Product · Models

Production models

A model is a set of dataflows that together produce an output for a business subject area. These six were built with our customers and run on their schedules today – executing as pushdown SQL in their own Snowflake.

Customer LTV
Lifetime value by acquisition cohort, channel and product.
Subscription14 stepsMon · 06:00
Deployed running daily
Blended CAC
Revenue after ad spend and blended CAC across Meta and Google.
eCommerce11 stepsdaily
Ready Deploy
Retention & repurchase
Time-to-second-order and repeat-purchase curves by cohort.
eCommerce9 stepsweekly
Ready Deploy
RFM segmentation
Recency, frequency and monetary scoring, out of the box.
eCommerce12 stepsdaily
Ready Deploy
MRR movement
New, expansion, contraction and churned revenue, broken out.
Subscription13 stepsMon · 06:00
Deployed running weekly
Churn & cohorts
Subscriber churn and retention curves by signup cohort.
Subscription10 stepsweekly
Ready Deploy

Categories reflect where each model was first built. The dataflow logic – cohorting, attribution windows, revenue movement – applies to any subject area with the same shape.

How they work

A deployed model is just a dataflow

There is no black box. Deploying a model writes an ordinary Refyner dataflow into your workspace – every join, filter and calculated field is visible and editable, the generated SQL is there to read, and you set the schedule. The AI assistant can map a model against your real schema and explain what each step does before you run it.

Deploy one, or build your own.

Connect Snowflake and push a model to your warehouse – then change whatever you need.