Commentary
Is The TV Business Ready For Ad Ratings?
- by Dave Morgan , Featured Contributor, June 9, 2022
I’m sure that many of you more digitally focused readers might be a bit surprised to learn that most TV ad campaigns in the U.S. are bought, sold and measured not on how many people actually watched the ad, but instead on the average number of people who watched the entire show. Of course, as any of us who has watched TV since the early 1970s knows, the broad-based introduction of the remote control has guaranteed that channel surfing is endemic during ad breaks.
Why does this matter? Because program ratings are typically 5% to 15% percent higher than ad ratings.
advertisement
advertisement
As those in the world of advanced TV advertising know, when you measure the audiences of ads versus average audience ratings across the entire 30- or 60-minute show, the difference is never less than 5% and frequently closer to 15%.
Importantly, not all ad breaks retain the audience the same way, either. Those airing at the beginning of a show and those that are shorter retain better. NFL broadcasts are among those that retain the best. Local ad breaks with lots and lots of ads retain the worst.
I’m really excited about all of the talk of new currencies for TV as it merges with CTV and the digital ad ecosystem. It's great that we’re now seeing companies like iSpotTV, VIZIO and Samsung bring massive directly measured panels to a market that previously only had Nielsen’s highly curated but relatively small panel.
This competition has caused Nielsen to create Nielsen One, its hybrid offering for the future. It’s also caused dozens of other companies to enter the TV ad measurement field.
However we go forward, and with whatever new suppliers, I hope that -- at the least -- we attack some of the real, low-hanging fruit in the gaps in today’s TV ad measurement. Content versus ad ratings should be first in line.
How can we make TV and CTV ad buys apples-to-apples when the former’s audience numbers are inflated by people that watched the show, but not the ad, and the latter are only based on those whom the server counted as actually receiving the ad?
Simply, we can’t. So let’s fix it. What do you think?
Program ratings or deliveries haven't been the currency since C3 so I don't understand why the article seems to go there. Advertisers have always been offered fair rotations for their schedule of spots and the different performance by pod and pod position have been known for decades. Not sure any of this breaks new ground.
Echoing Jack - incorrect reporting here. What is about to change is availability of individual commercial performance instead of an average of all commercial minutes in a program.
I'm surprised by this one, Dave. National TV show ratings are tabulated by Nielsen based only on those minutes when commercials are the main content and where they were on-screen. Because Nielsen doesn't really know whether anyone is watching during a given minute of content---unless its system is notified otherwise it assumes that the person who claimed to be "viewing" when the channel was first selected is still "watching" when the commercials come on screen.
People mistakenly accept these "average commercial minute ratings" as representing actual commercial "exposure"---or "viewing", however it is clear from observational and other research that substantial percentages of program viewers absent themselves during commercials while many who remain pay no attention.
As for local TV the ratings are even more inflated as they are for quarter hour "audiences" not average minutes and commercial zappers are counted as if they "watched".
But back to the basic question, I stand strongly behind the need for attentiveness measures in any new national TV rating service and this applies for CTV as well as digital video. However the industry isn't ready for such a vital improvement as the sellers ---who will foot much of the bill for any rating service---- will, no doubt, veto such a refinement.They want big numbers---no surprise there. As advertiser CMO's don't seem to care and are unwilling to spend any money for better ratings we will be stuck with set usage as the basic "audience" measurement metric and 'big data" panels melded together to generate huge samples for "granular"analyses.But we will still not know who watched the commercials.
Jack, Ray, yes. I didn't craft this piece very well. I should have pointed out that plenty of deals are done on six minute block average, which frequently show the same level of deviation between the ad minutes and the content minutes, but they're not being done on actual ad exposure. Certainly a big issue is that Nielsen's AMRLD is minute by minute, not second by second, but that data can be released at the second level (probably will be part of NielsenOne, I hope) and for sure many of the new callengers have their data avaiable at that granularity. I should not have made the point of the column only about average rpogram ratings (of course, in Canada it's still about average season ratings in many cases) since the six minute blocks show plently of deviation. My point is that we can now meausre at the ed exposure level and we should make that the default and move away from any averages that include content viewing.
Ed, very good points. Yes, the local quartrer hour have plenty of variabiilty between the ad and content viewing numbers.
And I totally agree that ad effectiveness measures will become an increasingly bigger part of the picture, but I don't see that as mutually exclusive to more granular measures on the ad exposure itself and getting rid of any averages that also include content viewing.
Dave, while I'm probably the biggest proponent of attentiveness measures that you will find---been fighting this bttle for 50 years---- but I see little chance of it being included in the next round of national TV rating services, including Nielsen's new, "big data" service. The sellers won't allow it.
But even if they did and we had access to attentive audience projections for every commercial, all that would happen would be the creation of a new GRP currency---producing much lower GRPs than the old one, in aggregate. The problem is how do you define "attention". If you take the IAB's definition---the two-second rule--and count any viewer as "attentive" if he/she had their eyes on the screen for at least two seconds this will seem to favor short commercials over longer ones---but suc ha definition does not account for probable levels of message communication. How long does it take for a typical "15" viewer to get the message? Is the learning threshold, 5 seconds or time7 seconds? What about "30s"? Is the average learning time threshold 8 seconds or 10 or 15?.
There are some who suggest that time spent be used as this avoids the definitional issue I just raised. Under the time spent concept every second of attentiveness has equal value. It's simple and workable---but, in my view far too simplistic. Many seconds of attentiveness---all at the low end of the scale---probably have no value while those at the higher end probably have more value as they flesh out the story telling function of most commercials for those who have chosen to watch the entire message.
Unfortunately, we don't have answers to any of the basic questions about what the data means and how it might be used. And advertiser brand management---the COMs---are notably absent from such discussions and, to be truthful, from the entire debate about what should be measured by a new national TV rating service. So we will probably wind up with nothing more than a seller orchestrated , "big data" service where we have a huge sample size that provides "granular" data based on device usage with no idea what the information means---if anything---- or how to interpret it.
Sorry for several typos in my last reply---for example, I meant advertiser CMOs---not COMs. Sigh! If only we had an editing option.
Ed, a quick question.
I'm not sure whether you are referring to the 'attentive audience' as the average attention during the ad break, or the attention level to each ad within the ad break.
I suspect you mean individual ads.
First question - as the broadcaster does not 'own' the ad who would pay for the individual ad attentiveness. In my decades of experience I think it is a nice round number. Zero. If I was a broadcaster I wouldn't be keen to pay for someone else's research.
Second, how will we define 'attentiveness'. Do we use a duration threshold a la IAB? That is more of a metric of presence - (begrugding?) tolerance, ennui or just couldn't be bothered getting off the lounge rather than attentiveness. Then we need to know whether the mere presence for a duration produces attentiveness.
Third, does this mean that every ad's attentiveness and for every day. With around 16 minutes per hour of ads on TV, and if they are all 15s that would be around 1,500 ads to be studied per day per braodcaster -probably closer to 1,200 ads to be studied.
In principle I agree. I just don't know how in the real world that this could be done economically.
Plus many smaller advertisers couldn't give a tinker's cuss. Maybe advertisers who are ultra keen on individual ad attentiveness conduct the requisite research to establish a norm for their ad content. Oh, and don't forget that creative content fares differently depending on the programme content that it is shown in as well as Position-In-Break.
Great points John. The network has no control over the creative of the ad. Should they suffer if an ad doesn't create attention, regardless of how that is quantified? Should the ad that decreased attention, or perhaps drove viewers away, be ostracized or penalized? These questions have not been ignored. They are among the many reasons attention is not used to buy and sell ad time. We still will also have to deal with the sticky issue of ad exposure. Is the two second rule to be applied here, as in the digital space? That would seem to be a step backwards in measurement if second by second precision is what's coming.
John and Jack, no one is suggesting that the seller is responsible for the impact that an ad message has upon its audience---at least not I.
