Method
How Bayesin
reads your spend.
Bayesian media mix modeling and multi-touch attribution, fit on your own data on a schedule, reported with the interval showing.
01
Ingest
Ad platforms, CRM, warehouse, and commerce data land in one touchpoint schema: a touchpoint stream, conversions joined to it, and identity resolution with consent recorded per subject.
02
Resolve
Identifiers resolve to subjects your organization owns. Suppression and erasure are first-class, so a lawful basis exists for every person-level row.
03
Fit
Attribution runs and Bayesian MMM refit on a schedule. Adstock and saturation are estimated per channel. Nothing is fit to a sample dataset.
04
Report
Every figure carries its interval. A page with nothing to read says so instead of standing in a demonstration.
Principles
Industry-agnostic by construction
No vertical is baked into the code. Segments, channels, units, and groups are data your organization defines. Vocabulary presets change labels only.
Intervals, not points
A contribution without an interval is a guess with a confident face. Bayesin reports posteriors and shows the width.
Your rows, every night
There is no demo dataset and no signed-out preview. Models are fit to the organization's own data on a schedule you set.
Partner-funded spend is spend
Co-op, market development funds, and channel programs are modeled as their own inputs, because they move revenue like any other dollar.
Docs
Want the model card?
Priors, adstock and saturation forms, and the diagnostics we report per fit are documented for your data team.