Commentary
Heeding ANA Call To Reform TV Buying
- by Dave Morgan , Featured Contributor, May 28, 2020
With this terrible public health crisis unleashing extraordinary disruption on every part of the media industry, there’s no time like the present to bring some much-needed transformational ad reform to the business. As noted economist Paul Romer has observed, “A crisis is a terrible thing to waste.”
Apparently, the Association of National Advertisers agrees. Less than two weeks ago, its Media Advisory Board called for reform of the television advertising upfront marketplace.
Here are its directives:
"Recognizing the opportunities for change brought forward by the current crisis, the ANA Media Advisory Board advocates that:
• The television upfront shift from a broadcast year to a calendar year to reflect and improve business planning, elevate marketer decision-making, and align television buying with most marketers’ fiscal years. This is an immediate priority.
• The ANA addresses other critical issues pertaining to the marketplace. These include greater visibility to the marketplace, ratings estimates rooted in more realistic research, increased financial flexibility throughout the year, and an improved ability to measure business results from our investments.
advertisement
advertisement
• The ANA elevates long-term media transformation initiatives, including cross-platform measurement and the transition to a digital ecosystem without third-party cookies."
I believe the ANA is right. The implementation of its directive would significantly improve the lot for many media practitioners, not just advertisers. Here are my reasons why:
Timing alignment will only help increase TV ad spend. The timing of the upfront today does not align with how most marketers operate their businesses. Simple. The spring timing of the upfront is an artificial relic of a fall-dominated TV programming debut calendar.
Yes, changing the timing will add further disruption to how TV networks and agency buyers run their businesses -- but creating better alignment with advertisers can only make it easier for them to spend more on TV.
Fix forecasting. At the core of tens of billions of dollars of upfront TV ad commitments each year are sets of planning models produced by the TV network groups and tested against planning models produced by the agency buyers. These models have been much more wrong than right for most of the past 10 years. They are a vestige of the days when ratings kept going up each year, particularly in cable, so if you overestimated your ratings you could always kick the can down the road with make-goods.
This was also one of the reasons that TV ad sellers and buyers have had to stick with such a broad, generalized currency like gross rating points and sex/age demographics. Trying to guarantee for true strategic targets was too hard when your projections were so broad and not very accurate. And none of this process has gotten better, as both agencies and TV companies have gutted their research departments over the past years.
Delivering transparent, accurate and granular audience estimates is certainly quite possible today. All TV ad sellers today have access to very robust predictive analytics, massive sets of directly measured TV viewing data, and lots of data scientists to do the work.
Yes, those models can even be adapted to extraordinary changes in viewing behaviors like we’re seeing today. Fixing this can only create more trust and comfort among advertisers, and result in more spending.
Make cross-platform measurement real. De-duplicating audiences and ad deliveries across television and online video is not rocket science. Both the ANA and the WFA are driving initiatives to help solve this issue, and there is strong participation from major measurement companies and major digital platforms like Google and Facebook. Roku just announced this week that it is going to sell based on proven unduplicated reach on its connected TV ad platform.
This is very doable. Let’s make it happen industrywide as soon as possible.
What do you think? Is this a good time to help advertisers better leverage their TV ad spend -- and ideally spend more?
Dave, if the networks went to a calender year upfront, they would probably have to start the year with a load of reruns in January and part of February as these are traditionally low spending periods for many brands easing into their calendar years. This means that most original programming would probably start later---like March, and the networks might have to back this up with new slates of originals in the fall---the last upfront quarter---- as this is a big spending period. That complicates things as the nature of the fall shows would not be known a year in advance---making rating estimates and audience guarantees very difficult. In fact the sellers might not offer guarantees at all for the fourth quarter. Or they might go to a nine- month upfront and then sell their Fall GRPs in June as they do now.
