And so the many years of "experiments" in digital media have produced few results that can be compared alongside those of better-funded media such as TV.
"The mix model was created in an era when everything was big," said Karna Crawford, director-media and interactive at Coca-Cola Co., at the Digital Media Measurement and Pricing Conference in January. "We're now in a world where small, incremental things tied together is what works best."
Coca-Cola often measures the effectiveness of digital campaigns against objectives other than sales, such as impact on brand-loyalty surveys. But, Ms. Crawford conceded, Coca-Cola's "marketing science" staff still wants to see the model impact.
"What happens is the companies that use marketing-mix modeling -- and the package-goods companies in particular -- treat it almost as gospel," said Rex Briggs, CEO of Marketing Evolution.
As a general rule, a digital campaign needs to reach at least 20% of a brand's target consumers to be measured by a marketing-mix model, Mr. Briggs said. That usually requires spending of about $2 million, he said, a level few digital campaigns reach -- in many cases because that kind of money requires prior success measured by the models.
Isolating campaign returns
But Doug Brooks, VP of Aegis Group's Marketing Management Analytics, one of the original modeling firms, said it's possible to get a reading on campaigns in the mid-six-figure range. Mr. Brooks points to an apparel marketer that found its Facebook campaign delivered sales gains of 7% to 10% and returns of $5 for every $1 invested. While the marketer doesn't use TV and isn't as huge a spender as some package-goods brands, MMA was able to separate the impact of Facebook and its other digital efforts from its print campaign.
One way to get measurable results from smaller campaigns is to concentrate digital spending in shorter periods and "build measurable variation" into the campaign by changing media weights from week to week or concentrating spending on certain geographic areas, Mr. Brooks said. Dixie tableware, a unit of Koch Industries' Georgia-Pacific, has been learning to live without marketing-mix results from its six-figure sponsorship of the MommyCast podcast, now entering its third year. Indeed, one objective in sticking with the program so long is to develop better ways to measure its impact, said Dixie Marketing Director Erik Sjogren.
Dixie has used surveys to find that category usage among podcast listeners is 19 points higher than average, at 54%, and brand penetration is 17 points higher than average, at 42%, Mr. Sjogren said. A fifth of listeners reported an increase in Dixie usage vs. a year ago, and 53% reported recommending Dixie to someone else in the past six months.