However, I'm surprised at the resistence to measuring attentiveness to whatever is on a screen---ads or program content. For, example, let's say that a properly sampled and maintained national rating panel was set up of whatever size is deemed necessary to provide useful information for the vast majority of TV viewing situations. This panel would have every set monitored by an "eye camera" ---perhaps in the same manner as TVision's current panel. The "eye cameras" would determine which household members were watching any portion of each program tuned in. They would also record attentive ---eyes-on-screen--- viiewing for every commercial that appeared on the screen, regardless of duration. The data would be "granular"---second by second---hence the only question would be how to tally the information.
So what's the issue? For years we have been getting "average commercial minute viewer " projections from Nielsen---which vastly inflate actual commercial viewing. Nielsen also can supply such information commercial by commercial and second by second and has done so for a number of the large agencies who hoped to find additional insights ---but got very little out of the exercise. Why? Because the data was not sensitive as it didn't reflect actual viewing. All I'm proposing is that we get accurate program and commercial viewing information via attentiveness measures---instead of using inaccurate estimates. In short, the new information is nothing more than a correction of the ratings the industry has been using---not a change in direction.
As to how the information is tabulated, that can be left up to an industry debate about reporting every commercial separately, or averaging them by length or showing break by break data, etc. Some will favor counting all attentive seconds as worthwhile and using time spent as the metric, others---like myself---will sugges tthe creation of attentiveness time thresholds for each commercial length--like count only those who saw at least 8 seconds of a "15" and 12 seconds of a "30". But all of this would be determined---hopefully in a logical manner---once we had attenntiveness measures.
John,
Very good points about attentiveness. While I belive that attentivess and effectiveness will become more important measurements, I doubt we will see them in TV & CTV for other than secondary promises in unique situations for premiums on top of normal CPM's.
However, as far as measuring indvidual video ad niews, we do have guidance from teh digital video world. Facebook wanted credit for an vvideo viewed more than 0 (to get credit for video in news feeds that users scrooled by), and the industry rejected it, finally settling on 2 seconds, in most cases, though some require 6 seconds of viewing. Under any circumstance, now that it is possible to measure each TV ad view discreetly, I beleive that we should transact that way, and get rid of averages that inlcude content viewed other than the ad in question.
Dave, how does FB or any platform that sets up a time-on screen definition of "viewing" know that anyone is "viewing? They don't---unless there is some sort of overt action---like clicking through to an advertiser's website---but such actions represent a tiny portion of the "audience"---often far less than 1%. What about the rest?
As an additional comment on the subject of attentiveness I recognize the terror that any shrinking of the reported "audience" holds for sales folks. I've seen it in print media where hte push for first "total audience", including pass- along "readers" generated bigger numbers and, later, when the recent reading methodology produced much higher 'readership" numbers than the through-the book, visually aided, recall method. Publishers thought that bigger numbers meant greater ad revenues, yet exactly the opposite happened. When advertisers wanted, say, 50 GRPs per month via magazines and bigger audience figures showed them that this could be attained with fewer insertions, they simply directed the money "saved", elsewhere---mainly to TV.
The same point applies to TV with attentiveness metrics. Yep, suddenly the numbers of commercial viewers would drop significantly from what's being used as the current commercial viewer"currency". And this would, apply to all forms of "TV"---CTV, AVOd, etc. not just "linear TV". Would this cause advertisers to desert the medium and switch their spending to radio or magazines? Of course not. Far more likely---once the shock of the truth hit them, many advertisers would consider increasing their "TV" ad spend. That's my belief. Why? Because most branding advertisers are wedded to TV-style communications and they will not desert it---even if their CPMs are calculated on a more meaningful basis.
Ed, it is only an impression number, just like legacy ad metrics, and doesn't speak to attentiveness. My point is that we have a broad, averaged impression/expsure number today that represents the currency and it can and should be made more accurate. And, importantly, I don't see counting impressions better as mutually exclusive with adding attentitivess and effectiveness measures.
So is the proposal to measure attentiveness but not use it to buy and sell the ad inventory? If you create an 8 second rule or a two second rule for ad exposure, aren't you back to counting and paying for impressions?
Jack, yes. I think that metrics like attentiveness and effectivenss will drive price, but the transactions for linear and streaming TV will be impressions for a long time going forward. I don't know any other way to drive a scaled $85 bilion US spend category ($65B linear & $20B CTV) any other way and have any sense of comparability, stability, etc.
Guys, for national TV we have been using a "measurement" that, in effect, tells advertisers how many people--by demo---"watched" their commercials---or very close to that via an "average commercial minute viewing" metric. And anytime an advertiser or seller wants to bore down to spcific commercials, Nielsen can do that for them---at a price. The data is there. In other words, the data is being used as if it represented attention.
What's being proposed is nothing more than doing what do now--- but doing so correctly. How? Simple. Measure---via an observational method--- exactly who in the household is present at any time in a telecast when program or commercial content is on the screen--- and do so second by second. No more assuming that the claimed program viewer "saw" the commercial. Only count those who actually did so.
As to how such data would be reported, that's up to the various parties---sellers and buyers,mainly. But the data would be available on a graular basis---just as it is now---except we wouldn't be counting a claimed program viewer who left the room or didn't look at the screen when a commercial played out as "watching"every second of the telecast when the channel wasn't changed.. The "viewer"would have to have been in the room and looking the screen for a period of time to qualify as a program content viewer. The same applies to being counted as a commercial viewer. Again, the exact tabulating specs would be decided by an industry committee but the detailed data would be available for those who wanted it.
Of course, this would greatly reduce the size of national TV commercial ratings as, for the first time we would get the truth instead of findings that tell us that a typical viewer almost never leaves the room and "watches" almost every commercial that appears on the screen---both being nonsense.
Being a realist I don't think that the sellers---who rule because they do most of the funding--- will allow attentiveness to be included in national TV rating studies---especially the one that eventually wins out and becomes dominant. Instead, we will get vastly inflated "impressions" and advertisers will continue to be mislead as to the size of the audiences to their ad campaigns.
I am perfectly content selling opportunities to see. The advertiser can then add any factor they want to that so long as they accept responsibility for it. Attention will be driven more by the ad creative than anything else. It the advertiser wants an attention metric, fine, so long as the publisher/seller doesn't get penalized for the lack of attention.
Jack, what does an "average commercial minute viewer" stat, as reported by Nielsen, mean to advertisers and time buyers? Is is "opportunity to view" or is it what it is called---"viewing"---in other words, attention to the commercial. Even if your answer is "opportunity to view" how can that be if 30% of those who were in the room just prior to an average break---per TVision's "cameras"---- aren't there during the average commercial? What opportunity to view did they have?
As for attention being a function of creative and other advertising variables---of course---but sellers can and do affect the levels of attention that advertisers attain in a number of ways. One is the degree of ad clutter in their breaks. The greater the clutter the lower the attention--as more people leave the room or pay no attention. The nature of the program content is another key factor. A seller who offers more engaging original program fare---usually dramas---"delivers" more of the program's audience to each break as people are afraid of missing the return of the content when the break is over---so many stay put. In contrast, put out frivolous reality or sitcom progrmming or varieties and your holding power diminishes---meaning that fewer of the program viewers see the average commercial in such shows.
The fact is that what we have now is an "attentiveness" measurement that is called something else--"average commercial minute audience"----but in reality it is taken to mean attentiveness---"watching the commercials"---by just about everybody. The reality also is that the TV rating services were never designed to measure actual second by second viewing---only program "audience"---but not on a granular basis. We have allowed ourselves to confuse the seeming precision of meters with audience measurement---hence the demand that the ratings be broken down only for those minutes when commercials appear on our TV screens. Unfortunately, set usage meters can't tell us who is present and who is watching---nor how much they see. It's about time that we went for a people measurement and stopped kidding ourselves.