As for better rating estimates, these have always been an issue---mainly for primetime. Even in the 1960s and early 1970s, as one analyst---Marv Antonowski, then of JWT----noted, if you simply left all of the incumbent shows at exactly the same rating as they attained the previous season---despite new competition and time slot changes---and credited all new shows with the average rating for new shows, you got as good an estimate as the "experts" provided after screening pilots, reviewing the plans for the shows, etc. This is still true today, however, the sellers don't guarantee telecast by telecast ratings for specific shows, but only the total GRPs for all placements combined for a given period---usually a quarter. Moreover, the ratings are tracked carefully so if a shortfall seems likely, adjustments are usually made very quickly---which is why makegoods traditionally amount to only 1-2% of the GRP's bought. That percentage has risen, recently, due to sloppy forecasting regarding matters like "cord cutting" and the growth of SVOD---but it's only a point or so higher. All that is needed is a sensible forecast of the extent of the GRP erosion for the upcoming year as well as a fix on the performance of a given network's program lineup. Yes, it's true that the buyers "reward" networks which have had good seasons in the Nielsens by assuming that they will repeat the following year---but this is a problem that can easily be dealt with if the buyers stop this nonsense and start with zero based estimates each season.
As far as cross platform metrics are concerned, this is not an easy nut to crack as true comparability is lacking. You can't equate digital "page views"---no matter how big the screen, where the "exposure" takes place, what else may be on the screen, etc. with average commercial minute in-home program "viewing" nor can you assume that a "30" has twice the "value" of a "15" simply because its twice as long.
One last point regqarding those outmoded "targeting" metrics---18-49 and 25-54. Since sellers and buyers can make audience guarantees work only if there is a single currency for a buy, these age/sex conglomerate "demos" have served this purpose for a long time. But they are not targeting instruments. If we went to a system where each brand could negotiate using whatever metrics were appropriate for its particular circumstance--still one per buy---you would have a very wide variation from one brand to the other--which could probably be dealt with by both buyers and sellers ---once they got used to such a system. However, those corporate buys would largely have to go and each brand would buy its own TV time.
This is not as easy as it sounds. Approximately 400 corporations account for 90%---maybe more---of national primetime ad dollars. But going brand by brand would replace these huge purchases with separate negotions and post buy stewardship for something like 7,000 brands and likely more---depending on how one defines a brand. The buyers and sellers would require much larger staffs---even if computerized systems were brought in to help---and this means higher costs to the advertisers in time costs and agency fees. Would they pay these costs? Also, many small brands might get short shrift in such negotiations and the agencies who handle lots of brands couldn't share their knowledge of overall pricing with each advertiser as many varying targeting definitions would be in play---the results not being comparable. On the other hand, if we went to a two-upfront plan---as I keep suggesting---one a brand-specific upfront; the other the usual CPM-driven corporate buying upfront, the former might account for only a third of the ad dollars, which makes this idea more managemable.
Ed, thanks for your good points, but I don't think that it is reasonable to assume that the TV networks will actually run their programming schedules much differentlly if they did their upfront at a different time These are massive, well financed companies and they no konger draw a straight line between upfront deals and programming financing. In fact, globak syndication now frequently plays a bigger role. As to having sloppy forecasts because of streaming viewing, we all know that the sloppiness is intended and the total amount of ADU's is now massive. However, since both buyer and seller are part of it, having big liabilities has just become part of the business, and it doesn't help anyone. Streaming viewing is quite predictable with predictive analytics, and linear viewing enormously so. Everyone could negotiate on better numbers if they wanted to.
Curious about how using more realistic ratings estimates will impact both sides of the market. If networks aren't able to increase their CPMs to compensate for the lower estimates, their upfront take could be impacted by as much as 10%-15% just due to negotiating based on more realistic estimates. On the other hand, advertisers are submitting GRP requests during the upfront process that are somewhat unrealistic, unless media companies do a great job of stewarding the buys and delivering ADUs within flight.
Howard, I believe (hope) that introducing a more accurate middle will center negotiations on something much more realistic than just wondering how much criziness is baked into either side.
Speaking about using more realistic rating estimates for an upcoming season, I find it hard to believe that the sellers and buyers aren't, at the very least, guided by what happened during the previous season regarding overall rating levels and the effects of fragmentation. If, for example, the average primetime 18-49 or 25-54 rating on a particular broadcast network was 1.6, does it make sense for the estimate for next season to be padded---baring any unusual new programming initiative----so the audience guarantees are based on an assumed average minute rating of 2.0 and the buyer knows that the network will give make goods to account for the difference. To make this work, the network must either withold a substantial portion of its GRPs to use as makegoods ( not a good idea )---or add commercial clutter to generate the promised GRPs. In the latter case, the make good spots would find themselves in cluttered breaks which means lessened impact.I wonder to what extent this type of negotiating is common on the broadcast networks as opposed to cable---the latter being more likely I would think.
The networks like audience guarantees more than buyers do. It let's them go after more dollars and protects their upside.
When the buy underdelivers, the network typically offers inferior programmimg as A.D.s , something buyers don't like to admit to. And yes, they increase the amount of commercials they air, too.
Syndicators are especially egregious at this practice, with unrealistic guarantees even for returning programs let alone new or new-to-syndication shows.
I've tried to offer networks deals without guarantees, but when you tell them it means less money (because the buyer's rating expectation is more realistic than the seller's), the salesman typically goes for the guarantee with more money.
Dave I agree with the sentiment, but the devil is in the detail.
There is also an element of advertiser expectation exceeding advertiser (non-monetary) participation.
But there is a plethora of uncontrollable variables that makes any audience guarantee virtually impossible. So, why have up-fronts? The answer is simple - to provide a degree of surety that advertising will deliver what is promised. Any TV network that guarantees campaign sales results is kidding themselves. Any advertiser that expects it is even more naive.
Some basics first.
Those who know TV well (and indeed any medium) know that a rating is a representation of 'most likelihood' audience - the average minute audience. For those who have access to more granular level data (e.g. at the minute level) realise that average minute audience is the best expectation - or indeed 'the least wrong' estimate. It is common that the lowest minute may be half that of the highest minute. Any network that trades in peaks is crazy. Even crazier are networks that trade in reach which is an even more fraught metric. [Remember that Reach > Peak Minute > Average Minute).
Further, when you do longitudinal studies of regular programming, the variance from episode to episode can be massive. For example, a programme was up against 'ordinary' competitive line-up, then the competition launched a 'tent-pole' or 'block-buster' competitive programme and the original programme takes a hit.
The thing is ... this is how people watch TV. The more episodes broadcast the greater the variability can be. What some see as 'noise' in the data is actually factual representation of everyday viewing behaviour.
PART II
Then you throw in (say) catch-up and non-linear viewing. For catch-up/time-shift it is quite common that viewing to ad-breaks is only 20% that of viewing to ad-breaks during linear viewing (leaving aside Ed's favourite that people aren't necessarily watching the ads during ad-breaks). But you have to ask yourself why? Is it the network's fault? Or is it bad ads from the advetiser. Should bad ads be rewarded with make-goods? Do we support a race-to-the-bottom?
On the issue of more advanced demographics I agree. These are (in the main) modelled (i.e. not measured) audience data. So go for it, model away. They are targetting demographics (i.e. content skew, lifestage skew, inter-media support ad-weight etc.). But I believe that this modelled cohort needs to be used in conjunction with the 'peer' age/gender/geography 'currency' demographic. You judge effectiveness against the modelled cohort, and efficience against the currency cohort.
Moving on to 'cross-platform' ... this is the BHAG - Big Hairy Audacious Goal.
As Ed has pointed out, you need a high degree of concordance between all the media to make it usable. So leaving aside the differences between all media, let's consider one or two examples.
As described, TV uses (the often much maligned) average minute audience for video. What is the online standard for video (which is now 25 years old)? Judging by the PR releases that flood the inboxes, there is none. For example, some speak in terms of subscribers, some use active subscribers, some use stream starts (handy if your server is flaky and iut depends on whether a start after an ad-breask is counted), some use 'users' with no minimum tresh-hold or re-starts).
So while, overall I agree, I think the issue of consistent standards is THE most important. Mind you, some people will be in for rude shocks ... which rhymes with road-blocks unsurprisingly.
John, thnaks for the great comments. I do agree that it's fundamentally a question of expectations. However, I do think that TV sellers can push things much further on the forecasting (and guaranteeing) front. With massive directly measured TV/AVOD viewing data sets, it is now possible to operate with balanced "panels" in the tens of millions of viewers with lots of longitudinal data on each member (deindentified for privacy protection) spanning back years. While programs are not very predictable anymore, people and their viewing are very predictable, and massive data sets make it not only accurate enough to make business bets on, but can help support the scheduling/optimization of ads on an ongoing basis to insure thawt campaigns projections (and guarantees) are met. This can be done much deeper than just demo GRP's, but work well for reach and many of the business outcome targets that scare folks who use conventional TV planning and trafficking systems.
Of course, what we need most basically, as you point out, is consistency across the various video platforms. I agree. Once you get to apples-to-apples, many might not like what they see.
As has been noted already in this discussion, the use of padded rating estimates is a standard ploy of syndicators launching new shows and this has always been the case. The reason is simple. Buyers pay more per viewer for shows with higher average minute ratings. This applies to the broadcast networks, the cable channels as well as the syndicators and is a long standing tradition in TV. So a syndicator with new talk show, who secretly believes that this new show might garner a 1.0% average minute rating will promise a 2.0 or more since buyers will accept higher CPMs for such a show. Later, when the ratings come in at only a .1.1 and many make goods are needed to pay off the buyers, these are delivered from time that was witheld specifically for this purpose. Cable channels frequently use the same ploy for the same reasons as there are CPM tiers they must cope with that generate higher CPMs for higher cable ratings. The broadcast networks play this game also, but more selectively---often only on certain shows---as their performance is tracked more minutely by the buyers.
My point is simple. The advertisers' media directors, who approve all TV buys, must be perfectly aware of the padded rating shell game as they see the estimates going in and later, once the real ratings are known. Until the advertisers demand that this system be changed---why should they pay higher CPMs for being in shows which attract more viewers---though not necessarily their trageted viewers?-----nothing will happen. Paying more might have made sense in 1974 when a show like "Happy Days" reached 30% of all 18-34s per minute and it made sense from a reach standpoint as well as an image/merchandising standpoint to pay more. But those times are long gone. Now a couple of rating points separates the "high rated" shows from those in the mid range and many programs do well to reach 1% of all adults per minute. It's all well and good to theorize about the availability of granular data, models that can predict ratings more accurately, etc. but we are discussing how TV time is actually bought and sold, not research methods. If advertisers really want change, they should demand that within particular network types and program genres, CPMs should not vary significantly based only on rating level. Rest assured, the buyers and sellers will instantly tidy up their rating estimates.
LOL Ed. I'm trying to straddle both the research methodology and the applicatiion of those ratings when buying a campaign.
While Australia is a smaller market, sellers that gild the lily (to be polite) are pretty much caught out straight away. The programme debuts andf flops. The following day the network gets calls for immediate compensation in a 'near equivalent' programme (demo, cumulative audience level and reach) which is essential for FMCG brands. The week 2 episode is downweighted, and a make-good programme. If Week 2 flops then the campaihn buy is re-assessed. Your point that the days of 30% reach is very valid. The flip-side of that coin is that as viewing is now done in small pockets of audience, small rating shows can add disproptionately high incremental reach (but you need many more spots).
Dave, I also agree that large longitudinal panels are lip-smackingly appealing. A researcher's delight - if the various third-party providers use the same standards, though that is not insurmountable. There is also an issue of how much 'depth' they provide, as opposed to 'can provide'. Privacy looms large. Behavioural data sets are so fluid that they often add noise rather than clarity. But the biggest issue I have confronted is that the data sets (and I am talking big global businesses) all too often have missing data, are late, used incorrect parameters etc. The data sets are ingested and processed, and it is not until you see the resultant data output that you may suspect an error. If the error is massive it can be detected and rectified relatively promptly (a couple of days). If the error is small in magnitude (say, only a sub-set of the data pool is affected), then it gets lost in the day-to-day noise. It may take weeks to find the error, and even longer to get the corrected data set, and then even longer again to re-process the data. Moral of the story - beware, you may get what you asked for.
Ed, excelent pionts. Yes. The bad practicies aren't about a lack of data, but an unwillingness to drive for change. You hit the nail on the head on the average rating point issue. Removing that premium therer would do a lot. Of course, the client media directors need to truly want change. They have the power.