You thought saved money buying low CPM ads through programmatic?

Nope, you wasted more money instead.

Advertisers often cite the advantage of buying through programmatic channels is “cost efficiency.” That’s just a different word for low CPM prices. Take the following example — the vast majority of the ads were purchased for $2.65 CPM (73%) and up to $5.33 (another 22%).

Did the advertisers save money by buying these low CPM ads through programmatic channels versus paying much higher CPMs by buying directly from mainstream, good publishers? To find out, we looked at FouAnalytics data for both the impressions themselves and then the clicks.

IMPRESSION quality

By measuring where the ads went (which domains and apps showed the ads) and also whether bots caused the ad to load, we can see that half of the impressions were of bad quality (dark red). If you have to discard half of what you bought due to the ads being shown to bots, then you’ve effectively paid double the CPM prices. So the $2.65 just became $5.30 CPMs and the $5.33 just became $10.66 CPMs. This form of analysis focuses on what was wasted — the 50% dark red.

CLICK quality

If we flip it around and look at what is valuable — real clicks from real humans, the calculations become even more extreme. When measuring in ads we don’t get to see human interaction events unless and until the user clicks the ads. Once they do that we have more data points to analyze. So while clicks are not the best signal to use to judge the effectiveness of campaigns, it is the data that we have. So we analyze the clicks to see if they came from humans or bots. From the same campaign as above, we can clearly see that 52% of the clicks came from confirmed bots (dark red) and another 14% came from declared bots (orange). Only 7% of the clicks came from humans (dark blue).

If the clicks from humans are what will drive business outcomes for you, consider what this means for the price you paid to get those clicks. In the above example, if only 7% of the clicks were dark blue, you’d have to buy about 14X more impressions to get the same number of human clicks as in this campaign. So you didn’t save any money by buying $2.65 - $5.33 CPM ads; since you had to buy 14X more. Or put another way, your effective CPMs are 14X higher — $37 - $75, respectively. Wouldn’t it have been better to buy real ads from real publishers with real human audiences in the first place?

Real clicks from real humans seeing ads on real publishers’ sites

In another consumer-facing campaign, when we analyzed the valid clicks only, we can clearly see the clicks came from ads shown to real humans on mainstream publishers’ sites, like the ones listed below. NONE of the valid clicks came from ads shown on long tail sites or in-app environments. This reinforces the recommendation to reduce or eliminate spend in programmatic channels and shift the spend back to as much direct buying as possible with a tight inclusion list of good publishers — ones that you’ve heard of or seen other humans actually visit.

And the click locations make sense too. On mobile devices, real humans have to touch the screen to click something (left chart below) and the (x,y) location of the clicks should make sense too. These were clearly clicks on a normal 300x250 display ad. All data and charts from FouAnalytics — this is the level of analysis that real practitioners look for, when verifying the effectiveness of their own campaigns.

Real practitioners know that the above are observations based on specific campaigns. Your mileage may vary and your campaigns may have different goals and be set up differently, of course. The key point here is that with the additional details that enable you to verify which clicks are real, vs faked bot clicks, you can better analyze and then optimize your own campaigns.