TV Advertising's Future Will Focus On What We Actually DO With Data

Data is not going to define the future of successful TV advertising, any more than it has defined digital advertising. I know that’s not what you hear at most conferences, but as someone who has worked at the center of data-driven advertising for the past 25 years, I can say this with certainty.

Data is not the new oil. I know you hear that phrase about data being “the new oil” all the time at cocktail parties and conference keynotes. You see it on cool-looking slide presentations in conference rooms. You read it in business publications. You even see it on magazine covers. But that doesn’t make it true.

Oil is valuable not just because it can be directly converted into harnessable energy, but because it is finite and scarce, expensive to extract, refine and transport.

Data has none of those qualities.  It is not scarce, but the opposite: plentiful, and becoming more plentiful every day, growing exponentially as more digital systems interact with more people and more systems, creating more “data exhaust.”  



Further, data does not produce value independently. You can’t just refine data and burn it as you can oil. Not only does most data need to be refined and cleaned, it also needs to be combined with other data, normalized, analyzed -- and most importantly, inserted into an algorithm and applied to some other process before its value can be fully realized.

Algorithms and processes matter more than data. Don't get me wrong. Data can create value, but its relative contribution to most situations, particularly in the world of advertising, is much less valuable than the algorithms or processes with which it interacts.

People were searching for things on the web for years in the mid- and late 1990s, but until Larry Page and Sergey Brin created the page rank algorithm and launched Google, users’ searches were not particularly fulfilling, despite the fact that what they were looking for -- as well as the data of the web -- was all in existence and quite accessible.

Likewise, until Bill Gross and his team created the bid-based cost-per-click monetization model at (Overture), search was not a very valuable or salable advertising platform. Today, thanks to those algorithms and processes, search advertising is expected to reach $95 billion globally this year, according to Zenith.

We’re pretty mediocre at leveraging data in digital display. How will we do with TV? Data is not the future of TV advertising. However, what we do with data will be TV advertising’s future.

There’s no question that the TV advertising experience today leaves much to be desired. Viewers get too many redundant and irrelevant ads. Advertisers have campaigns that reach way too many of the wrong people and contain incredible amounts of wasted frequency. As Erwin Ephron famously told us years ago, “Frequency is crabgrass.”

So, if we want to know how data is likely to be used in TV advertising, looking at how folks have used it in digital display is probably a good place to start. My assessment: Data-driven improvement in the digital display ad world today is mediocre at best. Fundamentally, from a user perspective, we’re not much more evolved than highly scaled e-commerce retargeting: the famous pair of shoes that follow you around.

Incredibly, our industry started doing that kind of retargeting 20 years ago, and built industrialized systems to do it in the mid-‘00s. We haven’t yet seen holistic solutions that collectively consider the user, the advertiser and the publisher, leveraging available data and truly delivering fewer, more relevant ads that not only yield better for advertisers and publishers, but truly delight users by providing offers, information or entertainment.

All this will happen someday, but that lag is not caused by  lack of data.

Today, lots of folks are talking about building and deploying industrial systems to leverage “massive pools of data” to make TV advertising work like digital. I am hopeful, but I also hope we don’t think it’s really just about the data. It will be about what we do with the data.

For sure, we need to do much better than we’ve done with digital display. Television and its massive loyal audiences and advertisers are too precious to squander. Let’s make sure to get the algorithms and processes right.

What do you think?

16 comments about "TV Advertising's Future Will Focus On What We Actually DO With Data".
Check to receive email when comments are posted.
  1. Mark Green from TVision Insights, July 12, 2018 at 4:31 p.m.

    Agree that data is not the new oil. The other question about data is: what is the information that it provides? Take Nielsen's gold standard peoplemeter data. Does this really report who is watching television? It tells us the period of time between pushing buttons and what the television is tuned to during that time segment. On the digital side, there is a good deal of analysis and testing of noisy data feeds to separate signals from the noise. Terms like IVT and viewability are outcomes of this processing. And there is always talk of changing collection methods to produce less noisy data. Maybe, it is time to start thinking about this on the television side too. Let's identify the seconds people are looking at the television instead of just pushing buttons. Maybe looking at the television could become the new "watching". It would certainly be a less noisy signal of watching.

  2. Ed Papazian from Media Dynamics Inc, July 12, 2018 at 6:21 p.m.

    Dave, I agree with Mark about the quality  and relevance of the "data" we use for TV time buying and media planning, which generally overstates advertising exposures by 50% of more in the true sense--was anyone "watching" the commercial. In that context, Erwin's oft-qouted comment that "frequency is crabgrass" is difficult to swallow as he was accepting the simplistic audience measurements  still in use as if they signified real world frequency---they don't. When the surveys and our R&F formulations tell us that an average prospect---no matter how well defined---is reached, say, 6 times per month while 20% of our audience is reached 12 times and we feel that both--- and especially, the latter--- are just too much we seek ways to reordrr the "balance of our exposures". But the rreality is that the person reached 12 times actually remains present in the room and watches all or part of our commercial, maybe 5-6 times while our average targeted person, who we think is exposed at a 6 sfrequency is actually "reached" a mere 3 times. In short, our audience data, which has little to do with advertising exposure or impact, may be misleading us.

