Commentary
ARF Day Two: 'Who Do You Trust?'
- by Tony Jarvis , Op-Ed Contributor, March 27, 2025
Day Two of the Advertising Research Foundation’s (ARF) Audience x Science Conference in New York City provided insights and warnings on AI, outlined the status of Association of National Advertisers’ Aquila venture, and raised questions on the evolution and role of media planning in today’s technopolistic marketplace.
Multiple challenges, limits and dangers associated with human-AI co-evolution were addressed by Northeastern University Professor Tina Eliassi-Rad.
She underlined that the paradox of Big Data is there is not enough information or measurement on any one person to predict a detailed profile of that person.
Consequently, this requires the use of proxy data that, in and of itself, is inadequate or inaccurate. The result for AI analytics and outputs, as she suggested, is that “it is not clear what is fake and what is real.” Generative AI has “lots of risks.”
advertisement
advertisement
With problems identifying “The Truth” – notably in the ad and media measurement world – the question she raised is, “Who do you trust?”
This echoed Rashid Tobaccowala’s stress on the value of trust from the first day of the ARF event.
As, who I would call, an “over-seer” in understanding and measuring the feedback loop between us and AI agents, I believe we can take comfort that Tina and her associates will help us balance exploitation versus exploration in this technological advance for society, which is so critical to enhancing the science of our ad and media industries.
“Education in its broadest sense will save humanity,” she remarked. Hopefully, she is right.
The industry’s on-going issues with the opaqueness of walled garden data, notably by the social media “death stars,” was the opening line from Aquila President-COO Tina Daniels.
This ANA LLC initiative, at its fundamental level, hopes to provide the industry with unduplicated cross-platform reach and frequency at relatively low cost based on “normalized total impressions” (not defined), especially for use in marketing mix models and understanding campaign outcomes.
This undertaking parallels a similar initiative – ISBA’s “Project Origin” in the U.K. – which has been encountering a myriad of complex concerns still to be resolved. Thankfully, neither project is aiming to be, or has ambitions to be, a “media currency.”
However, both are based on the World Federation of Advertisers’ (WFA) “Project HALO” open-source cross-media reach and frequency model.
Importantly, both U.S. and international discussions so far have shied away from the use and value of a meaningful and relevant definition and derivation of an impression to be used in those models.
On top of which, the primary funding and model development is being driven by Google, as well as other major social media platforms. Surely that’s problematic?
Which is why TV broadcasters in the U.K., to say the very least, are leery of involvement or use of their audience data in these cross-media projects and have significant concerns with Google’s influence and reluctance to involve any more than YouTube data.
As Daniels stated, Aquila needs many more partners.
Achieving real, meaningful and relevant unduplicated cross-media reach and frequency based on persons-based measures of actual content exposure is a Holy Grail that is desperately needed and the WFA, ISBA and the ANA are to be applauded for their initiatives.
The objective has become ever more complex, including issues with the use of Big Data, VIDs (virtual IDs) and calibration panels in measuring and understanding our exploding digital world. So, there is a very long way to go to ensure that these terrific initiatives are truly comprehensive and unbiased.
The industry’s eminent economist, Madison and Wall’s Brian Weiser concluded the conference by outlining the status and questions as to whether the future of planning is the platform? The power and influence of the “death stars” that can do it all – from planning to buying, as well as outcomes measurement – in many different formats and via many different devices.
From a detailed analysis by his company, the gap between plans and buys have become ever wider which results in the “platform” becoming the ad spending “line item” rather than media like TV, newspapers, radio, out-of-home, etc. The horizon for this evolution? Likely five years.
His analysis and assessment importantly underpin the current questions and concerns regarding principal-based buying by media agencies, which at one time was considered unethical, possibly even illegal. If this continues to evolve it could fuel the rise of boutique creative agencies, as creative is still king.
In addition, he asked, how would other media owners evolve and compete? And would the “Platforms” measure and grade their own homework? More urgent topics for ARF to address?
It's amazing to me that so much emphasis is on the need for cross platform reach data---by which they mean linear TV versus streaming and, perhaps, digital video. They certainly don't mean radio or print or outdoor media.
But returning to this huge "problem", suppose that a media planner has reach information---or the ability to estimate it based on tabulations of past ad schedules---for a hypothetical linear TV schedule. It's estimated monthly reach is going to be around 45%. The planner is also considering the addition of a streaming buy targeting the same kinds of consumers, which it is believed will deliver a 35% reach. Now the planner has to consider how much reach the streaming portion of the TV plan adds to the linear component. And bear in mind that in a media plan we are talking about next year and not about specific networks or shows--indeed many of the shows that might be employed haven't even been produced as yet. So there's no data to process. It's all theoretical.
OK, so linear TV delivers a 45% reach and streaming a 35% reach. What about the two in combination?
Well you can't go lower than 45% as linear gets you that and you can't realistically think that every person reached by the streaming schedule would be new to the campaign---meaning that the combined reach is 80%. So the "truth" lies somewhere in between these extremes.
If you assumed random duplication your answer would be 64%. But supposing that you had a source like Nielsen, that could process current data across platforms. You might try to simulate a future ad schedule, using networks and shows that are currently being measured as surrogates for what your time buyers might buy and this might tell you that your reach is not 64% but 65.76%
Horrors! You are off by acouple of reach points. Or are you? After all, the data you are using has little to do with your client's ads being seen. Indeed, it vastly overstates ad exposure---in terms of reach and, especially, frequency. So what exactly have you accomplished? Isn't it good enough to tell your client that about three out of five---or 60-65%--- of his consumer prospects might see his ads per month?
PART 1 OF 2
Hey Tony. Sorry you couldn't make it in person.
As you know (because Tina said so), I consult for Aquila. So allow me to address a few things.
First, I want to make it clear to your readers that the phrase "death stars" as a characterization of the so-called walled gardens was not Tina's phraseology. I suspect you put this in quotes to signify that it was your characterization, but I want to make clear it was not Tina's. As a 10-year Google alum, that's not a thing she would say. It is fair to say that she positioned Aquila as the solution to the industry's walled garden problem, about which more in a bit.
Second, I know quite well that the definition of impression is one of your causes celebre (did I pluralize this properly?) I would note that the question of what constitutes an impression is by no means trivial, and that the Aquila intent is to comply with MRC standards for impression measurement and reporting for a cross-platform measurement service (because one of the central mandates of Aquila is governance, which includes submitting to an MRC audit.) I'm guessing that now comprises, what, 1100 pages of standards? Toward that end, the MRC has been advising Aquila from the start (I don't want to name any names, because I'm sure Ron Pinelli doesn't need the attention.)
Third, as regards the role of Google and other platforms in development of Halo, Origin and Aquila: we all know that the notion of "Walled Gardens" is the bane of the buy side's existence. I would suggest that the platforms' collective interest in Aquila is not to somehow "jury-rig" the system, but rather, to embrace a solution into which they are each comfortable sharing their impression-level data, to help remedy the "Walled Garden problem." The privacy compliance they require, and which is inherent in the VID construct, will likewise benefit all consumer-facing media companies with impression-level data who want to provide advertisers with maximum flexibility and transparency while protecting the privacy of their users (and, let's not be naive, their own company privacy, as in, from competitors.)
(Like what you've read? Then don't miss part 2!)
PART 2 OF 2
Where was I? Oh yeah.
This leads directly into point 4. Aquila (and CMM and Halo and Origin) is indeed engineered for a world where the lion's share of impressions are delivered digitally-- addressed, as it were. The largest, best-known TV programmers (and audio programmers) in the US and UK are deeply committed to and invested in streaming. I can tell you that these companies on the left side of the pond understand the value of Aquila as a mechanism for assuring their inventory is surfaced in the best light possible for advertisers. I can't speak for Origin and the UK (and in fact as a contractor, I probably shouldn't speak for Aquila either.) But I can say that every media company Aquila has talked to that has digitally distributed impressions has, to steal a phrase, "leaned in."
And finally point 5, which is really point 1. CMM, Halo, Origin, and Aquila are all the direct outgrowth of the articulation of a need expressed by the world's largest advertisers, for cross-media reach and frequency. I read Ed's comment above, and what I'd say is that this whole thing is being engineered in response to the articulated needs of the advertiser community. Advertisers are concerned about things like how to efficiently extend reach; how to avoid excessive frequency (no one wants to be a part of a poor consumer experience), and how to make every media dollar-- no, every dollar, period-- work as hard as possible. A few percentage points of misallocated reach can mean tens of millions of dollars to a major advertiser. No one thinks this is trivial, certainly not in this day and age.
Granted, CMM/Halo/Aquila/Origin comprise a grand vision with a lot of moving parts. And there are other fine companies also building solutions in the cross-platform measurement space (and I worked at two of them-- three if we count Nielsen, who acquired my first employer, Arbitron.) But in the age of big data, identity, privacy, and digitally distributed impressions, this whole thing is inarguably forward motion. Bumpy, sure. But forward.
Josh, the basic problem with the concern about cross platform reach is the assumption by many advertisers that the industry's "audience studies" are measuring the audience to their ads---or at the very least, the potential audience. Which is not true. Some of the people designated as "reached" do look at or read or listen to the ad but many aren't even present or paid no attention. Yet, ask any CMO or brand manager about this and they will be surprised. They just don't know that.
When a media planner estimates the reach and frequency for the various components of next year's plan, this is nothing more than a theoretical estimate as the buy hasn't been made and the commercial positions that will be bought have'nt been allocated to the brand. So, if the planner is telling the brand that his proposed linear and CTV ad schedule will "reach" 64% of the target a8udience per month, this is a guesstimate, that's all. Indeed, the actual eyes-on-screen reach would probably be closer to 45-50% for the brand's commercials and the average frequency would be half what the planner is promising.
Let's face it, we are asking far too much of our "audience" data. It's simply not good enough to provide accurate real world projections of an ad campaign's true reach or frequency. It was never designed to do that. We need attentiveness measurements not just ad-on-screen measurements.Without that we are just kidding ourselves.