Months to deploy.
Black box results.
Someone else's cloud.
There's a better way.

OpenRetailScience is open source retail analytics that runs in your database, handles billions of rows, and gives you back control of your KPIs.

The OpenRetailScience O logo rendered as a splash of green liquid glass

Built with expertise doing analytics for scale-ups to multinationals

LoblawsIKIDominos PizzaMigrosSephoraSainsburysNectarMetroColesGANNIMindful ChefAuchan

What's included

A complete toolkit for retail analytics

From customer segmentation to revenue trees. Every module is tested, documented, and ready to use.

Your cloud, your data

Runs in your database. Your data stays safe and secure in your environment.

Value in days, not years

Modules for churn, gain-loss, cross-shop, segmentation, revenue trees, and more.

Own your KPIs

Every calculation is auditable and yours to extend, with no dependency on a third party's roadmap.

Built for scale

Python only orchestrates. Tested with retailers on their own transaction data.

Raw data in.
Decisions out.

Databricks, Snowflake, BigQuery

Among ~20 supported database backends

Billions

Of rows handled. The heavy lifting happens in your database, not in Python

Days, not years

From data to business value. Pre-built modules for the analyses you keep rebuilding

Open Source

No vendor lock-in, no black boxes, no data leakage

See it in action

From quick analyses to end-to-end workflows, see what you can build. Every chart here was drawn by OpenRetailScience with its default styling.

Waterfall chart: new customers add $610K, basket growth $380K, lapsed customers cost $290K, price adds $170K and returns cost $110K, for a net $760K lift
Visualization

Revenue bridges in one call

waterfall.plot() turns a list of drivers into a board-ready revenue bridge.

Three-way Venn diagram of Fresh, Bakery and Health and Beauty shoppers
Analysis

Cross-shop analysis

See how customers shop across categories, brands, or stores with automated Venn diagrams and overlap metrics.

Cohort retention heatmap: the December cohort has the highest retention in each of its first four months
Customer

Cohorts and retention

Retention, spend and churn by acquisition cohort, stepped into readable bands.

workflow.py
from openretailscience.segmentation.hml import HMLSegmentation
from openretailscience.segmentation.segstats import SegTransactionStats

segments = HMLSegmentation(transactions)
joined = transactions.left_join(segments.table, "customer_id")
stats = SegTransactionStats(joined, grouping_sets=[("segment_name",)])
df = stats.df.set_index("segment_name")
share = df.spend / df.spend.sum()
print(f"Your top 20% of customers drive {share['Heavy']:.0%} of spend")
Workflow

Composable workflows

Chain modules together. Use the output of one analysis as input for another to build complete analytical pipelines.

Your database. Your analytics. Get started.

Install OpenRetailScience and run your first analysis in minutes. Open source, well-tested, and ready for production.

pip install openretailscience