Two seemingly unrelated pieces of news this morning made me wonder whether it might not be time for media planners and buyers to start negotiating some clauses in their union contracts to protect them from displacement due to AI. Then I remembered media planners and buyers are not part of any unions, although maybe they should be.
So you probably already figured out that the first piece of news was that the Hollywood unions have reached a tentative deal ending one of their longest strikes ever. And while the terms of that deal will not be disclosed until the agreement is finalized, it is expected that it provides some protection from future uses of AI displacing human writers and actors.
While that would be worthy of a discussion for planners and buyers in its own right, it was the second piece of news -- the release of a World Federation of Advertisers study finding that media teams are playing backseat role in setting the AI strategies of big marketing organization -- made me think it was time for planners and buyers to have this conversation.
advertisement
advertisement
On the surface, the WFA study might seem like good news for agency media executives, because most of the respondents said they were mainly focused on the role AI can play in creativity, content creation, customer experience, and things like that.
The reason I'm not so sure, is that I believe the role of AI in media planning and buying is already baked in, and sadly, the media teams are only a secondary voice in setting the strategy of it going forward.
How is AI already baked in for media? Well, think about it. While the ad industry has talked about creative use cases for years -- ie. "dynamic creative optimization," "personalization," etc. for years -- most of the actual earliest uses of machine automation have been in the media planning and buying process: ie. programmatic.
And while yesterday's programmatic isn't today's or tomorrow's AI-powered media services, it doesn't take a great mental leap to see how advances in machine learning and AI could ultimately displace -- not augment -- what human planners and buyers have long done, especially decision-making.
Some of the earliest uses of machine learning in advertising have in fact been about media decision-making and or allocation.
Take marketing-mix models, which processed the Big Data of their day to explain which parts of the media mix were performing best. It's still used today, and various iterations of real-time attribution that have grown from it.
In fact, one of the first examples of machines processing data that previously were analyzed by people was the introduction of audience reach optimizers in the early 1990s.
When I first started writing about optimizers, I asked the late media-planning guru Erwin Ephron why agencies needed powerful computers to analyze what people used to do manually.
After thinking about it for a while, Ephron told me to write some numbers down on a piece of paper: 1.125 followed by 12 zeroes.
It was the first time in my life I had written the number quadrillion, but Ephron told me that was the number of permutations that a planner would need to consider to process all of the possible options in the national TV programming marketplace of broadcast and cable networks and syndication.
And that was 30 years ago.
Over the next several decades the progression of machine automation of media planning and buying has accelerated, but has largely still been people controlling what machines processed, or at least making decisions on what they outputted.
I don't know exactly when machines will be able to do the second parts as good as or better than humans, but I know -- just like Erwin Eprhon did -- that it's just a question of math. Not if, but when and how much.
So here's my advice to any planners or buyers interested in being around for the long haul. Get a copy of the final Hollywood agreements, read the parts related to the use of AI and how it can or cannot explicitly displace human labor, and think about framing something similar for media services.
I'm not going to suggest what you do with it, because Madison Avenue isn't exactly a collective of union shops -- but at least you'll know what the baseline is for a comparable industry going forward.
Interesting piece, Joe.
When BBDO and a number of ad agencies introduced the concept of computerized "media selection" in the early- to mid -1960s one of the reasons given was that the computers would be able to analyze all of the countless media possibilities that the planners would otherwise have to review each time a plan was called for. And that was at a time when there was much less media and very much less audience data available to say nothing of qualitative information concerning ad exposure. Of course the real motivation was to conjure up something to entice might be new business clients besides "creative" which was always a chancy sell. So when BBDO announced its "linear Programming" system the other major shops all rushed to develop their own counterparts.
While we solved the missing data problem enough to feed the system, the end result was that the stupid computers didn't like what most clients liked----prime time TV, TV sports, TV Golf tournaments, certain must buy magazines, etc. Instead, they kept recommending daytime TV, ROS radio, outdor billboards, etc. because of the vast CPM differentials. Result: the models were all quietly retired.
As for the claim that the computers saved the planners the work of evaluting all of the possible media options they would have to otherwise investigate that was a totally bogus claim. Media planners ddn't waste their time pondering the merits of thousands of radio stations or out-of-home billboard venues and other postings or every newspaper or magazine in the country or every TV station---especially when they know that their client would not tolerate mention of same. And I suspect that this is true to a large extent today. Sure, we have programmatic buying for online media---where the computers supposedly evaluate all possible buys---but all too often the advertiser's ads go to bots, or to bad places or to sites where nobody is paying attention or where the ads are not very visible---and half of the ad dollars go to middlemen not the websites themselves.
Yep, you guessed it. I'm not so sure that AI is going to be a boon to media planners---or buyers. Actually, I see it as representing an excuse to automate those functions and get rid of the expense of having human planners and buyers---and, ultimately, sellers. And I expect that top management will fall for the inevitable sales pitch---that AI can do it better, faster and cheaper---until the inevitable debacle that will result hits. So maybe you are right---planners and buyers---you better set up a union.