Ad fraud is any deliberate activity that prevents the proper delivery of ads to the right people at the right time, in the right place. This is an all-too-common issue for modern markets. Fraudulent clicks can heavily skew results and potentially hurt your brand's reputation. Fortunately, we have solutions to reduce your risk of ad fraud.
These are basically scripts that are generated from a particular server, such as Amazon Web Services. They are called “simple” because they are fairly simple to identify: they have a static user agent, IP address, cookie ID, and etc. All these features make it fairly easy to fingerprint and block them.
Sophisticated bots are much more difficult to track down because they use more sophisticated methods like random proxies in order to alternate IP addresses, alternating user agents, to mimic normal CTRs, and sometimes even mimic actual mouse movements that have been captured from browser activity. These tactics make it quite difficult to fingerprint so they can be blocked.
Botnets are comprised of a large group of personal computers, usually residential, that have somehow been comprised by unsavory individuals. These people have gotten control of these computers and use them to load and click on ads. This activity generates what looks like legitimate impressions and clicks on ads, but in reality they are all fake.
Human traffic is often more deceptive than nonhuman traffic since the end users are actually real people even though the impressions and sometimes clicks being generated are fraudulent. It may be a lot more difficult to detect this kind of ad fraud if you are only on the lookout for bots.
Publishers who are in the business of defrauding advertisers can “conceal” ads so they meet the criterion for “no chance of ever being viewed” by any humans. They can do this by “ad stacking” or stacking the ads on top of one another. This is also called “impression stacking.” When an ad is hidden from view behind another ad the publisher can generate any number of impressions for one page view, even when the ad on top is the only one being seen.
This is one of the most despicable types of traffic fraud perpetrated by humans, yet it is rarely reported. It can come in many different shapes and formats, like display ads and often video. Essentially, a fraudulent publisher will buy web traffic at rock bottom prices and offer it for resale at exorbitant prices. For example, a publisher may buy questionable web traffic to his/her site for pennies and then sell it to advertisers for an average of $5 CPM.
In an environment of real-time bidding (RTB), publishers are on occasion permitted to put up their own domain as well as the label of their website ID. This gives the fraudulent publishers a chance to lie about or “spoof” their inventory. For example, they may say they are a well-known publisher like huffingtonpost.com, but in reality the ad was served on a completely different domain. At other times, the ad-serving domain is being spoofed in the actual bid request. Either way, it’s fraud.
In the RTB environment, inventory is organized according to site ID with each site ID correlating to one single domain. However, often a lot of ad exchanges and publishers bundle whole networks of domains together under a number of single site IDs. Advertisers think they are purchasing ads on abc.com, but in reality their ads end up appearing on def.com.
Due to adware plugins and browser toolbars, ads can actually be injected or replaced on any website without the publisher or user even knowing that it’s happened. This can cause ad inventory to appear on ad exchanges as if it’s Facebook.com, even when it is not at all from Facebook’s true ad inventory.
Cookies act as a mechanism that allows the ad tech environment to target specific audiences. This is why they’re so important. Cookie stuffing has been going on forever and there have been some well-known instances where the practice was used to generate huge affiliate revenues. With cheap sources for buying Internet traffic, cookie stuffing in an effort to misrepresent or even dilute audience data happens quite often and something real to be aware of.