Home / Platform

From raw events to a saved account

Retain is four stages. Everything you configure sits in one of them, and you can see exactly what happened at each one for any account.

Stage 01 · Ingest

Your data, not an industry benchmark

Retain reads what you already produce: product events, invoices, support tickets and CRM fields. There is no shared model across customers, because churn at a billing tool looks nothing like churn at a marketplace.

Connected sources
Product eventsSegmentlive
BillingStripelive
SupportZendesklive
CRMHubSpotsyncing
WarehouseBigQuerynightly
Why this account scored 86
Weekly active seats−42% in 21d+31
Unresolved tickets3 open, 9d old+24
Payment1 retry failed+19
Exec sponsorStill active−12

Stage 02 · Score

Every score explains itself

A number nobody understands does not get acted on. Each score comes with the signals that moved it, weighted and ranked, so a CSM can read it in ten seconds and argue with it if it is wrong.

Stage 03 · Play

The play is the product

Pick what happens at each risk band. Retain runs it and records whether it worked, so the playbook gets better rather than staying a document nobody opens.

High risk · Enterprise

Human first

Task the account owner with a call inside 24 hours, with the three drivers and a suggested agenda attached.

Used on 148 accounts31% saved
High risk · Self-serve

Targeted offer

Two months at 40% with a one-click accept, expiring in seven days. Never sent twice to the same account.

Used on 1,902 accounts19% saved
Medium risk

Re-onboarding nudge

An in-app sequence pointing at the feature they stopped using, triggered only if usage is still falling after seven days.

Used on 3,410 accounts12% recovered
Any risk · Billing

Failed payment rescue

Smart retries plus a card-update link sent on the channel they last replied on, before the subscription lapses.

Used on 820 accounts64% recovered

Stage 04 · Measure

Held out, so the number means something

Retain holds back a control group from every play. The revenue we report as recovered is the difference against that control, not the total value of everyone who happened not to churn.

Q3 · play performance vs control
Failed payment rescue+64% vs 9%₹18.2L
Human first+31% vs 14%₹11.7L
Targeted offer+19% vs 11%₹6.4L
Re-onboarding nudge+12% vs 10%₹0.9L

See it run on your data

Send a sample export and we will return a scored cohort with the drivers behind it, within two working days.