Revenue and customers
Churn & Retention
Which customers will leave, why, what to do about it, and what that was worth. Scores and reasons land in the CRM your retention team already works in, and every result is measured against a holdout.
The problem
Customers leave quietly, and most save efforts begin after the cancellation request arrives. Where teams do act early, offers tend to go to everyone at risk, including customers who would have stayed anyway and some who leave because they were contacted.
Who owns it
The head of retention or customer success, with the chief revenue officer. Product and pricing owners receive the driver report, and a data counterpart signs the definition of churn.
What it learns from
- Billing and subscription history
- CRM accounts, activity, and field history
- Product usage
- Support tickets
- Marketing engagement and consent state
How it works
Predict, decide, act, measure
The same four steps in every solution, in this order, each one visible to the team that owns it.
- 01
Predicts who will leave, and why
A model trained on point-in-time history scores each customer's risk of leaving within a window your team can act in, with the top reasons on every score. Features that leak the outcome are excluded by construction, and customers are resolved to the account or household the team actually acts on.
- 02
Decides through an explicit policy
Risk, value, and the dominant reason place each customer in a treatment: a call from their success manager, a plan review, an onboarding intervention, a service escalation, an automated journey, or no contact at all. Discount caps, contact limits, budgets, and exclusions for customers in collections or with an open complaint live in the policy, not in the model.
- 03
Acts in the CRM and success tools
Scores, reasons, and tasks with an owner and a due date land in the CRM and the customer-success platform, and journeys trigger in marketing automation. The score appears where the owner works, never in a separate dashboard.
- 04
Measures against a holdout
A randomized holdout in each risk tier receives no model-driven treatment. Incremental retention is the treated group minus the holdout, never raw campaign response, and a side-by-side run against the current process comes before any routed action.
What it moves
The business levers
Retained revenue
Recurring revenue kept that would otherwise have left. This is the primary lever.
Retention spend
Offers concentrated on the customers a contact can keep, instead of spread across everyone at risk.
Early warning
A usage collapse in a key account surfaces while there is still time to act on it.
Fixes upstream
A standing driver report gives product, pricing, and service owners the causes to fix, not only the customers to call.
Proof
How the return is proven
The return is the revenue retained in the treated group over the holdout, net of offer costs, stated against the return model agreed before work starts. Guardrails compare complaints, discount spend, margin, and satisfaction between the treated group and the holdout, so a gain is never bought out of sight.
What you see: the ROI readout
Every solution produces the same readout, assembled from the system's own records rather than built by hand.
Correctness
The checks the pipeline passed on your own data: leakage, point-in-time, and entity resolution.
Incrementality
The incremental effect against the holdout or equivalent design, with its confidence interval.
Return
The measured return against the investment, set beside the return model agreed before work started.
Guardrails
Complaints, cost, margin, fairness, and contact limits, each shown against its bounds.
Audit extract
Every score, action, and outcome, with the model and policy versions behind it.
Governance
What is governed
Access, approvals, and the audit trail are part of the solution, not a review step added at the end.
Identity-scoped access
The scored list is commercially sensitive. Access runs through your single sign-on, and each person sees the accounts they own.
Approval gates
Concessions above the cap, playbook changes, and any exemption from the holdout need a named approver, and exemptions are counted.
Audit trail
Every score, action, and outcome is logged with the model and policy versions that produced it.
GDPR and sector rules
Profiling has a signed lawful basis under GDPR, and vulnerable-customer suppression is a rule in the policy. In banking and insurance, the same audit and validation records support model-risk review and the NAIC model bulletin.
Why it holds up
Trained on your data, framed by the outcome
This solution trains models on your own data, checks every stage from data to decision, and is measured against the outcome it was built for. Because its predictive power is measured, it also shows the drivers behind the metric, so your team can fix causes as well as act on scores.
Why solutions hold upStart with one decision
Bring the problem and the systems your team works in. We scope the first solution with you, agree how its return will be measured, and run it in production with your people.