The Perils of Econometrics

Mar 8, 2004  •  Post A Comment

A group of about 300 media planners last week heard that there is a need for their input somewhere in the heart of the sometimes mysterious process of econometric modeling.

Econometric modeling, one of this year’s buzzwords in media circles, is designed to use sophisticated data processing to determine which types of marketing and media spending produce sales results.

But a panel assembled at a conference sponsored by TelevisionWeek at the Grand Hyatt Hotel in New York said econometric modeling-like the much-ballyhooed optimizers, designed during the ’90s to mathematically determine the best television mix-is no panacea. Indeed, questions abound regarding what types of marketing can be put into the model, whether results other than sales can be predicted and whether the modeling should be done by a media buying company or an outside consultant.

To a degree, how some econometric modeling is done is kept secret on purpose. “It’s proprietary. I can’t talk about it,” joked one of the panelists, Randolph Stone, CEO of Marketing Management Analytics.

More seriously, Mr. Stone said, modeling is designed to find out what works to increase the odds of success. He said econometric modeling started with packaged goods companies with gobs of supermarket scanner sales data to work with, and now “the discipline is migrating into other categories very well.”

He added that in addition to deciding how a brand spends its marketing resources, econometric modeling can be used to decide how much return on investment a marketer gets by supporting each of its brands before allocating marketing funds.

Krishnakumar Davey, general manager, Insight Partners, added that while modeling formerly was used primarily as a report card on how well marketing spending worked, it’s increasingly being used to make decisions. “It makes you smarter and helps make better media decisions,” he said.

Mr. Davey stressed that for the model to work, it needs detailed input on what works for specific brands.

That’s when the media planner should have input. “Meet the people who are modeling in person,” he told the audience. “You know the data.”

Mary-Ellen Vincent, senior VP, director of research, insights and accountability at MediaVest USA, added that the models are “not just an algorithmic formula. A lot of assumptions go in.”

While many models may do a better job of evaluating media with well-established measurements, especially television, Mr. Stone noted that his company has measured the marketing impact of more obscure marketing tactics, such as handing out items at a ballgame. He also said one movie company sought to qualify the public relations value of a star who was in the news for a morals charge as opposed to a star who simply made the rounds of the talk shows.

But Mr. Davey disputed the notion that the models handle those sorts of things well. “I’m not a big fan of that,” he said. Gregg Liebman, senior VP and director of strategic resources at Zenith Media, added that it’s difficult to model media that a client doesn’t use. But if you test a new medium, you find out how effective it is. “That’s the benefit of econometric modeling,” he said.

With clients looking to econometric modeling as a way to validate the effectiveness of their marketing spending, some question whether the modeling should be done at a media agency or by an independent company.

Mr. Stone said it could look like “the fox guarding the henhouse” because a media agency could find that the best idea is to buy 30-second spots, which gives the agency its best profit margins. But the data can’t be manipulated, he said. “You can’t make up these inputs,” Mr. Stone said. “The answer is what the answer is.”

Ms. Vincent said that now that the agencies understand modeling, they don’t actually have to do the calculations themselves. “We don’t have to be the programmer, but a system analyst,” she said. “We’re happy to do either.”

Mr. Davey insisted that doing it within a media agency is best: “There are a lot of nuances to the data that I didn’t understand from the outside.”

Inside there’s a better chance to work together to drive a client’s business, he added.

More work has to be done to focus econometric modeling on long-term results. “It’s tougher and you need more data,” Mr. Stone said. One impediment is that marketers don’t want to know what will happen in 2007. “I’m more concerned about how I’m going to make it till June,” he said, echoing client concerns.

Mr. Davey downplayed the distinction. “If it doesn’t work in the short term, it won’t work in the long term,” he said. “The long term will take care of itself.”

But Ms. Vincent disagreed and cited the quote that “in the long term, we’ll all be dead.”

A second panel focused on the phenomenon of communications channel planning. The panelists were Marston Allen, senior VP and director of communications architecture at Universal McCann; Enza Chiodi, executive VP and director of planning at PHD; and Carla Loffredo, senior partner and strategic planning director of MindShare. The moderator was media consultant Erwin Ephron.

Integrated channel communications planning looks at marketing messages from the consumer’s perspective, Mr. Allen said, instead of from that of the media or the marketer.

One company, Unilever, has put ICCP at the hub of its marketing planning, but other companies have not fully embraced it yet.

While ICCP can be threatening to individual marketing disciplines and media choices, it doesn’t mean that all marketing services will be integrated into a single company, Ms. Loffredo said. “You need a team of specialists and to respect their ability to contribute to our brand.”