Commentary
Nielsen, We Need A Gauge With A True Ad-Supported View
- by Dave Morgan , Featured Contributor, Yesterday
First things first. Nielsen, thank you for creating a new view of The Gauge that breaks out the viewing of content that is ad-supported. Previously, the report didn’t differentiate the portions of U.S. TV viewing that were on ad-free streaming and whose audiences, therefore, weren’t available to advertisers.
It shows, for example, that “[i]n Q1 of 2025, 72.4% of TV viewing was on ad-supported platforms, compared to 27.6% for ad-free platforms. Traditional TV (cable and broadcast) are tied at about 29% to account for 58% of total ad-supported TV.”
The idea is to give advertisers an idea of where they can connect to their desired audiences.
While I laud the new release, I would like it to go further. Nielsen, how about making it a gauge of the number of ad exposures and multiple-time reach available across broadcast, cable and streaming channels?
Why is a true ad-exposure Gauge important? Because the vast majority of streaming services are “ad-light,” carrying far fewer ads and ad time than their linear brethren. Plus, streaming viewership tends to have a very strong skew toward heavy-streaming viewers. That's one of the reasons that we see the same ads over and over again when we’re watching popular streaming programming.
advertisement
advertisement
Is it hard to make a true ad-exposure Gauge? Of course not. Nielsen has all of the data to do it, as do many ad platform companies that work across both linear and streaming. In fact, in October of last year, comScore released a report with media analyst Evan Shapiro called “Setting the Score” that did exactly that. It showed that while streaming might command more than 45% of all content viewing time in the U.S., it commanded well under 14% of ad-viewing time.
But Nielsen can take this even further for us, and make the Gauge really valuable. How about including a simple graph to show the curve of the declination of reach with the increase of frequency? This is the biggest problem today for CTV advertisers, and it would be great for Nielsen to demonstrate leadership in presenting something that would be enormously beneficial for advertisers, agencies and publishers alike.
What do you think, folks? Should Nielsen lead the way with a true ad-exposure Gauge for TV?
Widespread adoption of the 4As AD-ID would make this easier and more accurate. Nielsen is great at program level ratings because they watermark and index all programming and schedules. But they have a separate system for tracking ads and metadata and it does not have full market coverage. If they adopted a standard like AD-ID they could provide ad unit level ratings that would provide much more transparency into R&F for ad units. But as we know, Nielsen is not fond of standards that they do not control.
Certainly these would be great additions to the landscape. It requires using actual schedules (a large sample) to have the data RxF data right, or some large number of AI produced schedules. If Nielsen won't (or can't) which enterprising company should take the lead to establish itself as a leader. It is worth the effort if the report is shared widely by the online trades etc.
Bruce, Good points. Ad-ID would be a good identifier, but ISCI would also work. Fortunately, Nielsen (and many others) have both syndicated ad occurrences and ACR ad occurrence at a volume necessary to do the Reach & Frequency well. The industry needs it. I hope they consider it -- if not, someone else will, for sure :-)
Jack, Yes, and as I wrote in my comment to Bruce, there isn't a lack of data from ACR, set-top box and panels to do all of the RxF deduplicaed across all forms of linear and streaming with a very high degree of accuracy. I totally agree that it is worth the effort.
Guys while I agree that this is a forward step by Nielsen let's be fair about what we are asking for. Nielsen has a good fix on national set usage and can identify almost all such program content and know how it was accessed and where it originated from. Nielsen also can ID most national commercials--but this is not true--as far as I know--for many local commercials as there are huge numbers of spots put out by small, local advertisers which may not be coded for Nielsen to identify. So asking Nielsen to tally all commercials for "linear TV" is asking too much. I suspect that Comscore faced similar issues.
However, If Nielsen stuck to national TV it should be possible with some effort to calculate the gross number of GRPs--based on average commercial minute ratings--for national ads on broadcast TV, cable, national syndication and CTV. As regards reach, this would require a huge and extremely costly effort to take tens of thousands of ad schedules to calculate some sort of weighted average monthly reach and frequency portrait that compared the various platforms. Even if it was attempted, the result would reflect how many brands are buying time--many with different goals--- but not, necessarily, the true potential of each platform if approached with common goals. So I think that we would be satisfied with an ad GRP comparison on a national basis, which would reflect the way ad "impressions" are split by platform --but for only about 75% of all ad-supported viewing.
I should add that it's not difficult to estimate the approximate split of ad GRPs by platform based on what info we now have from Nielsen. We have done thai for some time for our MDI Direct subscribers and before Nielsen realeased its first report we figured that streaming garnered about 17% of the total. Based on the latest Nielsen findings and our own, exclusive estimates of the average ad loads per platform, it apears that the actual figure is about 16%--meaning that we were very close.
Ed, I agree that tallying all ads by unique ID is a lot of work, many of us do that today for almost all national ads, but that's not I'm calling for. I suggest that the ad supported gauge focus on ad exposure viewing, both number of ads vieeed and time, and build a simple RxF curve or deciles or quintiles chart to show reach across delivery types, top channels and top services.
