• by Joe Mandese  @mp_joemandese, April 19, 2011

What would you do if you wanted to create a media agency designed not for the traditional world of reach & frequency media planning and buying, but an era in which changes in technology and consumer behavior are reshaping faster than anyone can plan? That's what the team inside Interpublic's Universal McCann unit began doing a couple of years ago, and the result is quietly being rolled out now - initially to some of its biggest and most progressive clients - account like Johnson & Johnson, Schwab and Wells Fargo - with others likely to follow over the next year. The new model is based less on conventional media planning and buying, and more on things like Wall Street's and the travel industry's sophisticated "yield" optimization systems that utilize powerful models tied to some new, exclusive Nielsen data, including Nielsen's joint-venture panel with Facebook.

The new approach, which agency insiders have dubbed "UM 3.0," has quietly been integrated over the past few months with some of the agency's leading clients, and will, they say, greatly differentiate it from other leading media services agencies, including sister Interpublic shop Initiative.

The cornerstone of the makeover is an extremely sophisticated modeling system that incorporates many of the elements of the kind of econometric modeling systems that have helped transform the way marketers measure the ROI of their marketing mix over the past couple of decades, but instead of being purely regression based, it has been adapted to work on-on-the-fly, with more predictive modeling built in. Equally significantly, the system incorporates much more discrete media channels than traditional marketing modeling systems look at, with the ability to get as granular as the effects of a specific TV show or magazine buy on a media plan, and the contribution it makes to the marketing goals.

UM executives don't have a fancy new term to describe the new media operating system, but generally refer to it as their "data ecosystem."

"The problem with relying only on econometric modeling is that is that you usually get stuck at the channel level," explains Hari Abhyankar, senior vice president-business insights & analytics at UM, who helped design and build UM's new system in conjunction with Nielsen. "Those models are good at inferring relationships between advertising and sales but they have some drawbacks."

The main drawbacks, he says, are that regression models are generally looking back at past results and inferring future consumer behavior based on what worked in the past. Another major drawback is that those models were too broad to look at the discrete effects of specific media buys. The new system developed between UM and Nielsen, he says, enables UM's team to "observe rather than infer" in real-time, and to analyze the impact of extremely discrete media buys - even the daily contribution that TV advertising budgets are having on a marketer's sales for "one show vs. another."

The new system, which includes a two-year exclusive partnership with Nielsen, was alluded to two weeks ago by Mediabrands CFO Jeff Lupinacci during an Interpublic briefing with Wall Street security analysts. While he did not elaborate on who it worked, Lupinacci said the system would enable Interpublic's agencies to "align" their compensation with their client's business outcomes.

While the new Nielsen data will ultimately be shared with other Interpublic agencies, including Initiative, UM executives said the model was developed exclusively for them, and UM CEO Jacki Kelley said the compensation model was developed for UM by big management consulting firm McKinsey & Co., which typically works directly with Fortune 500 marketers to structure how they work with their ad agencies and other service providers.

Kelley tells MediaDailyNews that the new "pay-for-performance" compensation model also is being rolled out on a client-by-client basis, in conjunction with the new data and modeling systems, which will enable the agency to measure its real-time performance on client's brands, based on specific KPIs, or "key performance indicators."

She said the new compensation model was developed by McKinsey after looking at a wide range of other industries, including the oil drilling business, but is unique to UM.

She says the new compensation model only works because UM has reliable data to predict and measure its performance against.

"If you have the data and the structure to tie it to much more meaningful values, we should be compensated on it," she explains, adding that in many cases, UM will be hedging its compensation based on future outcomes for its clients, which is radically different from the standard compensation models utilized by Madison Avenue media shops, which generally are tied to some form of cost-plus fee arrangements based on the number of full-time employees allocated to their clients' businesses.

"We will hedge based on our ability to influence the business outcome," she confirms, adding that the new compensation model may not be right for all clients, and that some clients may want to minimize the degree of financial risk UM takes on their business, but that the agency is prepared to take those risks because it is so confident in the quality of the data, its modeling systems, and the ability to reliably project its clients' business outcomes.

"It's definitely a mash-up," she says. "It's like everything in life. It's about balance, and not all clients will want to work this way. It's not for every brand."

UM's Abhyankar says the data ecosystem is still a work in progress, and while new Nielsen databases and data integration strategies are the core of it, he says UM is bolting on other data and its own proprietary insights to fill in the gaps.

Without going into specifics, he says the system is the most highly integrated approach to Nielsen's disparate array of both media measurement and consumer purchase tracking data to date. Among other things, he says it includes Nielsen's traditional media measurement panels, its consumer purchasing data panels, its "single-source" database with Catalina Marketing, and its new, humongous social media tracking panel with Facebook, a company that Interpublic owns a minority stake in.

Nielsen and Facebook have said very little to date about their joint-venture to turn some of Facebook's half billion users into a panel for measuring marketing and media. The deal, announced more than a year ago, simply said the two companies would collaborate.

During a presentation at MediaPost's 2010 Outfront Conference, Nielsen Online chief John Burbank said the panel was still in its seminal stages, but that it potentially could be as big as all of Facebook's user base, though it would depend on who actually opted in to the Nielsen joint venture.

Asked what he anticipated that opt-in rate ultimately would be, Burbank demurred, noting that it would remain a "trade secret for a very, very long time."

Along with the new data, panels and modeling systems, UM executives say they've begun to restructure their organization to emphasize the kind of talent and experience that is more akin to the way other industries like Wall Street and travel operate, than traditional media services shops.

UM's Kelley says the agency has been recruiting executives who are, "former consultants, former peer point managers, and data jockeys," as well as a number of people from "media owners."

While the bulk of the agency's organization will continue to be comprised of media planners and buyers, she says they are developing new skills and are evolving to make the traditional part of their business "smarter and more automated" and to also have hteir results measured against specific client performance objectives that may have little to do with traditional media plans and measures like, "CPMs."

Kelley says the transformation is consistent with the vision laid out by former UM chief, Matt Seiler, who is now CEO of Mediabrands, and the motto he coined when he joined the organization from Omnicom: "Curious Minds For Surprising Results."

But with better and more predictive data, planning systems and expert talent, she says, those results increasingly will become less of a surprise.