Recent statistics show that digital advertising reached $ 140 billion in 2020 in the United States. It is well known that there is a large amount of digital ad fraud, although the exact amount is disputed, ranging from 1% to 100%. It is also well known that the advertisers of the biggest brands like to buy large quantities of ads, at low CPM prices, with the appearance of high “performance”. They choose to ignore the fact that large quantities, low prices, and high clicks are only possible from bogus sites and apps with bogus users – bots (automated software) that load ads and click on them. There are not enough humans on earth to generate the 500 trillion advertising opportunities per year; but this is trivial for large botnets, such as those previously used for DDoS (Distributed Denial of Service) attacks.
If a marketer genuinely cares about achieving real business results with digital marketing, like small and medium businesses, how can they buy digital media while minimizing their exposure to ad fraud? Keep in mind that these small businesses cannot afford any fraud detection technology. But as you will see below, these detection technologies are not necessary if you know how to buy digital media while minimizing fraud in the first place.
The following are general recommendations. Of course, there are exceptions and nuances, but these are meant to provide starting points for digital ad buyers who want to do more on their own, in order to minimize fraud and maximize business results.
Start with strict inclusion lists
If you buy ads through programmatic channels, and that includes buying through Google Ads, the best way to avoid 99% of fraud is to start with very strict inclusion lists. What does “strict” mean? It means very very short. Let’s start with a brainstorming exercise: “Name the top ten sites you visit each day, as quickly as possible”. How many did you throw before you started slowing down? For most people it’s between 5 and 8 and then they slow down. Do the same with mobile applications. If you can’t quickly name ten sites and ten apps that you use every day, how many millions of sites and apps are used by real humans? Probably not much of the 1.5 billion registered domains, or the 350 million active websites, or the 10 million mobile apps.
I’m sure you’re worried that you won’t have “scale” if you take this approach. But think of it this way – the grand scale you hear about comes from long tail sites that have few human visitors; these sites generate advertising revenue by buying traffic from bots. If your ads appear on these sites, they are shown to bots. So that whole scale won’t produce meaningful business results. These are just big numbers on a spreadsheet. Do you want to pay for expensive spreadsheets or do you want to do digital marketing?
A note on “direct purchases” is warranted here. Buying direct from a publisher is a good thing; this minimizes the supply path – the number of middlemen who profit from your dollar before it reaches the publisher, to run the ad. However, buying direct does not solve the ad fraud problem if the publisher is a cheater. For example, if the publisher is the one buying traffic from the bot, or if they do other shady things on the site to increase revenue, like stacking ads (stacking 50 ads on top of each other in the same ad space), pixel stuffing (filling advertisements or entire web pages in hidden 1 x 1 pixel windows), popunders, refreshing ad spaces, etc. Even if you are buying direct, you should have analytical tools to monitor your listings and look for anomalies.
For example, in the graphic above, there are 2 examples of hourly models. Each green bar represents volume per hour. Note that in the first case, virtually all of the volume is blown out at midnight; just at the end of the day, the volume is at maximum, leaving little or no at all for the rest of the day, when humans are awake and in line. The second graph shows the volume a little more spread out, but most of it is still used before noon, there is nothing left for the second half of the day. As you might have guessed, these are bot traffic patterns – where bots deliver traffic and clicks in the fastest possible time. If you don’t have hourly data you won’t see this and think you have a lot of traffic. See: Soufflé between midnight and 4a
How to buy at “Walled Gardens”
You may have heard the term “walled gardens” used to refer to Google and Facebook. This term comes from the fact that they define their own rules and the fact that most of the users are logged into Facebook or one or more Google services at some point – e.g. Gmail, Android, YouTube, etc. if you buy media and market on these platforms, the same principles mentioned above apply.
If you run search ads on Google, make sure they stay on google.com. This minimizes ad fraud because scam bots have no financial motive to cause large amounts of ads to load on google.com i.e. they cannot make any money from it . Fraudulent bots go to sites that pay them for the traffic. Sites in Google’s search partner network have the motivation (make more money) and the means (those scam bots) to increase their own ad revenue by paying bots to generate and click on ad impressions. So if you limit your search ads to the primary property, you are already reducing exposure to spoof bots (you do this by disabling “search partners“).
The same goes for Facebook. You can avoid most simple advertising scams by disabling FAN (“Facebook Audience Network”). FAN includes all sites and apps outside of Facebook that monetize using its advertising platform. They also have the motivation and the means to increase their own ad revenue by using bots. So, by disabling FAN, you avoid most of this type of fraud. The ads on Facebook itself, including Instagram, have proven to be very effective. Just be prepared for much smaller amounts and fewer clicks than if you left FAN on and let the deluge of false impressions and clicks run into your campaigns.
How to buy YouTube and CTV (Smart TV)
Finally, if you buy video ads from YouTube and CTV, avoid the spray and pray. Ads served on YouTube are at high risk of appearing alongside unsavory or downright inappropriate content. It is UGC (“user generated content”) after all. Are you okay with showing your ads before, during, and after the yoga, beheading, bomb-making, terrorism, and child abuse videos on YouTube? Of course not. So, if you buy ads from YouTube, be sure to select channels and creators who have a demonstrable history of producing quality and useful content. There are many gems on YouTube; you can find the channels and creators that are right for your brand and your audience. It’s exactly like the inclusion list approach above; include specific YouTube channels in your purchases to avoid pretty much all the bad stuff.
The same goes for CTV. Manually organize CTV apps that have humans and have appropriate audiences for your brand. If you haven’t heard of a CTV app, chances are other humans haven’t heard of it either. It doesn’t matter that there are tens of thousands of apps in the Roku app store. How many of them are used by humans? Select them and include them and you will avoid the vast majority of frauds that already plague CTV media.
After reading the above, I hope you will notice the recurring theme. Buy digital media from a carefully curated list of great publishers, real mobile apps, real YouTube creator channels, real CTV apps, and more. There are a finite number of humans who use a finite number of sites, apps, CTV channels, etc. for finite periods of time and generate a finite number of ad impressions. The vast scale that programmatic digital media claims to have is mostly bogus sites and bogus apps that have bogus users generating bogus ad impressions. “Missing” this “ladder” is a good thing; it will definitely improve your digital marketing. If your ads are shown to humans, you will get better business results. Let other gullible traders buy this scale. You know how to avoid fraud and do better digital marketing.