Human CPM (hCPM): Optimize for Humans to Avoid Ad Fraud
As an advertiser, what would you rather pay for – actual human eyes seeing your content, or software-generated traffic? If the question seems absurd, think again.
False non-human traffic is a serious issue in the advertising industry and one we have previously discussed, for example here. Various studies reveal that fewer than half of all impressions are actually seen by humans and $6.3 billion of ad spend a year is wasted on fake traffic, or clicks that appear to be views but are actually the work of software.
Ready to dive deeper? Check out our selection of takeaways from a subject matter expert and learn why you should make sure only humans view your content:
- “Just add humans”. According to Augustine Fou from FouAnalytics, for the last decade, marketers have purchased large numbers of ad impressions and gotten large numbers of clicks, but no one thought to ask if those impressions were shown to humans and whether those clicks were clicks by humans who actually saw the ads. Imagine this scenario: Even when marketers paid for fraud detection and viewability measurement and fraud detection vendors identified a certain % IVT (percent “invalid traffic”), for example 1% IVT, that doesn’t mean the other 99% that was not marked as IVT were actually human. It just means they failed to detect any of it to be invalid. Also, to save costs, many advertisers opted not to measure 100% of the impressions. Instead, they paid for 1 in 10 post-impression measurement – the so-called sampling. That means the fraud detection vendor didn’t even look at 90% of the impressions.
- How many humans were your ads shown to? According to Mr. Fou, if you ask this question, you’ll be surprised that NONE of the fraud vendors or analytics platforms can tell you. In an experiment run by FouAnalytics to measure programmatic ads, across a sample of 22 programmatic campaigns, looking at approximately 45,000 impressions, “unique humans” were as low as 17. In contrast, FouAnalytics measurement on a sample of 11 real publisher site reveals there is a lot of humans that can be confirmed.
- A human visit or a bot. How do we measure for humans? It starts with obvious things like mouse movements, clicks, page scrolling, touch events, etc. However, Mr. Fou warns that there are obviously many other things that we have to “check out” before we label a visit to be human. The reason? Bots regularly fake mouse movements, clicks, scrolling, and touch events too. The same principles apply to in-ad measurement too. Different humans click on different places on the ad. It would be very hard to coordinate lots of humans to click on the exact same pixel location (x,y coordinate), i.e. repeated clicks on the same pixel location reveal a bot.
- Not concretely traceable is where fraud occurs. According to the ISBA 2022 study, only 4% of the ads (61 million / 1.3 billion) were matched from participating advertisers to participating publishers. This was up from 2% (31 million / 1.3 billion) in 2020. But still at 4% match, that means the other 96% was not matched. As Mr. Fou explains, that does not mean it was fraudulent or bad, it just means that log level data analysis was not able to track 96% of the ad impressions through the supply chain. Those impressions went to non-participating publishers and “the long tail” where it is common knowledge that lots of ad fraud occurs. To be clear, it’s not 96% fraud; it was 96% not matched from end-to-end. But would you be OK buying something where 96% of what you bought is not concretely traceable? Ad buyers can avoid all of this by buying direct from real publishers, so they concretely know where their dollars went and concretely know where their ads were shown.
- Human CPM (hCPM) changes the game. If you ask for human CPM (hCPM) you will find that you paid far higher costs per human when attempting to buy low CPM inventory. You will also quickly see how few vendors actually know and were actually delivering to you. They delivered trillions of impressions to you. What to do? Ask for impressions shown to humans. Instead of just CPM, ask for hCPM – i.e., which clicks were humans. Instead of a CPC, calculate your own hCPC. Instead of % invalid traffic, ask for % humans. Once you focus on humans, you can not only optimize away from IVT and fraud, you can also optimize towards humans.
Novelty Media, via SmartAdd, has developed a unique mobile media technology framework for video and static content delivery endorsed by Mobile Operators. Unlike other measurement vendors, we can tell you the exact percentage of your ads that were shown to humans. That’s because 100% of our impressions are delivered to verified mobile network subscribers, hence the clicks you get are clicks by humans on the actual click-through URL. Get in touch with us to learn more about our solution and discuss the next steps in optimizing your ad budget.