  3. Dave Morgan from Simulmedia replied, July 13, 2018 at 5:34 a.m.

    Mark, no question in my mind that we're giong to see our tlevision and video ad measurements enhanced more and more with directly measured data that captures not only what the vewers are actually doing, but then what they subsequently do - purchase for example. I don't think that it will happen super fast, but it is certainly coming.

  4. Dave Morgan from Simulmedia replied, July 13, 2018 at 5:39 a.m.

    Ed, no question that , on average,ach "frequency" on TV today certainly represents something less than a fully attentive viewing of the ad, and I'm excited about the future when we enhance our ad viewing measurements with more robust data as Mark suggests. However, I do think that Erwin would still stick with him core mantra that +1 reach is always better than +1 frequency, assuming the quality of each impression is the same. That is the context of the frequency is crabgrass point - additional friequency may be more needed today than before, but sacrifincg oincremental reach to get it isn't the answer.

  5. Neil Ascher from The Midas Exchange, July 13, 2018 at 9:47 a.m.

    I miss the days when planners actually understood what you guys are talking about.  Regardless of the quality of the data, unless we really train entry level people to understand the concepts behind the numbers, I'm afraid we're just talking to ourselves.

  6. Dave Morgan from Simulmedia replied, July 13, 2018 at 10:14 a.m.

    Neil, your point is so important. The investmetns and training in TV planning have gone down so much as clients have squeezed agencies on fees, that these kinds of insights and opportunities just aren't translating into activated plans and buys. And you're further right on target that we need to find ways to bring this into the ecosystem operationally today, since otherwise we're only just talking to ourselves.

  7. Ed Papazian from Media Dynamics Inc, July 13, 2018 at 11:08 a.m.

    Having been very actively involved at what many believe was the inception of "media planning", I have seen how this function rose to prominence---for a time---so long as the agencies thought they could promote it, mainly in new business pitches. However, once they understood what the numbers were saying, namely, not to buy big primetime TV sponsorships or sports or big event specials as well as the advertisers' favorite magazines and other "sacred cows", agency management quickly lost interest---as did most of their client marketing directors. Even when the "media" and "creative" people worked in the same office building, they rearely talked and there was no effort to integrate thses functions, leaving media planners little option but to give clients what everyone knew they wanted--with minor polishing, here and there, like recommending a tad more daytime TV or adding a few pages in a particular magazine ---if the client would allow that.

    The sad truth is that this kind of thinking is still widely in force----despite a few notable exceptions I have encountered and lots of BS to the contrary. Yet this is a time when many  new media options are emerging while traditional media are adapting and becoming more flexible about how they target audiences and how they sell time and space to advertisers. Seemingly oblivious to this, many advertisers are operating as if this was 1988 not 2018 and their entrenched buying systems are not being tempered by the kind of thinking a well versed but fully integrated planner might supply---so little progress is actually being made. Will this unfortunate situation change? Near term, probably not; long term, who knows?

  8. Gerard Broussard from Pre-Meditated Media, LLC, July 13, 2018 at 12:04 p.m.

    All excellent points, especially about the erosion of the conceptual art of media planning.  It seems that when device graphs are connected with consumers, a cross platform measurement-and-reporting victory is declared.  But this technical feat is useless without full knowledge of the conceptual underpinnings on how advertising works.  The TV industry now appears to be committed to learn more how advertising works by packaging ad inventory with the goal of achieving business outcomes.  Encouraging.  Ths movement could be a catalyst for media agencies to intensify efforts to advance their own TV buying models and train personel on how best to tweak TV (and all media) investments to improve client business.  When I was overseeing research and analytics on the media agency side, more often than not the strategic planners would ask my department to run the more complex media planning software modules.  I kind of wonder if things have changed.  If not, they need to.   

  9. Dave Morgan from Simulmedia, July 13, 2018 at 2:07 p.m.

    Gerard, I am hopeful that the recent moves by the TV industry to better understand how their advertising works and, particularly, both complements and competes with digital will catalyze agencies and others in the eco-system to refocus on stronger, foundeational media planning. It needs a lot of work but I'm hopeful that it's coming.

  10. John Grono from GAP Research, July 13, 2018 at 6:15 p.m.

    I 100% agree with the comments about 'attention' to TV.   The current method is a proxy for attention.

    However, this plagues every medium.   For example, those who believe that slavishly analysing transaction data for a website or a video stream are deluded.   Simple example - I currently have 12 MediaPost tabs open as I type this.   All are crediting time.   Just this tab is in focus.   Not to mention that I have two browsers open.   In comparison, TV's proxy is probably better.   It is possible to include real-time facial recognition but I would seriousky question what that would do to the quality of the panel and their 'button-pushing'.

    Magazines and press have the same problem.   We ask about 'read or looked into' for titles - not ads.   Radio diaries ask for 'listening to at least 8 minutes per quarter hour', and electronic measurement has it's issues with carry-rate - which still don't address attention.   Out-of-Home 'fliters' the traffic/footfall data to exclude people who didn't have 'eyes on' the panel - but nothing about attention.

