Hershey Turns Marketing Mix Modeling into Monthly Process With AI Agents
Hershey is automating marketing mix modeling-a statistical technique that measures how media spending drives sales-by partnering with analytics platforms Mutinex and Tracer. The move compresses what has historically been a slow, backward-looking analysis into a monthly cycle.
Previously, Hershey ran marketing mix modeling three times a year for about five brands. Results arrived months after the data was collected. "We were getting the full read of 2024 data midway through 2025, while we were planning for 2026," said Vinny Rinaldi, vice president of media and marketing technology at Hershey.
The new system processes data in as little as three weeks. Hershey now measures its entire brand portfolio monthly instead of quarterly, meaning the company can adjust media and trade spend decisions 12 times a year instead of three.
How the system works
Mutinex, built on Claude and Gemini, functions as a multi-agent system where each AI agent specializes in a specific domain-marketing econometrics, competitive pricing theory, or model diagnostics. Tracer cleans and standardizes fragmented data across Hershey's marketing and retail systems so the models run faster and more reliably.
Sarah Martinez, chief commercial officer at Tracer, said the bottleneck isn't artificial intelligence. "Most companies don't have an AI problem. They have a data readiness problem."
The budget justification problem
AI for Marketing measurement addresses a long-standing credibility issue. CMOs struggle to justify media budgets as investments rather than costs because attribution methods have been unreliable and slow.
Lou Paskalis, market advisor at Mutinex, said the problem runs deep. "Marketing is not viewed as credible when it comes to investment. A lot of that has to do with skepticism around how attribution has been done historically."
Faster, more frequent analysis should change that calculation. Hershey expects to increase revenue attributable to media by 4% to 5% once the system fully rolls out. With two billion dollars split between media and trade marketing annually, monthly optimization decisions could significantly affect allocation efficiency.
Data Analysis at this frequency allows marketers to test hypotheses and adjust strategy in real time rather than defending decisions based on stale information.
Your membership also unlocks:
