Hershey has deployed AI agents to run media mix modelling across its entire brand portfolio on a monthly basis. The system uses Mutinex (powered by Claude and Gemini) and Tracer, replacing the traditional quarterly MMM cycle with continuous, agent-driven analysis.
The shift from quarterly to monthly modelling is the important detail. Traditional media mix modelling is expensive, slow and typically runs on a 90-day cycle. By the time the model produces recommendations, the media landscape has changed. Hershey's approach uses AI agents to ingest performance data continuously, rerun the model monthly and surface budget reallocation recommendations in near real-time.
Expected revenue lift from Hershey's AI-driven continuous media mix modelling
The company expects a 4 to 5 per cent revenue lift from better budget allocation, which on Hershey's roughly $2 billion annual media spend translates to $80 to $100 million in incremental revenue. That is the kind of number that makes CFOs pay attention to marketing measurement.
Mutinex, the Australian company behind one of the AI agents, has built its modelling approach on Claude and Gemini. The agent does not just run the statistical model. It interprets the results, generates natural language recommendations, identifies which brands are over- or under-invested by channel and flags where diminishing returns have set in. Tracer handles the data integration layer, pulling performance data from media platforms into a normalised format the model can consume.
For Australian marketers, the Mutinex connection is particularly relevant. This is an Australian company deploying AI-powered MMM for one of the world's largest advertisers. The same technology is available to mid-market Australian businesses at a fraction of the cost. The gap between enterprise and mid-market marketing measurement is closing because the AI layer makes sophisticated modelling accessible at lower price points.
The broader trend is that AI agents are moving from ad buying (programmatic) into measurement and strategy. An agent that can reallocate $2 billion in media spend based on monthly modelling is making strategic decisions, not just tactical ones. The governance implications mirror what the IAB's Programmatic Governance Council is working on: who is accountable when the agent's recommendation does not work?
Why it matters
Media mix modelling has been a tool reserved for companies with $100 million-plus media budgets and six-figure consulting engagements. AI agents are compressing the cost and cycle time to the point where monthly MMM is practical for mid-market businesses. If your marketing budget is above $50,000 per month across multiple channels, AI-driven MMM is now within reach.
What to do about it
Investigate AI-powered MMM tools like Mutinex, ChannelMix and Lifesight. The cost has dropped significantly in the past 12 months. Start with a quarterly model to establish baselines, then move to monthly as you build confidence in the outputs. The key input is clean, consistent spend and outcome data across all channels. If your data is fragmented, fix that first.
