Neszed-Mobile-header-logo
Thursday, October 9, 2025
Newszed-Header-Logo
HomeFood & DrinkHow Kellanova uses AI to predict creative performance and drive KPIs

How Kellanova uses AI to predict creative performance and drive KPIs

This audio is auto-generated. Please let us know if you have feedback.

Kellanova, in partnership with data firm Vidmob and MMA Global, has found it can more quickly connect the dots between campaign creative and business outcomes by leveraging artificial intelligence to power predictive impact scoring, which crunches past performance data to pinpoint creative criteria for driving specific KPIs.

In an analysis of more than 443 Kellanova creative assets, the marketer of snack brands Pringles, Cheez-It and Pop-Tarts found that predictive impact scoring can forecast three-second view-through rates (VTR) with 83% accuracy, improve performance by more than two times and contribute to an 11% increase in ROI. 

“Predictive impact scores are like creative fingerprints – unique to our brand, actionable and incredibly powerful for driving both brand distinctiveness and performance,” said Charisse Hughes, senior vice president and chief growth officer at Kellanova, in a statement. “At scale, these go beyond growth and drive transformation, and with AI we can do this with increased speed and accuracy.”

The study, conducted between March 2024 and March 2025, analyzed assets that ran across Meta platforms in the U.S using Vidmob’s Aperture technology. The analysis was used to pinpoint 19 company-wide and 11 category-specific scoring criteria that are now being used to inform Kellanova advertising creative moving forward.

“Creative has always been a little bit subjective, a little bit art and science. But now, with the ability to process large sets of data at scale and be able to build more advanced analytics and modeling on top of that, we can get to more precision and more predictiveness of business outcomes,” said Nicole Vinson, vice president of digital, media and omni-shopper experience at Kellanova.

Inside the process

Kellanova’s latest predictive scoring work began a few years ago when the company implemented Vidmob’s fit-for-platform best practices and scoring capabilities — a first step in adhering to the basics required by each digital platform to ensure creative performance. When MMA Global approached Kellanova about testing predictive scoring, the company was prepared to connect scoring with specific KPIs.

“The advantage that you have as a marketer is that you can unlock specific insights that are specific to your brand and your distinctive brand assets. That’s your own secret sauce. That’s your competitive advantage versus my next competitor that’s using a similar tool,” Vinson said.

The study looked at high-production value, branded creative assets from 10 brands in Kellanova’s North America business across Meta platforms, analyzing more than several billion impressions over the year. The use of Vidmob’s AI tool allowed Kellanova to winnow down 20,000 creative decisions — like including positive emotion in the first several seconds of an ad, having audio or text overlays, the presence of brand logos and characters or consumption cues, for example – into 19 variables.

“As you think about the power of this, it’s not to create a paint-by-numbers solution for our creatives and our content production partners,” Vinson explained. “It’s to help them to understand what are those variables that have the highest predictability of improving performance so we can then start to inform future creative decisions, operationalize that and scale that across our business.”

After analyzing its creative on Meta platforms, Kellanova has run the exercise using the same framework and methodology across all digital media platforms and rescoring creative against new criteria to acknowledge, for example, what works better on Facebook compared to TikTok. The recursive process is allowing Kellanova to drill down into what works for specific brands and audiences.

Source link

RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Most Popular

Recent Comments