Commentary
My Wish List For The Alt Measurement Movement
- by Dave Morgan , Featured Contributor, January 20, 2022
We have media owners like NBCU and ViacomCBS announcing pilot projects for alternative or secondary promises for video ads sold this coming upfront. We have a range of emerging measurement suppliers like iSpot.TV, Comscore, TVSquared, EDO, TVisionInsights, Samba and VideoAmp attacking parts of the video ad measurement world, many powered by newly public Vizio.
And, critically, long-time measurement leader Nielsen has announced plans to totally transform and replace its historical TV ratings with a Big Data-driven cross-platform video measurement product called Nielsen One, currently in beta.
It’s going to be an exciting year. While I personally think we’re not going to see much change to the primary currency of Nielsen sex/age defined gross rating points used in the vast majority of the TV upfront deals this spring and summer, I do expect to see a healthy amount of secondary promises -- delivering campaigns against a data-defined audience or business outcome metric supplied by one or more of these alt measurement folks -- in many of the deals by the time they get done.
advertisement
advertisement
I’m not at all concerned about the ability of these companies to add some meaningful value, layering more granular targeting and measurement on both linear and streaming ad campaigns. Data-driven and audience-based campaigns have been part of TV advertising for years -- they just haven’t had the widespread adoption and acceptance that many of us have been working toward. Big props to the VAB and Sean Cunningham for really moving the needle here lately.
But there are some specific, acute challenges in the cross-channel world of TV and CTV advertising that really need to be addressed before they get out of control. Here are my top three:
Cross-channel audience overlap. As anyone who has combined viewership data from linear TV and streaming well knows, there is a massive overlap of audiences in the U.S. who watch both -- and do it regularly. Just as we need to plan, buy and measure linear TV audiences on a de-duplicated basis, it’s critical that we do it to access all TV viewership, whether it is delivered through a cloud service, cable box, satellite dish or digital antenna. Pretending that audiences on each are accretive to each other is dishonest and wrong.
Ad frequency control. Conventional TV has not done a great job with frequency control over the past years, bombarding heavy TV viewers with the same ads over and over again. However, as bad as that has been, it pales in comparison to the ridiculous amount of uncapped frequency viewers are hit with on ad-supported streaming channels.
Anyone who watches much streaming knows of what I write. Incredibly, CTV ads are delivered through digital ad servers that could actually fix that. Yes, I know it’s hard to holistically control ad deliveries when there are so many different distributors, but we don’t need to let the pursuit of perfection get in the way of better. This should be measured, called out, and iteratively fixed.
Bots. Smart TV’s hardware and software closely mirrors that of mobile devices. Thus, all of the bots, torrents and ad fraud we’ve seen on mobile is already being seamlessly ported into the TV world.
However, those in mobile haven’t worried much about this issue, since most mobile ad campaigns are bought, priced and measured on performance, not really on impressions, brand impact and reach.
But that’s not the case with TV. We must, must, must solve for this, and only independent, verified panels can do it. Bots can’t join and participate in appropriately verified panels. When channels that no one has ever heard of are claiming millions of viewers and impressions, let common sense dictate. Don’t trust; verify.
What do you think? What are you hoping for from the alt measurement movement?
Dave, as you know, we are on the same wavelength on much of this. However, I'd like to make a few points. Regarding traditional TV and heavy viewers, it's not the fault of TV programmers that some people are very heavy consumers of their content while others are not. Imbalanced frequency of consumption is a fact of life for virtualy all activities. Did you know that 20% of coffee drinkers drink 55% of the coffee; 15% of all adults have 75% of all sexual experiences;10% of the population accounts for 55% of all podcast exposures; 5% of the lying politicians do 67% of the lying, etc., etc. In the case of TV heavy usage is a function of time availability and IQ. You can't fix that by targeting light viewers---young and "smart"?----with your programming unless you are willing to get very small audiences and have all of your shows cancelled. And does it really matter whether some people are "overexposed" to your ads when so much of the time they aren't paying attention anyway?
A number of years ago I took data from MRI and Simmons and compared the share of market for hundreds of brands by heavy, moderate and light viewing TV quintiles. Typically these brands allocated 65-85% of their media spend to TV, hence they all "underexposed" light viewers and vastly "over exposed" frequent viewers---yet their share of market stats were about the same---or close to it---in all of the TV quintiles. How can this be? Simple. Since all of the competing brands in each category had similar media mixes---very heavy in TV----the resulting brand shares---to the extent that they reflected the impact of competitive advertising ---were probably a function of "share of mind". In other words even if you "overexposed" some people rival brands were doing the same and the effects cancelled each other out. Ditto for the light viewers. They got few commercial exposures for all of the brands. Result: no change in brand shares.
The CTV situation is very different and is caused largely by computerized buying which overloads a few low CPM venues with too many spots---at the sacrifice of reach. That needs fixing.
Continuing my reply, the fact that some TV buys are now being refined with added metrics and guarantees based on those metrics is probably a good thing as it makes people think. However, no seller is going to guarantee a buyer whatever the buyer demands----like a 20% increase in sales---or a 15% CTR to the client's website. Rather the sellers will closely examine data for each metric to determine the usual result and the high-low range and make guarantees in those contexts so that they are assured of delivering the required goods , say, 90% of the time. If the remaining 10% fall short of what was promised the makegoods ---or other compensation---- are easily paid for by the 25-35% CPM premium that the buyer paid for this kind of transaction. In short, the sellers are really calling the shots. While its good to see something different, I doubt that big things in ad impact and advertiser product sales will come of it---but this remains to be seen. Let's hope for the best.
