But did it sell Pop-Tarts? How Kellogg’s decided if its Pandora campaign was worth it

But this isn’t a campaign story focused on creative brilliance and genius execution. It’s a story about results. And when all was said and done, Pandora reported that 347,000 members of the ad campaign’s target group added Kellogg’s branded radio station to their playlists sometime during the course of the campaign. Average time spent listening clocked in at 81 minutes, and users spent a total 466,000 hours listening to the Kellogg’s station.

A lot of brands would have stopped there. It was easy to assume that Kellogg’s had achieved its campaign goals: to drive awareness and consideration towards Pop-Tarts among teens.

“But so what, right?” said Heidi Browning, svp, strategic solutions at Pandora. “What does that mean in terms of impact on actual sales?”

To answer that question, Pandora and Kellogg’s compared its first-party data, gleaned from its logged-in users, with Nielsen Catalina Solutions’ (NCS) local retailer data to identify consumers who were exposed to the campaign. It was a complicated mapping effort, mainly because it was the first time that NCS had worked with the mobile channel.

The next step was comparing the identified households to those that were not exposed to the campaign, going back as far as 12 months before Pandora started collecting data for Kellogg’s. This allowed NCS to control for factors that might have influenced purchase decisions, like media consumption and lifestyle.

“The industry as a whole buys into the fact that advertising works,” said Thalya Hamilton, vp, delivery and analytics at NCS. “The next question is: What works better? What part(s) of those campaigns are really driving that incremental impact?”

Of course, there was the question of whether this analysis was being conducted at the appropriate scale. “Was there enough of a sample?” asked Hamilton. There was a question as to whether the team would be able to use “traditional methodology” or “look in our toolkit for something a little more sophisticated.”

In the end, the crucial component for correlating campaign exposures on Pandora with offline purchases was the log-in. “People log in daily with us,” said Browning, “so we’ve got [updated] cross-device information.” She added, “We were able to measure audiences across platforms and then ultimately take that data, map it against a third-party data set, and measure the impact on offline sales.”

Results showed that the Pandora campaign had indeed been a success, inciting a seven-percent incremental sales lift. Additionally, every dollar that Kellogg’s spent on Pandora drove three dollars in revenue. The work also revealed how much of all incremental sales could be tied back specifically to the Pandora partnership.

This was by no means an easy problem to solve, and yet it represents just the first step in one of the most daunting marketing goals today: linking online actions to offline conversions. This scenario only accounts for the impact of a single stimulation channel on offline sales performance. The challenge quickly multiplies as more digital touch points and platforms are added to the mix. For instance, in addition to being exposed to audio and display ads on Pandora, a user may have watched a video ad on Hulu, clicked on a paid search ad, or been exposed to numerous other marketing moments that ultimately influenced their purchase decision.

At that point, advanced attribution becomes a necessary tool for any marketers looking for a holistic picture of performance; to know with certainty which channels and tactics actually led to or influenced a conversion. Otherwise, allocating budgets to the best-performing channels and tactics can never be more than educated guesswork.

But in the end, some vital factors popped to the top for Kellogg’s: good first-party data and the ability to map it to retailer data was a necessary starting point. “Those were two key pieces to this puzzle,” said Browning.