Dave , I think I understand what you mean by "ad exposure" and "time spent"---the latter taking into account commercial length. In other words, ad GRPs calculated in total ad seconds. If I'm right, that would require a massive new programming effort even if done only for national commercials and I'm not sure what that would get you as the split between commercial lengths is not very great for most platforms. About half of their ad messages are "30s" and about half are "15s' with a few other lengths representing roughly 5-6% of the total. So GRPs figured the current way--without regard to length ----would probably give you the same answer.
The reach and frequency request continues to puzzle me, however. How would Nielsen go about that? Wouldn't it have to tally random samples of schedules--not all schedules---to save on the huge expense of trying the latter? And if it selected random schedules--say a thousand for each platform---wouldn't it have to equate their ad spend in some manner? Also many media plans involve more than one platform. A brand may use linear for reach with a mix of cable and broadcast and lots of prime but rely on CTV only for added frequency against some demo--say younger viewers. Another brand might use CTV as a short term reach extender and cable for added frequency, etc. How would Nielsen know what the goals were for each buy?How would we interpret such media mix-driven data as it appplies to each platform? I think that we are asking too much of Nielsen.
Ed, yes. With all of the second-by-second viewing data from ACR and set-top boxes, many companies, mine included, are already capturing every ad exposure individually (creating true commercial ratings in a sense). It is hard, but lots of data, smart folks and well designed software can do it well. You can cume it up to GRPs, but that's going backwards in many ways. Easy enough to have 95% accuracy at audience cohort/exposure level.
In TV ratings, if you 'watch' a 30 minute program (as an example) but you are bored or interferred after 10 minutes you don't count as a viewer, but as one-third of a viewer.
Why aren't ads calculated in the same basis. I recall some ads (mobile not TV) with claims of massive viewer counts. After a little bit of experimenting it was clear that there was no minimum duration. When a bunch of ad vendors were competing for the spot they were all pinging in competition - one winner but all were claiming they had posted the spot because of the ping. A minimum duration was eventually agreed ... 2 seconds. Why wasn't a proportionate duration introduced rather than a small mimimum?
If we calculate the 'program audience' we rely on (in main) 15-second data.
John, th TV ratings will count you as a "viewer"--of program content or commercials so long as the TV set remains tuned to the content, whether or not you are bored or distracted or leave the room. So both program content and ads are treated the same inflationary way. The only exception is when you notify the system every time you get bored or leave the room--which few panelists ever do--despite occasional prompts to do so. It's simply a case of asking too much of the people meter panelists.
Great point John, streaming and linear show "ratings" and ad calculations are not apples to apples today.we have the data and systems to count linear ads at the "commercial ratings" point, and to normalize streaming to the "average show ratings" used in linear. Do you think the industry is ready for that much truth?
Ed. you are right ... in a way. The TV ratings don't measure 'appreciation or boredom'. Yes of course not every panellist will be 'pressing the button' but many do (can be better in larger homes when little Johnny is bored and leaves). I would even suggest that the person who correctly press's buttons is paying less attention to the content than button pressing.
Dave, I sure hope that the industry is ready for a system that is a 'good' estimate. We're going through with that often in AU.
Overall, there is an old saying ... no research has ever been 100% accurate ... just strive to get as close as possible.
Spot on John. It will be critical not to let the pursuit of perfect get in the way of "good" ratings ... giving us something better than the status quo.
John, I must disagree about the people meter panelists' cooperation levels re indicating that they are or are not watching. The people meter assumptions--based on what info is supplied by panel members regarding on and off "viewing" while content plays on-screen indicate that over 90% of all TV commercials are "watched" --yet camera studies have long shown that the real percentage is about 50% and even this is only a "start to watch" figure. Less than half of the ad's content --once it is being viewed--- is looked at. That's a big difference and the "camera" and other observational studies have been telling us this for many years--as far back as the 1960s. They have shown that about 15-25% of the time the "viewer" is not even present while the rest of the non viewers are so distracted that they don't even look at the screen for two seconds.
Thank you Ed, you make valid comments. They do seem to be based on US research.
'People Meters' were an improvement back in the 1980s and 1990s. But I am not sure where the source for the 90% of all TV commercials being watched came from. We didn't see that magnitude in AU.
Camera studies are generally done on a small cohort knowing exactly what the research purpose is and they strive to be "good citizens" and co-operate. In fact most 'camera studies' are now done on mobile phones and/or household digital equipment. In AU we have devices within the household's rooms that have a TV et. al. Roughly half co-operate, and the other half are extrapolated based on the co-operation data.
I agree that Attention is a very desirable metric. There are some methods of attention rates, but given the plethora of ads I suspect that it will be quite some time before being measured is a standard. P.S. I would wonder how they would be able to measure the Sony Bravia bouncing balls ... I know I loved it but didn't buy one. Yep, research is never 100% correct.