In a conference room at the world’s biggest media-buying agency last week, the research guru was showing off an expensive new high-tech toy that will eventually change the economics of television.
There was just one problem with the “optimizer” developed by West Hollywood-based Western International Media: It had a few bugs.
“Something obviously strange has happened to the system today, which is very frightening to me,” said Cheryl Idell, Western’s president of strategic planning and research, as the machine spat out a buying plan that ignored many obvious choices for the best reach with a selected demographic.
Embarrassing glitches aside, Western’s optimizer usually works. Already, optimizers have become the dominant method for buying TV time in parts of Europe, and it is now being adopted by big advertising and media-buying agencies across the United States.
Of course, the domestic TV landscape is tremendously more complex than in these countries, and as was made obvious last week, American optimizers have plenty of bugs to work out. There are also lots of additional variables yet to be included in the systems.
Theoretically, an optimizer is supposed to tell advertisers the most efficient way of buying national television time. You just program in the amount of money you plan to spend in a given week, the type of audience you’re trying to reach women 18 to 24, say, or adults 25 to 54 with annual household incomes of $75,000 or more and it tells you which TV shows to buy time on to reach the greatest number of eyeballs fitting your description.
The people who sell advertising time on TV are not necessarily thrilled about the innovation.
“We’re going to be selling to a computer,” grumbled one sales rep to another during the demonstration.
“It was bound to happen sooner or later,” the guy sitting next to him replied.
Actually, as Idell was quick to point out, the computer won’t really be making the decisions at Western or at any other agency that uses optimizers. For one thing, it almost never spits out the same buying plan twice. The number of choices among syndicated, cable and network TV is so great that there are many different permutations of what constitutes an efficient media buy.
For another, there are some variables the optimizer doesn’t factor in. Because time on NBC’s Thursday-night lineup is so expensive, for example, the optimizer rarely advises buying it. But some advertisers really need to be there; movie studios with films opening on Friday will pay a premium for big TV exposure on Thursday night. The optimizer doesn’t know that, but the human buyer does.
Buyers and planners typically run a set of variables three to five times, look for patterns, and use their own experience and knowledge to come up with a final buying plan, Idell said.
The use of optimizers in the United States will likely revolutionize the TV industry. Even though the final decisions will still be made by humans, the optimizers point out choices that the humans would likely never have considered and eventually, preconceived notions about the best way to buy television time will give way to pure mathematics.
What this does is level the playing field between relatively obscure cable channels that reach a small but desirable demographic and don’t charge much for advertising time, and the big broadcast networks that reach a mass audience and make you pay through the nose.
During last week’s demonstration, representatives from Fox and NBC grumbled that their programs weren’t showing up on trial runs by the optimizer. Part of that may have been because of the bugs, but in fact, optimizers do tend to shift the balance of buying recommendations away from expensive network TV and toward cable.
Idell asked the optimizer to find the best possible buy to reach men 25 to 54, spending $3 million, during a given week. The machine concluded that only 3 percent of the money should be devoted to primetime TV, the showcase daypart for networks.
“Unless we are explicitly saying it should buy a certain amount of primetime, it tends not to buy very much, because it’s looking for the best possible reach for the money and you can’t usually find that on prime,” Idell said.
How this will ultimately affect the economics of television is anybody’s guess but there’s little question that it will.
“We’re not going to show up on any of these (optimizers) if it’s just about cheapness,” said Keith Turner, president of sales and marketing for NBC, who attended last week’s presentation. “If you talk about program ratings, if you talk about attentiveness (the amount of attention paid to a show by its audience), that’s a whole different thing.”
Attentiveness and assorted other audience variables may eventually be added to Western’s optimizer, which has been dubbed “WestOpt” by the company. But the current system is already being used as a tool for buying TV time. It was unveiled last summer when advertisers began buying time for the 1998 upfront market, with the optimizer using viewer information that came from the November 1997 sweeps by Nielsen Media Research.
Once this year’s November sweeps figures are available, they will be programmed into WestOpt, which Western plans to update with new Nielsen information at least once every quarter.
News Editor Dan Turner writes a weekly column on marketing for the Los Angeles Business Journal.
