Beneath AnalyticsSaaS
Churn prediction model for a SaaS business losing 4% per month.
A B2B SaaS company was losing 4.2% of customers monthly with no early warning system, while product usage sat in a database customer success could not reach. Beneath Analytics unified the data in BigQuery, built a 45-day behavioural churn model at 78% precision, and cut churn to 2.8% in four months.
B2B SaaS company, project management tools
What was going wrong.
Monthly churn was running at 4.2%. The customer success team was reactive: they only heard about at-risk accounts when a cancellation came in. Product usage data existed but sat in a separate database that nobody in customer success had access to, so there was no early warning system.
How we built it.
Steps in delivery order, each one depends on the one before it.
BigQuery data warehouse unifying product usage, billing, support, and CRM data into one model.
Churn prediction model using 45-day behavioural signals, validated at 78% precision against held-out cohorts.
Customer health score dashboard for the customer success team, updated daily.
Automated Slack alerts when an account crossed a risk threshold, so intervention happens before renewal.
What changed.
Every metric measured against the baseline we inherited, reported the same way we report to clients.
4.2% to 2.8%
Monthly churn: reduced in four months
45 days
Churn visibility: earlier warning on at-risk accounts
About €280K
ARR retained: estimated, in year one
10 weeks
Delivery time
The solutions this ran on.
Each of these engagements drew on the solutions below. Their pages set out the capabilities, platforms, and outcomes you can expect.
- AI Consulting & Development
Data Science
Data science answers the questions your dashboards cannot: who will churn, what demand will be, which segments matter. Beneath Analytics builds churn models, forecasts, and segmentation on your own data, integrates them into the tools your team already uses, and validates them with properly designed experiments.
View solution - Tracking & Analytics
Analytics & Business Intelligence
Analytics and business intelligence means Beneath Analytics turns your tracked data into decisions: attribution modelling, predictive models, and dashboards built around the questions your team actually asks and adopted by the people who use them.
View solution
Common questions.
Monthly churn fell from 4.2% to 2.8% within four months. Customer success received health scores daily and Slack alerts when an account crossed a risk threshold, so outreach happened 45 days before renewal instead of after cancellation.
78% precision, validated against held-out cohorts. The model uses 45 days of behavioural signals from product usage, billing, support, and CRM data unified in BigQuery, rather than demographics or a single engagement metric.
About €280K of ARR was retained in the first year, estimated from the accounts that stayed after intervention. Earlier visibility at 45 days gave customer success time to act before renewal conversations began.
Ten weeks from data unification to the health-score dashboard and Slack alerts. That covered the BigQuery warehouse, model training and validation, the customer success dashboard, and automated risk-threshold alerts.
Get your Free AI Readiness Snapshot
A 30-minute session, no pitch: you get a one-page opportunity map showing where AI can make a real difference in your business.