Dave, you succintly narrowed it down to three critical issues to address. Perhaps I am somewhat naive given their recent track record, but I'm hopeful that Nielsen's new measurement product will at least address the first two. Not sure what they can/will do about the bots.
share of voice is certainly a relevant measure; was using it 40 years ago planning/buying for p&g and then brought the practice to General Motors. the thing is, however, that as viewing fractionalizes, measuring and executing against a media strategy that encompasses SOV becomes very difficult if one is not willing to etend reach protocols aggressively. in this case, victory goes to the aggressive marketer who will be a first mover in adapting and creating new measurment grids on new platforms. i fear that many in the business will be slow or simply don't have the depth of talent to get to a refined level of execution and then keep evolving as the learning process grows. and many eassurement protcols will bite the dust without the capability to blend input.
look at it this way --if you implement an aggressivedreach strategy through experimentation and measurement you WILL certainly increase you impact against an audienc segment, your choice, that your competitors are ignoring. bingo.
Ed, yes. We are largely in agreement here. To be clear, I don't blame TV programmers for the fact that some veiwers watch more than others. I blame them for not better managing their ad schedulng to improve the frequency of ads they see. They watch programming predictabilty differently than light viewers and their ad experience can be made demonstrably better with some work.
Neil, I'm with you. I do believe that Nielsen is going to make a difference with Nielsen One and that it will help address some of the problems.
Phil, for sure. We need to courageous marketers willing to push the envelope to make a difference. They will be rewarded with better results if they can push their reach.
Dave, traditional TV time sellers typically rotate advertiser commercials on a fair basis across the various shows they buy time in and they also ensure that the same message ---or those of competing brands ----do not appear in the same breaks---unless the buyer wants it otherwise. There isn't much else that they can do regarding how many commercials a heavy viewer is "exposed to". There are exceptions, such as extended baseball games where the advertiser's ads are repeated ad nauseum--but what's the seller to do----not run half of the ads and give back the money? Also, some advertisers still use what's called "impact scheduling"---airing the same commercial twice in the same break---but this is rarely seen these days. The over scheduling problem is really a CTV issue as it has been organized along digital media not traditional media lines---yet CTV sellers covet the big bucks of "legacy TV" branding advertisers, not direct marketers. Something has to give---I wonder who will blink first?
Ed, very true, but it doesn't have to be that way. Ad rotation on teh linear TV porgramming could be done much smarter, and still stay within with advertisers and programmers constraints. It is not. Almost everything is blindly rotated as long as it meets deal terms, and that creates sub-otimal viewer experiences and poor frequency management for many advertisers. Progrrammers can and should do it better, and then charge premiums for the incremental reach they generate by using the same inventory and same ad schedules smarter.
For sure, as you mention, the additoin of streaming/CTV into the mix makes it more complicated, but not to computuers. It's a very solvable porblem if folks want to solve it - maybe not perfrfectly, but appropriately.
Dave, to do what you suggest, the sellers would have to find a way to schedule each screen's ad messages individually based on the screen's overall frequency of exposure to an advertiser's total ad schedule across venues. They simply wouldn't have the information to work with on an ongoing basis what with new shows appearing, old ones being cancelled, programs gaining or losing viewers, etc. Moreover, sending ad messages to TV sets or digital device screens---no matter how you configure it---- doesn't tell you if anyone watched the commercials or, in most cases, who watched. The digital folks think that they are able to control how often a consumer "sees" each message and that this can be applied to "linear TV" Sorry---they don't and it can't.
Ed, the data and technology exists today to do this. It starts with hybridizing Nielsen's respondent data and massive data sets of direct viewing, which is now available on many tens of millions of viewers and what they watch on both linear and on-demand on a second-by-second basis, and then building a prediction model on what every screen and viewer will watch in the future at 300mm+ P2+ and append that to a demo/purchase model for each one. You use that prediiction to drive planning and activation and, to the extend that the stewardship systems permit, at the show and pod level. The CTV ad server is the easy part. Getting folks running the stewardship systems to input the data is harder ,which is why it is more of a business reengineering and social engineering problem, not a technical one.
We do it an Simulmedia today, and while we only apply it to our own campaigns, we have given netwokrs the output many times to the extent that they have wanted to try to better optimize their own inventory, particulalry promos. But, as you would imagine, since neither the agencies nor the clients are jumpping up and down about the ad frequency problems, most networks/porgrammers see this as a problem where the juice doesn't justify the squeeze.
Dave, I'm well aware that this kind of thing can be simulated or seemingly handled by statistical models---and, I consider that to be a useful first step and a good one for your company to develop and use for your clients. However the reality is that even if the pairings and matchups are quessed ad in a "scientific" manner and even if the sellers could use them to schedule and reschedule commercials so that each screen gets a certain dosage and no more---or less---we are using information which gives us a very inflated measure of commercial audience. Not slightly off but way off. So when you send a screen six messages the average consumer living in that home may be watching the program in question only half of the time and of those who are "watching" the program, only 35-40% even look at your message for a few seconds. In short, we have a very long way to go if we really think that we can dictate---or control---how often a consumer actually sees a commercial.
Ed, Dave. Great conversation. Two greats!