Consumer Goods
Traditional market research takes weeks and costs too much. Synthetic consumer testing gets you there in under 24 hours.
CPG brands face SKU complexity in shelf optimization, innovation pipeline lead time, cannibalization measurement, and the cost of traditional market research. Legacy tools give static answers. Bayesian models are transparent about what they don't know, combine expert knowledge with data, and update as new information arrives.
Nobody else is doing what we're doing in this space. The lack of sophistication at these places continues to surprise us.
Work in This Space
Colgate-Palmolive
Cannibalization & Incrementality Modeling
Bayesian Multinomial Logit discrete choice model measuring cannibalization when launching new toothpaste SKUs. Three innovation horizons modeled with counterfactual inference.
Colgate-Palmolive
Shelf Optimization
Nested logit discrete choice model for shelf assortment optimization. GPU sampling: 4 chains in 6 hours vs. 10+ hours per chain without GPU.
Colgate-Palmolive
Synthetic Consumers
Synthetic consumer panel replication achieving up to 90% alignment with real consumer responses across 9,000+ synthetic responses.
Trusted By
- Colgate-Palmolive
- Procter & Gamble
- The Coca-Cola Company
- Yum! Brands
- Nomad Foods
- Diageo
- Unilever Prestige
- Swarovski
What We Solve
Synthetic Consumer Testing
AI-generated personas simulate product reactions, pricing sensitivity, and concept ranking. Up to 90% alignment with human survey data. Research cycles compressed from weeks to under 24 hours.
Shelf Optimization
Nested logit discrete choice models predict sales impact of adding or removing SKUs. Optimizes margin and revenue per shelf configuration with GPU-accelerated sampling.
Cannibalization Modeling
Bayesian discrete choice models quantify how new product launches redistribute sales across own-brand vs. competitor SKUs. Counterfactual inference for launch scenarios.
Media Mix Modeling for CPG
Hierarchical Bayesian MMM across markets, channels, and seasonality. Budget optimization with full posterior uncertainty quantification.
Why Bayesian
Why Bayesian for Consumer Goods?
- Transparent uncertainty — Models that are honest about what they don't know, combine expert knowledge with data, and update as new information comes in. The opposite of black-box ML.
- Custom flexibility vs. legacy tools — Custom nested logit with arbitrary hierarchy and full uncertainty quantification, replacing static tools like Kantar RichMix.
- Speed of innovation — Synthetic consumers deliver results in under 24 hours at 90% alignment, replacing weeks of traditional focus groups.
“At Colgate-Palmolive, we really value the relationship we’ve built with PyMC Labs. They continue to deliver truly unmatched quality work on the hardest and most cutting edge problems we encounter.”