Hershey Accelerates Marketing Decisions with Agentic AI-Powered Mix Modeling
Key Takeaways
- ▸Agentic AI reduces marketing analysis latency from 6+ months to weeks, enabling brands to optimize spend allocation monthly rather than quarterly
- ▸Multi-agent systems—where each AI specializes in a specific domain—outperform single-model approaches for complex business analysis like marketing mix modeling
- ▸Data infrastructure readiness is as critical as AI capabilities; companies must clean and standardize data before agentic AI can be effective
Summary
Hershey is overhaul its marketing mix modeling process by partnering with analytics platforms Mutinex and Tracer, powered by Anthropic's Claude and Google's Gemini, to shift from quarterly analysis cycles to monthly decision-making across a $2+ billion marketing budget. Previously, Hershey's manual MMM analysis created severe lag—the company received a full read of 2024 data midway through 2025, while already planning for 2026. The new agentic AI system, where specialized AI agents handle domains like econometrics, competitive pricing theory, and model diagnostics, can now deliver complete models in as little as three weeks.
The shift directly addresses marketing's credibility challenge: demonstrating ROI and justifying budgets as investments rather than costs. Tracer handles data infrastructure—cleaning and standardizing fragmented data across Hershey's marketing and retail systems—enabling Mutinex's AI agents to operate faster and more reliably. By running models monthly across the entire brand portfolio instead of quarterly for just five brands, Hershey expects to increase revenue attributable to media by 4-5%, with the ability to make spend adjustments based on near-real-time insights rather than historical lag.
- As agentic AI matures, marketing can transition from a cost-center narrative to demonstrable ROI-driven investment, reshaping how CMOs justify budgets
