Friday Mailing: Netflix, Disney+ & HBO Max face the question, "What constitutes an impression?"

I recently moderated a panel discussion at the 7th Annual Cynopsis Measurement & Data Conference. It included senior executives from Nexstar Digital (below), Innovid’s TVSquared and Vizio, and we discussed who they are seeing winning the ad dollars in OTT/CTV. [1] 

The organizers recommended I show up earlier to watch previous panels to see if any themes emerged that I could leverage with my panel. So, I was able catch a debate between senior executives from Nielsen, GroupM and Roku on what constitutes (or, how we should define) an impression now that it has multiple definitions across linear, OTT and Connected TV (CTV).

Some punchy takes from GroupM’s Adam Gerber - the Executive Director, US Investment Strategy - pushed that discussion into the uncomfortable territory: “Because nothing is defined the same way, none of the ballparks are interoperable”, and that makes data like Automatic Content Recognition (ACR) “fairly worthless”. [2]

His point about “ballparks” is that walled gardens like streaming services and CTV devices are too walled off for their data on viewers to be more broadly applicable. At an extreme, he is reflecting a point that technology analyst Benedict Evans recently argued in an opinion for The Financial Times: “data is not one thing, but innumerable different collections of information, each of them specific to a particular application, that can’t be used for anything else.”

As Netflix, Disney+ and HBO Max pivot to advertising, I think there is an emerging question of how walled gardens - including their own - may have obstacles to success for these pivots. 

Defining impressions beyond “walled gardens”

The topic discussed on the panel “Impressions - The Great Equalizer?” is a complicated one. The simplest takeaway, there is no lowest common denominator standard for impressions. Meaning, an impression measured on a linear set top box [STB] is not comparable to an impression on a streaming device, even when measured in the same room of the same household.

One consequence of that, as Nielsen’s Chief Data and Research Officer Mainak Mazumdar argued, the “challenge is to turn devices into people… who’s watching the ad?” This is a point Mazumdar laid out more thoroughly in a recent blog post for Nielsen:

In addition to never being intended to be used for measurement, big data isn’t reflective of actual people. There is no mistaking the value of [data from STBs] and ACR, as they provide scale to measurement, but big data is reflective of devices, not of actual people. The data by itself can’t tell you who’s watching and who’s not—which is a fundamental need for advertisers. And when people are removed from the equation, the numbers just won’t add up.

That creates a challenge as linear and digital converge around CTV and linear: 

The increasing supply of new data sources, however, does add complexity to measurement, especially when it may not be connected to real people. Publishers and advertisers will always want the biggest reach possible, but certainly not without the analytical rigor needed to validate it.

So, one key conclusion of the panel was that there may be real value in the impressions data from CTV streaming. But, that data will have limited upside if walled gardens won't allow advertisers to tie it to individuals within households.

The Challenges for Netflix, Disney+ and HBO Max

These questions about how to define an “impression” emerge at a time when advertisers and investors expect big returns from ad-supported versions of Netflix, Disney+ and HBO Max.

All three certainly have the scale in the U.S. to deliver value for advertisers. But, as Jeff Fagel - CMO of infrastructure-as-a-service entreprise software provider Madhive - recently argued, the problem is they are all walled gardens: 

But rather than joining forces and making it easier for brands to get access to this sort of data, television providers, both linear and streaming, are mostly concerned with building walled gardens and using the data they collect for their own purposes.

This does not help our hypothetical brand manager who now has to manage multiple results from multiple providers, few of which provide any sort of easy apples-to-apples comparison that would make it easy to determine the ultimate value of the brand’s massive TV spend.

To put in real world context, Netflix may be able to offer advertisers some of the best data on U.S. subscribers out there. But, as Leichtman Research Group recently learned, across all households (including those with no connected TV devices) there is a mean of 3.9 devices per TV household.

This means the worst case scenario is a subscriber starting a Netflix show on a Vizio Smart TV and then continuing it later on a Roku device. Because Vizio and Roku are television providers who do not share data, nor are part of a consortium of data-sharing, there is no precise way to identify who is watching the ad.

So Netflix will be able to serve ads to a subscriber and tell the ad buyer what it knows about that subscriber, but the ad buyer tracking the consumption across devices will have difficulty piecing together a story about the effectiveness of an ad upon an individual subscriber because Vizio and Roku track that consumer differently. 

Without that precision, Fagel writes that ad buyers increasingly seeking to tell - “who actually saw which ad, what they did after they saw that ad, right down to the ultimate purchase decision” - will fail to generate an “easy apples-to-apples comparison that would make it easy to determine the ultimate value of the brand’s massive TV spend”. 

What if the data “can’t be used for anything else?”

These are not small problems: the Cynopsis panel on impressions was basically concluding that without better data-sharing in the marketplace - something every walled garden existentially refuses to do - there will not be the ability to target and track individuals watching the content.

In turn, that means the extraordinarily value of the data that Netflix, Disney+ and HBO Max are generating on consumers starts to decay across multiple use cases. Instead of becoming more valuable than linear data or even Meta or Google data, it becomes less valuable because the data sets become increasingly specific to the walled gardens where the impressions were served. The worst case scenario is the one both Benedict Evans and Nielsen’s Mazumdar outlined above: Netflix, Disney+ and HBO Max data can’t be used for anything else.

It makes me wonder whether my recommendation for Netflix back in April - it “may find more wins with upfronts and the old-school network TV model” of “non-targeted, highly limited, national advertising across very few breaks.” Even if still suboptimal relative to the power of data-driven, programmatic ad-targeting back-ends, it may result in fewer questions for brand managers and ad buyers to solve across data sets.

Footnotes

[1] The entire program will be online next Tuesday, June 14th. Cynopsis charges $399 to access the entire program, so if you have a research budget I strongly recommend making the time for both my panel and the discussions on measurement and defining impressions/ 

[2] ACR enables a smart TV to listen and/or see what’s playing on screen, thereby allowing advertisers to measure viewership and ad performance, to target ads and to personalize content recommendations.