It sounds like a joke, but it's a true story: Mattel's ghoulish Monster High dolls were eating away at Barbie. When the toymaker measured the effects of paid search marketing for its Monster High brand in test markets over a 13-week period, it saw a 3% boost in sales for the creepy dolls which go by names like "Draculaura" and "Howleen Wolf."
Yet, while paid search and other digital ads increased incremental sales volume for Monster High dolls, "We also found it cannibalized Barbie and other doll volume," said Bill Bean, VP of shopper insights and competitive intelligence at Mattel.
Mr. Bean, a CPG veteran who is helping bring shopper insights and measurement techniques more common in the packaged goods industry to Mattel, suggested he would have never known that the Monster High sales was crimping sales of its flagship doll brand were it not for a system that measured the effects of paid search, including ad volume, targeting and timing, in test markets matched to control markets.
"If we considered the cannibalization of the larger brands as well as the incremental volume on Monster High, the net effect of the marketing stimulus was negative," said Mr. Bean. In a less-sophisticated test that didn't isolate and track multiple data streams while controlling for other factors, he added, "You never would have seen that incremental Monster High sales came at the expense of your major brand."
For about two years, Mattel has worked with Applied Predictive Technologies, or APT, to develop testing based on the analytics firm's Test and Learn platform for measuring marketing, merchandising, operations and capital investments. The partners determined which test markets would provide the most representative samples, for instance. They tracked weekly sales in those markets while running campaigns promoting the Monster High brand.
APT was acquired in April by MasterCard for $600 million, giving it access to aggregated and anonymized data from its new parent company showing how people shop across multiple brands. For clients including Coca-Cola, P&G and Hershey's, APT uses store-level data along with demographic information, weather and economic data, even unemployment and gas price data, to understand the true impact of marketing taking into consideration real world factors. Mattel itself looks at sales, price and promotion data from more than 7,000 stores including Walmart, Toys "R" Us and Target locations.
According to Jonathan Marek, senior VP of APT, "It helps them figure out things like, Are there different characteristics of the stores? … Do higher income stores or lower income stores perform differently?"
Using the data, Mattel was able to choose markets that would respond best to paid search ads overall, while reducing cannibalization of Barbie and other brands. Now the company is using the test results data to help adjust its approach to paid search keywords. The research has informed plans for other brands, too, including Fisher Price, said Mr. Bean.
Campaign measurement and analytics isn't entirely new at Mattel, which had a division dedicated to traditional modeling and analytics such as marketing-mix modeling before Mr. Bean joined the company in August 2014, following more than a decade in research and insights positions at PepsiCO, Miller Brewing Company and Colgate Palmolive.
What he has introduced to Mattel, however, is more about direct measurement than modeling; in other words, APT is helping Mattel track what's happening in the real world rather than using models intended to represent what is expected to happen.
"We're measuring on the ground in markets the effects of our marketing efforts using controlled tests in stores," he said. "We used to say at Pepsi that nothing happens until something is sold."
He went on to conclude, "The standard we're trying to establish is that, and it's been my personal opinion for years, the ultimate measure of any marketing stimulus in a market is sales."