    Yes, we are right to call on better data around 'attention' for TV but do this in the context of all media.

  11. Ed Papazian from Media Dynamics Inc, July 13, 2018 at 6:32 p.m.

    John, the basic point I'm making is not so much that we don't have anything approaching ad viewing/reading/listening in the audience surveys utilized for media planning and buying, but that these in all cases wildly inflate the actual incidence of advertising noting and message consumption. When we become concerned that we are overexposing heavy TV viewers, for example, and accept the idea that such redundant exposures are "crabgrass", we may be deluding ourselves. In reality the extent of "over exposure" is not nearly as great as we think.

    Of course it would be desirable to control the number of times a person "sees" our message but in a practical sense we are not going to be able to measure this on a realistic basis anytime soon---if ever. Also, when a brand gets all hot and bothered about diverting "media weight" from heavy to light viewers, it should start by determining how it is selling among both segments and how this stacks up against competing brands who usually have exactly the same "problem". Often, the brand's share of market is about the same in both segments, in which case, shifting weight alters the competitive GRP balance, reducing  share of voice among heavy viewers while improving it on the lighter end of the viewing spectrum. Is this always a good idea? Sometimes, yes; sometimes, no.

  12. Jack Wakshlag from Media Strategy, Research & Analytics, July 13, 2018 at 6:37 p.m.

    What Dave is saying here is so much more than about media planning. Nielsen’s way of measuring viewing is at least transparent and we’ll known. And the market has not replaced it. What he is saying is that we have 25 years of online display advertising that has all this data but so little information. We are awash in data. It is not scarce or unavailable anymore. We don’t need more data. We need information and knowledge the algorithms can provide if we ask smart questions. We don’t.  We need the right questions. 

  13. Dave Morgan from Simulmedia replied, July 13, 2018 at 6:47 p.m.

    Ed, your points about the relative value of incremental  frequency in TV campaigns, particularly agagaint heavt viewers makes sense academically, but doesn’t hold up in the field now that we have massive data sets matching ad viewing to purchase at the person/impression level. That data makes clear that Erwin was absolutely right. When it comes to short term sales lift, the first impression is by far the most valuable. I have seen this across thousands of campaigns now. No question that many of those impressions never really landed, but the misses seem to be evenly dispersed.

  14. Dave Morgan from Simulmedia, July 13, 2018 at 6:49 p.m.

    Totally agree Jack. You said it so much more clearly than I did. Thanks!

  15. John Grono from GAP Research replied, July 13, 2018 at 7:30 p.m.

    I understand and agree Ed.   I've spent decades wrestling with this issue.

    We have audience measurement systems, which can exist at three levels.   1. Delivery 2. Consumed 3. Gained Attention.

    We're stuck at Level 2.

    Howvere, there are SO many vectors as to what attention is.   What gets my attention, but may not get you attention.   My belief is that we will not be able to measure attention with any depth and accuracy.

    Yes we can use al sorts of tools such as skin response, retinal studies, brain activity.   But that is 100% dependent on the content.   Should/will every advertiser test their ad?   Won't happen.   Should the media test their content in which the ad is placed?   Won't happen.

    At this stage I think the most likley approach will be to do bespoke research and tests using the mythical 'average ad' (by category?, by daypart?) for each medium so that strategists and planners can factor in the effect to get an estimated/likely "gained attention" number.   Maybe it is the CPMA - Cost Per Thousand Attention, whereas the existing CPM is Cost Per Thousand Consumed.

  16. Ed Papazian from Media Dynamics Inc, July 14, 2018 at 6:44 p.m.

    Dave, of course the first impression---when it is actually an "impression" or "exposure" will see the greatest lift but it's a mistake to assume that a second or third impression in a particular time frame has little value. The main thing is the time frame. Professor Jone's analysis of 78 packaged goods  brands as I recall, which was the genesis of Erwin's thesis, indicated considerable short term sales effects for homes "exposed" two or more times, but, curiously, he did not provide a breakdown at various frequency levels, so we are left in the dark, somewhat, in interpreting his findings.

    Returning to the issue at hand, If we select a week as our time frame, as Erwin did---though most brand campaigns with a particular positioning strategy play out for several years, not a week---- it makes sense to try to limit "excessive" exposures as people who have seen ad ad on day one are less likely to pay attention to it  a day or two later. However, if the time between exposures is streched out so that the ad is seen for the second time seven or eight days later, attentiveness will be higher. What Erwin was advocating was an every week schedule for every brand with a 70% reach and, if possible, a 1.0 frequency. Of course that was never a possible outcome---even 25 years ago. However, if you take Erwin's "ideal" plan, a 70% weekly reach with a 1.0 frequency and replicate this every week over the course of a year you are buying 3640 GRPs. Assume 90% reach for the year and you are hiting each consumer---on average---40 times. That's a lot of "crabgrass". Still, since the 40 expoures will take place every nine days, maybe the redundency is not as  bad as one thinks. It's all a matter of timing---just my opinion, of course.

Next story loading loading..