|Source: Parade, based on Information Resources Inc. analysis Note: Return on investment is incremental dollar sales of product for each dollar of media spending.|
At least that's what marketing-mix modeling the publication commissioned from Information Resources Inc. appears to show, in a move that increases the pressure on other media players to similarly prove their worth.
IRI's study of four package-goods brands advertised in the magazine in 2006 found the ads lifted sales an average of 20% the following week. Results ranged from 10% to 31% for the brands, which included Procter & Gamble Co.'s Prilosec OTC, Kraft Foods' DiGiorno pizza, Hefty OneZip bags and Campbell's cream of mushroom soup.
Even some who believe strongly in the power of advertising have trouble believing those numbers, which substantially exceed typical modeling results that show, at best, single-digit sales bumps for all media advertising combined over much longer periods.
Mike Hess, director-global research and consumer insights for OMD and a longtime marketing-mix analyst, was initially skeptical but after discussing the methodology with IRI, believes it's sound.
Other media executives also appear intrigued, or at least impressed with Parade's ingenuity.
"I find the idea [of media vendors using marketing-mix models] to be very, very interesting and compelling," said David Verklin, CEO of Carat North America.
"I applaud them for finding a way to prove their value to clients," said Brenda White, VP-director of print activation for Starcom. A strong believer in print, she wasn't surprised by the findings. She said using IRI, a firm many package-goods marketers use themselves, adds credibility, and said Parade's move pressures other media vendors to produce similar third-party ROI validation.
Marketing-mix modeling, which employs complex correlation-regression calculations, has become increasingly popular with marketers. But neither Mr. Hess nor Todd Brant, VP-Analytical Insights Group at IRI who ran the study, knows of any other media company having done a similar one.
Many package-goods clients were reporting strong sales lifts from Parade ads, claims Jim Hackett, senior VP-marketing and research, but wouldn't let their numbers or endorsements be used in pitches. So he said the magazine hired IRI on its own to do the modeling on ads that already had appeared.
The brands were selected because they had other significant spending in TV, print and in-store merchandising, said Mr. Brant. He acknowledges the results are surprisingly high, more akin to what he's used to seeing from trade promotion than advertising.
Mr. Brant said most other individual magazines or TV programs don't have the reach to produce statistically significant results. Mr. Hackett said Parade could use statistical modeling because of its reach -- 32 million homes and 70 million readers. It chose to look at following-week sales because, unlike most magazines, most of its readership takes place on one day -- Sunday.
All true, Mr. Hess said, but much of the eye-popping impact IRI found likely comes from looking at only one week vs. the quarterly or annual time frame in most analyses. He doubts Parade's result would remain so high if the same ad ran for four, or even two, consecutive weeks, adding that sales bumps so high are likely to have moved up some purchases that would have occurred in subsequent weeks without advertising.
But even spreading that one-week 31% bump IRI found for Campbell's cream of mushroom over 52 weeks equates to a sales increase of more than a half percentage point for a full year -- still good for a single ad in a single publication.
The model indicated TV ads collectively boosted sales another 49% and other magazine ads another 16%.