Remember Ad Blocking? How Much Money is Left on the Table?

Everyone has been parroting the ISBA study from 2020 [PDF], which found the same results as three previous studies -- that 50% of every dollar didn't make it from the ad buyer (advertiser) to the ad seller (publisher). Ad tech middlemen took their cut of 50%. That's like you getting half of the pizza or half of the gallon of milk you just bought. Who'd be OK with that?

The 50% was the average (left side), that went to toll-takers. More recent data from Adalytics (right side) shows an even more dire situation. Middlemen take between 20 - 60%, with one taking a 99% cut, leaving 1% for the publisher. The publisher doesn't know this and can't readily calculate this because they rarely know what the advertiser bid for the ad opportunity offered on their site. Advertisers might bid $15 CPMs for a site they recognize, but the publisher gets $3 CPMs. The publisher didn't know the advertiser bid $15. The advertiser didn't know the publisher got $3. If either of them found out, they'd be pissed. That's why adtech middlemen never disclose their mark-ups.

The other thing that most people readily remember is the "15% unknown delta." This meant that the accountants could not see where 15% of the money went, even though they were studying data from 15 mainstream advertisers and 12 mainstream publishers, supposedly a "well-lit neighborhood." The other important point that is missed entirely is that only 12% of the 267 million impressions could be matched at all. That's right, 88% of the impressions could not be traced through the supply chain from buyer to seller.

Combine this observation with what I have written about before: 1) placement reports show large buckets of "(not set)" where ads went to domains or apps that were not known, and 2) fraud verification reports have large buckets called "mobile in-app" with no details of which apps the ads went to. I won't belabor the point here; everyone should realize they are NOT getting what they paid for when buying ads through programmatic channels. And it's way worse than your worst estimates and worst nightmare. If, however, advertisers bought ads from real publishers, directly using programmatic technology, things are entirely different.

Let me reprise a slide from 2016, when I was at The Advertising Research Foundation. We were interested in knowing how much ad blocking was going on because ad blocking had been growing rapidly, and humans that block ads can't be shown ads. Those humans were no longer "addressable." How big of a problem was this? The slide above shows that human activity accounts for about 1/4 of the traffic online. The rest is automated traffic or unknown. The automated traffic is not necessarily the bots that commit ad fraud, but it is simply not human. Of those humans that are online, about 40% of them regularly use ad blocking. There are lots of nuances to the measurement, but here's a good summary of the different points of view -- the range is 20 - 40%. Even without deciding which number to use, you can surmise that millions of humans block ads and that means tens of millions of dollars of lost revenue for publishers on the one hand and billions of dollars lost to ad fraud for advertisers on the other hand. Ads that do get shown via programmatic channels are served disproportionately to bots than to humans, at a 5:1 ratio (74% vs 15%).

Quantify the dollar impact for publishers

The dollar value of lost ad revenue for publishers due to ad blocking is enormous; and it is larger for larger publishers. In the table below, you see some of the largest sites in the U.S. The CPM average comes from data FouAnalytics has observed over years. Monthly bid requests are estimated from the bid firehose. If you assume the low end of ad blocking - 20% - you can see MONTHLY ad revenue lost by each publisher due to ad blocking. At the high end 40% ad blocking means even larger sums of money "left on the table" because those humans that block ads are "not addressable." If publishers sold ads directly and displayed the ads using their own tech, not the 3rd party tech that is blocked by ad blockers, ads could be shown to these valuable humans. How can we say these are humans? It's because humans block ads and bots don't (because their job is to cause the ad to load). When we find a user blocking ads, it is sure to be a human (and therefore a valuable "eyeball").

Quantify the dollar impact for advertisers.

At the very minimum, if advertisers buy direct from publishers they will DOUBLE their money. That's right. If you save the 50% tax that went to middlemen, the full dollar goes to the publisher for showing ads. $1.00 is 2X the $0.50, right? Further, what if you could show ads to the humans that were previously un-"addressable" due to ad blocking? These are humans on major, legitimate publisher sites. If advertisers did this, even if they paid far higher CPMs, the fact that their ads were shown to humans means the ads have a chance of working. Buying ads through programmatic channels is fraught with problems -- 1) 50% tax went to middlemen, 2) 15% of the dollars went to an "unknown delta," 3) 88% of the ads could not be fully traced through the supply chain from end-to-end, and 4) ads are being shown to bots at a 5:1 ratio compared to humans. Take those dollars spent in programmatic and shift it back to real publishers with real human audiences. Real value is greater than "low prices" (which means crap ads). Watch your returns from digital marketing go back up, after 10 years of wandering in the wasteland.