Marketing-mix modelers shoot for real time as some users tap brakes

By Published on .

Delivering real-time sales data and TV advertising information has gotten so efficient thanks to new tools, marketers are now worried the analysis of that data may intensify short-term thinking. Some are even raising questions as to who should be trusted to see and interpret it.

After seeing strong double-digit increases in the past year in sales of marketing-mix modeling to ROI-conscious marketers, marketing analytics firms are aiming to go one better by slicing and dicing the data in near real time and delivering it in simplified dashboard-style formats.

But the new systems may be outpacing the capability of research firms to deliver TV advertising and retail point-of-sales data. And efforts to make the data and analysis more accessible are raising questions about who should dissect that information and just how beneficial getting ROI analysis done even faster will be.

"Marketing-mix models are wonderful for a point in time, but they're somewhat backward looking," said Ed See, exec VP-product development and consulting for Aegis Group's Marketing Management Analytics. "What's been missing is combining a mix model with forecasting and actual live data feeds. It gives you the power to watch your actual results vs. your forecasts, identify the causal drivers and find what to do to get back on track and optimize spending."

Not all the needs for mid-course corrections are driven by how the marketing campaign is working. "I might have built my plan on hitting a volume target, but now I'm halfway through the year and I get a budget cut that comes down from corporate and I need to re-optimize to hit a profit goal," said John Nardone, exec VP-product development and marketing for MMA. "As a former brand manager at Procter & Gamble, I know that's a pretty typical scenario."


Another emerging use for the new continuously updated tools is portfolio management, or monitoring how marketing plans for one brand could affect a company's other brands in the same category, Mr. Nardone said. "It's a very active what-if tool," he said. "And that's something that we really feel marketing managers have needed for some time."

Independent marketing-analytics shop ImmediateFX began marketing systems last year that can deliver marketing-mix modeling continuously for forecasting and trouble shooting-potentially helping marketers make on-the-fly adjustments to marketing plans.

MMA launched Avista, with similar capabilities, earlier this year; sibling media planning and buying shop Carat also uses the technology with clients. Other research industry heavyweights-including VNU in cooperation with IBM and Information Resources Inc. in cooperation with Acxiom-are also developing enhanced ability to provide continuously updated marketing-mix analytics.

Research firms are mum about which clients are using the models and how. But Procter & Gamble Co., long a client of MMA and more recently a client of Carat, is believed to be among companies trying Avista. Other package-goods and drug players, including Clorox Co., Gillette Co. and Energizer Holdings, also have begun using the continuous marketing-mix tools, according to executives familiar with the industry.

Having the capability to evaluate ROI and start the process of making adjustments immediately seems like a dream come true for analytically inclined marketers, but several impediments make it harder to do than say.

Marketing-mix models are complex tools that use econometric analysis of data sources that generally arrive with varying speeds and accuracy. Among factors that may need to be analyzed include a brand's own data on sales and marketing spending and how it was actually executed in the marketplace, the same information for all competitors, plus such general market-moving variables as consumer income, unemployment-even the weather. But some data streams are more reliable, and available, than others.

Syndicated retail data hasn't included Wal-Mart Stores data now for years, but some marketers who are "category captains" for Wal-Mart have complete data from the chain for all competitors, or use their own shipment data and panel data to fill gaps, Mr. Nardone said.

IRI recently invested $300 million in a new computing network that, among other things, gives such retailers as Albertsons, A&P and Walgreens a real-time point-of-sale tracking system similar in scope and performance to Wal-Mart's famed Retail Link. That can help improve data upon which models are based. But Michael Heberle, VP-analytic product development, notes that the brand sales and pricing data that go into models already came faster than some other pieces of the data set-notably TV gross rating points, which still have a four-week time lag.

John Hastings, senior VP-VNU Global Business Intelligence, noted that overnight Nielsen ratings numbers can help provide data faster. And Mike Noonan, VP-modeling and analytics for ACNielsen, added that VNU owning both Nielsen Media Research and analytics provider ACNielsen gives it an edge in "putting all the data together."

"Nobody's going to sit there and assess whether a new campaign that went on air last week is having any effect this week," Mr. Noonan said, adding that different clients have different standards for how long to wait to see impact from ads, but all generally give the ads more time to work than retail promotion events.

"Real time" is still more dream than reality for most of the data streams needed to analyze returns on marketing investments, Mr. Heberle acknowledges, with a two-to-three week lag common for many of the elements. But he said once links between IRI's MarketKnowledge platform and its Mix Drivers and Drivers on Demand products are fully operational, store-level sales, distribution and price information should only have a two-day lag.

who sees what?

Dashboard-style software that can let relative analytics novices (such as brand managers) monitor their marketing-mix performance are common parts of these real-time tools. But accessibility can also make the data prone to misinterpretation. As a result, the players in the emerging space provide client service as part of the package.

"Companies often lack the staff, and the question also becomes whether you want the people who are working with the marketers everyday to evaluate their programs," said Gregg Ambach, VP-marketing analytics for ImmediateFX.

The model makers also acknowledge that their systems can fuel short-term focus that favors trade promotion over advertising. But by maintaining quarterly service calls by analysts and offering tools that benchmark long-term impacts of marketing decisions, Mr. See said, "we will be helping people not drift further down the trade promotion addiction."

Another question is how much information is too much. "The reality is that most manufacturers can't [change marketing programs] next week or even two weeks out very much, because things are planned months in advance," Mr. Ambach said. "So you start to wonder if there's a point at which it sounds good, but having more data can be counterproductive."

Most Popular
In this article: