering high risk and medium risk traffic).
Adding Fraud Prevention to your stack
Possible applications are feed integrations, API integrations or real-time interfaces for use in the programmatic area, like pre-bid flagging towards non-fraudulent impressions.