The stage model
Every mature lifecycle program is a variation on the same five-to-six stage shape. Naming conventions differ by vendor and by writer, but the underlying states are consistent:
| Stage | What it means | Typical transition event |
|---|---|---|
| Acquisition | A prospect has been identified but has not yet converted to a customer. | Signup, trial start, first purchase. |
| Activation / Onboarding | The customer has converted but has not yet reached the product's first meaningful value moment. | First core action completed (e.g. first project created, first order shipped). |
| Engagement | The customer is using the product or buying on an ongoing basis. | Recurring usage or purchase cadence established. |
| Retention / At-Risk | Usage or purchase frequency has dropped below a defined threshold. | N days of inactivity, or a usage-frequency drop against the customer's own baseline. |
| Winback / Reactivation | The customer has churned or gone dormant and is being actively targeted to return. | Explicit cancellation, or a long inactivity window with no response to at-risk messaging. |
| Advocacy | The customer is engaged enough to be asked to refer, review, or expand. | Sustained high engagement or purchase frequency over a defined window. |
hypothesis: the exact thresholds (what counts as "N days," what counts as a "core action") are business-specific and cannot be given a universal number - they should be set from your own historical engagement distribution, not copied from a competitor's playbook. This is a reasoning-based recommendation, not a measured claim about any specific business.
Why the transition event matters more than the stage name
A stage label is only useful if a system can detect the transition into and out of it automatically. If "engaged" is defined as a feeling rather than an event (a logged usage count crossing a threshold, a purchase within a rolling window), no platform can trigger off it - see lifecycle triggers & automation for how the event side of this actually gets built, with or without a CRM in place.
Stages and the CRM-bridge question
Where a CRM already exists, stage state often already lives there (a deal stage, a lifecycle-stage property) and the engagement platform should read it, not duplicate it. Where no CRM exists, the stage has to be reconstructed from tracked behavioral events directly - see our CRM-and-analytics alignment use case for the version of this problem that shows up most often in practice.