• by Joe Mandese  @mp_joemandese, April 11, 2025

Following the March 18 release of a controversial study purporting to show a massive ad industry failure in detecting and preventing ads served to non-human “bots” on behalf of Fortune 500 brands, as well as U.S. government agencies, the Media Rating Council (MRC) this morning released a rebuttal asserting the researcher that conducted the study -- Adalytics -- failed to understand the actual business dynamics the ad industry uses to filter such “invalid traffic,” implying that industry systems failed and advertisers were paying for fraudulent traffic.

The problem, the MRC says, is that Adalytics' analysis focused entirely on so-called “pre-bid” detection and filtration, but the industry standard actually utilizes a “back-end” process that filters invalid traffic after ads are served -- meaning that advertisers do not pay for the non-human traffic after the fact.

You can read the MRC’s detailed 10-page rebuttal here, and you can read Adalytics’ 43,547-word report here, but the bottom line is that as solid as Adalytics research may have been, it was looking at the wrong thing and reveals that the fledgling digital advertising analytics “watchdog” is naive about how the ad industry’s business processes actually work.

Adalytics describes its process as “empirical observational research” that detects, observes and analyzes actual digital ad occurrences in the wild, and in its brief history it certainly has published some sensational findings in its reports.

In February, it published a report finding that ads for major brands were being served to sites affiliated with children’s pornography, which got the attention of lawmakers, as well as industry executives.

That study -- which included numerous, salacious images, albeit blurred or blocked out -- did not quantify how many of those ads were actually served to humans vs. bots. In fact, it included a boilerplate “caveats and limitations” statement at its end -– as Adalytics provides in most of its reports -- disclosing: “Interpreting the results of this observational study requires nuance and caution. This study is predicated on publicly available data.”

You can read it at the end of the study here, but it goes on to state that the advertisers it works with “were unable to confirm how many of their ads were directly served adjacent to child sexual abuse materials due to lack of page URL level transparency from their ad tech vendors.”

While the report doesn’t quantify how many -- if any -- actual people were served reputable brand ads on those pages, it raises an important philosophical question: If the ads were served only to bots showing up on those pages, was any harm done to the brands vis a vis those adjacencies?

A similar question could be asked of Adalytics’ more recent pre-bid analysis, which is: If advertisers never paid for the ads served to bots, was any harm done to the brands vis a vis those ads being served?

While arguments could be made that there were energy and time costs associated with the bogus traffic, or that there were indirect costs associated with the fact that advertisers and ad-tech middlemen had to pay for detection and filtration software on the back end to deal with it, but the point is that those processes are what actually deals with it.

Readers should review both Adalytics’ report, as well as the MRC’s rebuttal -- and for that matter, statements made by fraud detection and filtration providers such as this one by DoubleVerify -- and judge the business consequences of the pre-bid analysis for themselves.

But the MRC points out that there is another important reason why its standards currently do not advocate -- and actually discourage -- pre-bid filtration of ads served to non-human traffic in the first place: Because it can give bad actors data signals they could use to reverse-engineer and thwart the actual filtration process.

“We don’t do this lightly,” MRC Executive Director and CEO George Ivie explained to MediaPost, noting: “Adalytics has issued many reports and we’ve never issued a statement. We’re issuing a statement here, because we think there are some implications that Adalytics has drawn that are incomplete or inaccurate and they cross with our standards and the application of our standards, so we felt like we needed to correct the record.”

“Adalytics never showed us their blog post in advance. They never spoke to us about that blog post and never even tried to correct or verify any facts that they were putting forward before they published their blog."