Unfamiliar with a term? See the Glossary for definitions.
Marketing mix modeling is a statistical analysis technique that measures the impact of marketing activities on sales and other business outcomes. Unlike digital attribution (which tracks individual user journeys), MMM uses aggregate time-series data to determine how much each marketing channel contributes to your business results.
Simba uses Bayesian MMM powered by PyMC-Marketing, which provides causal attribution while accounting for saturation effects, adstock (carryover), and seasonality.
Simba is built on the open-source PyMC-Marketing framework — meaning full transparency, no black boxes. You can inspect every prior, every model parameter, and every assumption behind your results.
Key differentiators:
For a comparison against a specific tool, talk to us.
Simba is built for:
No. Simba’s interface lets you configure Bayesian models, set priors, and run optimizations without writing any code. Smart defaults and an Data Validator guide you through the process, so you can build reliable models even if you’re new to marketing science.
For teams that want deeper statistical control, the platform exposes all model parameters for fine-tuning.
At minimum, you need:
Yes — Simba offers a 28-day free trial you can start yourself: sign up, confirm your email and sign in. Every trial account starts with a sample model fitted on synthetic data, so you can explore results, the optimizer and the MCP integration before uploading your own data. The trial includes 10 model fits, 10 optimizations and 10 scenarios per 30 days, and one saved model of your own beside the sample.
Bayesian modeling is a statistical approach that combines observed data with prior knowledge to produce probability distributions over possible outcomes — rather than single-point estimates. This means you get:
PyMC-Marketing is the leading open-source library for Bayesian marketing analytics. Simba is built on top of PyMC-Marketing, which means:
After uploading a CSV, choose Start Validator Agent and choose a validation AI model to request a data review. The agent reports issues and suggestions for you to inspect before modelling. Uploading alone does not start the agent, and its findings do not certify that the data is suitable for causal inference.
Simba can model any marketing channel where you have time-series data: TV, digital display, paid social, paid search, OOH, video, audio/radio, email, print, and affiliate marketing.
Your data is encrypted at rest and in transit using industry-standard protocols, and access is isolated per project: only the owner and the team it is shared with can read it.
Yes. Simba is compliant with applicable data protection regulations. We practice strict data minimization and work with aggregated marketing data only — no personally identifiable information is required.
Your data is stored encrypted, with access isolated per project. Enterprise customers can discuss custom hosting arrangements for specific data residency requirements.
Simba currently offers Enterprise (full platform with dedicated support) and Managed (done-for-you modeling by PhD statisticians). Self-service tiers for individual analysts and smaller teams are coming soon.
Yes — a 28-day free trial with full platform access, 10 saved models, 10 optimizations, and 10 scenarios. No setup fees.
Yes. All plans can be cancelled at any time with no cancellation fees.
Can’t find what you’re looking for? Open a GitHub issue or email info@pymc-labs.com.