I’ve written extensively on digital ad fraud and how it negatively impacts marketers’ campaigns. Even having fraud detection in place is not enough because those detection technologies miss more fraud than they catch — due to errors in measurement or simply not looking for various forms of fraud that are not bots hitting webpages. But since they are “black box” you will never know if they are measuring accurately or not — i.e. they don’t explain how they measured it. More importantly, paying for fraud detection is kind of like paying for cough suppressant — you’re suppressing the symptoms of the disease, without curing the disease itself. Unless you “cure” the disease of ad fraud, you will continue paying for fraud while also paying for fraud detection tech that doesn’t detect it.
If you don’t know what the disease is, you can’t cure it. Having the analytics in place so you can see what bots are coming to your site or eating up your impressions enables you to know what disease you have and then take appropriate action. What follows are tips on how to improve your digital marketing campaigns across the popular tactics currently being used by marketers big and small. The insights come from 25 years of doing digital marketing, both as a small business owner, where I ran all the campaigns myself, and as a consultant serving some of the largest advertisers with digital strategy and tactical optimizations of their campaigns.
Display and Video Ads
The vast majority of display and video ads are purchased programmatically these days. That means algorithms automate the buying and placing of the ads across hundreds of thousands of sites and apps, most of which you’ve never heard of. That’s where the ad fraud comes in. Those fraudulent sites and apps can disguise their traffic and users and make them look realistic, in order to avoid getting caught. So fraud detection fails to detect them and reports to you that everything is fine, giving you a false sense of security. But despite spending more money on digital campaigns, and even getting more clicks and traffic to your sites, sales haven’t moved much. That’s because bots can create many ad impressions and even click on them, but bots don’t buy stuff, especially offline in stores.
The disease you have is ad fraud, caused by skilled bots. The best way to avoid this disease is to avoid catching it in the first place from the millions of long tail sites that your ads would have been shown on when placed through programmatic channels. These are sites and apps that you’ve never heard of. They don’t have many humans visiting them, because those humans have never heard of them either. So to increase their own revenue these sites simply buy traffic, generated by bots — software programs designed to repeatedly load webpages or continuously use mobile apps — in order to create billions of fake ad impressions out of thin air.
On the other hand, there are mainstream sites that you’ve heard of and other humans have as well. The most savvy marketers now use very strict allow lists or include lists, to ensure their ads go on these mainstream sites, instead of millions of unknown ones. By way of example, in 2018, when Chase reduced the number of sites showing their ads from 400,000 to 5,000 (a 99% decrease) they saw no decrease in business outcomes. That was their first step in reducing waste and exposure to ad fraud — by not having their ads sprayed out on all those long-tail sites. You should try this too.
Paid Search Ads
If your business uses paid search ads, that’s a good thing. Search ads are shown when users are searching for something, and they are even based on the exact keyword the users are searching on. And you don’t have to pay for the impressions; you only pay when you get the clicks. But beware of bots that click. Over the years, I’ve seen craziness in paid search campaigns like 50% click through rates, or even 100% click rates. Those are the bots that got greedy — they have to click, in order to get paid the CPC revenue share.
The best way to avoid the disease of click fraud is to avoid catching it in the first place. How? Search ads on the main platforms can go on the main properties like Google.com or Bing.com. But the ads can also be shown on outside websites that are part of “search partner networks” of these platforms. These sites earn a revenue share on every click generated on their sites; so there is an overwhelming temptation to increase their own revenue by using bots to click on the search ads.
Savvy marketers can avoid this kind of ad fraud, by turning OFF “search partner network” in the campaign settings and allowing your ads to be shown on the main properties like Google.com. Humans “google” things so you are far more likely to have your ad seen by and clicked on by a human when limiting them to the main properties. If you absolutely must use search partners, try to use an include list where possible, or aggressively exclude sites that are exhibiting strange data like abnormally high click rates. This is how you limit your own exposure to ad fraud, and improve your digital marketing campaigns.
Facebook, LinkedIn and Other Social Ads
A similar principle holds for ads on “walled gardens” like Facebook and LinkedIn. The ads that show on Facebook itself will be subject to far less ad fraud than ads showing on sites and mobile apps that belong to Facebook Audience Network (FAN). Again it boils down to simple incentives. Those sites and apps get a share of the ad revenue; they can increase that revenue by using bots. Those bots also click on the ads to simulate “engagement.” After all, it’d be too obviously fraud, if there were absolutely zero clicks.
The best way to avoid the disease is to turn off FAN and limit your ads to only the main properties and apps — Facebook, Instagram, and Whatsapp. If you have already been using FAN, be prepared for a sharp drop in the number of impressions and clicks, and the click through rate. Facebook might even try to scare you with that sharp drop in “engagement” or “performance” and higher effective CPMs. But bot clicks are not real engagement, and lower CPMs from fake, long tail sites is not better “cost performance.” Your campaigns will need far fewer ad impressions, and fewer clicks, to perform better. That’s because your ads are actually being shown to humans. Don’t you want your digital marketing to actually “perform” for real?
Similarly, be sure to turn off GDN (Google Display Network) if possible. And be on the lookout for abnormal numbers in campaigns run on other networks like Twitter, Snapchat and TikTok. When campaigns get 50 billion hashtag mentions, you should probably be suspicious of that. After all there are only 7 billion humans on earth, and not all of them use TikTok, right?
YouTube and CTV/OTT Ads
The common sense principles stated above also apply to YouTube ads and connected TV, streaming ads. If you’ve never heard of a YouTube channel or creator, or a channel is 1 month old but already has a billion views, etc. All those should be tell-tale signs that there’s some form of cheating going on. Do you want to show your ads on those channels? See: NYTimes – The Flourishing Business of Fake YouTube Views.
Similarly in streaming video, there are certain streaming apps that humans know and use. This article by Dave Morgan, CEO of Simulmedia, notes that “streamed viewing [is] at almost 24% of all video viewed on TV. The vast, vast majority of streamed viewing is on services with no ads or few ads — Netflix, Amazon Prime, Hulu, Disney+ and HBO Max. Of the dominant five, only Hulu carries ads, and only on part of its service, with a very light ad load relative to conventional TV.” This means, when you see tens of millions of streaming ad impressions from Roku apps that no one has ever heard of, it’s probably not humans streaming. Buy streaming ads from Hulu and Roku directly and keep a close eye on which apps are burning through your impressions and budgets abnormally. And aggressively block and exclude those. See: Forbes – CTV/OTT Streaming Ad Fraud.
The Rest – Influencer, Affiliate, Lead Generation
I won’t go into much detail here since it’s already been well covered by others. Influencer campaigns are subject to influencer fraud — fake accounts that inflate the follower counts, making it appear that a particular user has more “influence” than they actually do. See: NYTimes – The Follower Factory. Affiliate and lead fraud are also straightforward for bots to accomplish. They can trick the attribution platform to “give credit where credit was not due” and they can complete lead forms with accurate data (personal info stolen from data breaches). See: CNBC – Bots Kept Winning T-Mobile’s Promotional Contests.
Yes, some practitioners will criticize the recommendations above as not sophisticated enough. Of course they are right. There are nuances and many other details beyond what was presented above. The above is basic blocking and tackling in digital marketing. Many marketers are not yet doing these; so everyone should start at the basics, and then progress in sophistication as these foundational steps are accomplished.
And those marketers who want to buy the largest quantities possible, at the lowest CPM prices possible, so they can spend all their budget before year end, can just ignore all of the above. You won’t be able to do all those things without fraud. Oh, you don’t mind that there’s fraud or you have fraud detection reports that show less than 1% fraud to show your CFO/CEO? Ok. Carry on.