Many brand advertisers know there is fraud in the digital advertising ecosystem — fake traffic, click farms, sub-par leads, etc. Yet far too many dismiss it as “acceptable” or “accounted for” by the far lower CPMs they are getting on ad exchanges for the immense quantities of ad impressions.
While the premise may be that more ad impressions usually leads to more awareness and therefore a lift in sales, the underlying assumption is that those ad impressions were actually seen by humans. In the modern digital ad ecosystem, advertisers can’t afford this assumption.
Ecosystems of Fraud: Fake Impressions, Fake Clicks, Fake Leads
For those advertisers who still pay on a CPM basis (every thousand times their display ads or video ads were shown), there are botnets honed to deliver pageviews and therefore ad impressions to their exact specifications. Bad guys first create networks of sites, where content is automatically assembled or generated by algorithm. They stick many display ads on each page. Then, they use the botnets to generate large quantities of traffic to these sites and push such “ad inventory” into ad exchanges, for sale to brand advertisers. When advertisers’ bidding systems see desirable target traffic, they bid to place the ad. A similar sequence of events happens for CPM based video ads.
For CPC based advertisers who pay only for clicks, botnets are used as “click farms” to generate clicks on keyword content-based ads, so the owners of these networks of fake sites can harvest the CPC based revenue.
Finally, in the lead generation space, where advertisers pay for “qualified” leads of users who supposedly fill out forms requesting more information about products like air conditioners and services like insurance, bad guys can arbitrage the difference between what they make per lead and what it costs them to pay a worker in India to fill out forms with fake data.
Why Now? Rise of Ad Exchanges and Automated Buying and Placements
At every stage of the purchase funnel, from impressions, to clicks, to leads, there are proven forms of fraud. Advertisers know this is happening. The amount of such fraud is actually increasing, despite industry efforts to detect and mitigate it. This is due to the dramatic rise of the use of ad exchanges, where ads are “decisioned” and placed within milliseconds. Ad exchanges already account for $1 in $5 of display ad budgets. When more and more ad inventory is pushed into these systems and hundreds of billions of ads are placed per month, few advertisers spend enough time to analyze the results and the details of this truly massive quantity of data.
Typically, a simplifying assumption is made — the ad fraud and waste (up to 40%) is “priced in” due to the lower costs I now pay.
Industry Initiatives Are Useful, But Not Enough
However, this is no longer a valid argument, because even lower cost ad impressions or clicks are still worthless to advertisers if no human ever saw it or clicked it. Also, the lead forms filled in by workers paid to do so are not going to turn into paying customers.
The ad fraud problem is so pervasive and deep-rooted, it is hard for most advertisers to even fathom. Even as the drumbeat grows louder and more industry initiatives are launched to combat this rampant fraud, those efforts are largely not sufficient.
For example, Nielsen and the IAB focusing on defining “ad viewability” and “view through” is still missing the point that even if the ad is viewable, if no human saw it, it is valueless to the advertiser. Spider.io, WhiteOps, Integral, and many others are using bot detection algorithms to detect and mitigate suspect traffic. Unfortunately, it is hard to separate malicious bot traffic from legitimate crawler traffic (because those are bots, too) and it still doesn’t tell the advertiser which activity is a real human, who can potentially become a paying customer.
Finally, Google removing known fake websites from search results can temporarily mitigate the amount of free organic traffic these sites receive, but thousands of new sites and millions of pages laden with algo-generated content and ads spring up in their place.
Human Whitelisting versus Blacklisting Techniques
Many of the advanced technology approaches center around blacklisting — i.e. blacklisting sites that have known algo-generated content; detecting and foiling botnets that generate pageviews and clicks; Google maintaining a list of sites that serve malware, etc. The premise is that if we detect it is a bot, then it is not a human.
None of the approaches and technologies focus on positively identifying humans, though; those users who are able to convert into paying customers and should therefore be worth a premium to advertisers.
If advertisers were to shift their emphasis to human whitelisting techniques instead of botnet blacklisting techniques, they would be able to more efficiently identify profitable target customers (i.e. traffic).
Furthermore, advertisers should do some or all of the following:
- Don’t assume that lower cost is sufficient to account for the ad fraud because if you do, you are still wasting your ad budget.
- Shift from CPM based ad impressions to performance type ads like CPC-based search ads where possible.
- Regularly monitor your analytics to check the bounce rate, dwell time, pages per visit, etc. of users coming to your site (bots don’t stick around to read anything).
- Cookie-ing their own visitors, specifically the ones that have actually completed purchases (bots don’t complete purchases).
Diligently performing these steps can help advertisers prevent committing fraud against themselves.