All can be done but we will not be able to afford a tvision system to provide stable data on every single national avail much less local. Prices are set by supply and demand, just like for every product. If someone walks out when your ad is on screen, why should the network suffer. I'm sure the money would be found if the nets thought you new proposed metrics would increase demand and prices. Likewise, if P & G or other sophisticated advertisers thought it would generate sufficient ROI it would be there. In the end it isn't a methodology issue. It's economics.
Jack, I'm not even thinking about adding attentiveness to local TV ratings---you are right, it would be too expensive for the stations. But national TV is quite a different matter. Nobody is talking about a panel of millions of homes with all of their sets monitored regarding attentiveness. A far more likely scenario, as a compromise solution, would be for a metered and attentiveness panel of, say, 25,000-50,000 which would supply factors which could be applied against a much larger panel of set metered-only homes.
I should also add that there are many more uses for attentiveness data than merely time buying and selling. These are for TV programmers--- as a way to evaluate how well the shows are watched---not the commercials---and for advertisers to gain, for the first time, new insights on how their ad campaigns are working, when commercials are wearing out, what their real ad reach and frequencies are, etc. etc. The point being that the funding need not come mostly from the sellers---advertisers should pay for advertising impact findings and program producers or network program departments for valuable information about how their shows are being viewed, which stars or elements attract the most attention, etc.
Bear in mind that I don't expect any of this to happen. Rather, we will get "big data" set usage "impressions" just as we now gulp down such info from much smaller meter panels. Nielsen will, no doubt, add viewer- per -set ratios culled from its small people meter panel and everyone will be happy---except the resulting "audience" projections will not tell us who watched what and we will continue to accept that the average viewer almost never leaves the room and almost always "watches" every commercial.
Here we can agree. Not likely to happen unless there is sufficient belief that the investment will be profitable. Others have tried less expensive proxies but they weren't supported in the marketplace. We can agree there is little appetite to risk this as an investment.
Jack, I'm not sure that we agree on the likely outcome for the same reasons.
I doubt that it's a question of attentiveness being a risky investment. Rather its the fear that much smaller numbers will have a negative impact on ad revenues---a fear held by most sellers regardless of the medium.
Actually, if advertisers realized that they are getting much less real ad exposure in TV than they have ben led to believe, rather than flee to radio, magazines or digital display ads---which have exactly the same issues and are not able to convey their mesages as TV does---"sight, sound and motion"----many TV brands would probably consider spending more, not less, on TV to improve their coverage.
But try explaining this to a media time seller who has no idea how brands think nor how the advertising process eolves and works. In this case, I think that fear of smaller numbers is the main explanation not the cost of adding attentiveness measures.
No seller will be convinced to spend more to see lower numbers. A smart buyer with a way to include attention measures in their media purchase/scheduling process should have an advantage over competitors, no? If such a system will give an advertiser sufficient advantage, they should use it to their advantage, and happily pay for it.
These types of discussions have been taking place for at least two decades. The arguments you pose have merit, but it's been decades and nobody has found a way to make it happen. Whether for different reasons or the same, it's not happening.
I agree, Jack. It's not happening.
Hi Dave, I align with the idea that established currencies have created a measurement void in the industry and that this has definitely led new ground-breaking solutions to arise.
However, and as an answer to your final paragraph, new cross-media audience measurement solutions (that allow you to compare apples to apples) do exist, and are ready to accurately pinpoint consumer ad exposure in cross-media campaigns.
At Beatgrid, we track deduplicated cross-platform measurement from a single source for all campaigns. By applying ACR technology software, we use, from a single source, person-level deterministic cross-media ad exposure data (instead of claimed exposure), helping advertisers understand how their cross-channel campaigns are performing.
By giving media buyers and planners well-detailed campaign measurement data like incremental reach, frequency, ad cut-through, brand lift, in-store visitation uplift, and others, we can solve the everlasting problem of “half the money I spend on advertising is wasted; the trouble is I don't know which half", and help them optimize their media investments with greater efficiency.