Lifecycle classification
How the autopilot places every contact on a lifecycle stage from their behavior, and the six predefined business models it can use.
Classification is the foundation of the autopilot. On each cycle it looks at each contact’s mapped behaviors across three windows — all-time, the last 30 days, and the last 7 days — and assigns a lifecycle stage. The stage is stored on the contact, so it’s queryable, visible, and usable as a segment condition even before any campaign is generated.
How a stage is chosen
Every model maps behaviors onto a small set of roles — acquisition, conversion (the key milestone), activity (recurring engagement), and optionally intent, churn, and a repeatable behavior whose count tiers up high-value customers. A generic classifier turns those signals into a stage:
- An explicit churn signal always wins — the contact is churned.
- Converted and recently active → engaged (and, with enough repeat volume, a high-value tier like VIP or power user).
- Converted but only mid-window activity → at-risk; no recent activity → dormant.
- Acquired but not yet converted → new (very recent) or onboarding (gone quiet).
Contacts with no mapped-event signal are left unclassified rather than mislabeled. Classification uses no LLM — it’s deterministic and runs as batched, workspace-wide aggregate queries, so it stays cheap even for large contact bases.
The six lifecycle models
Each model supplies its own display labels for the generic stages, so the language matches your business.
- SaaS — new → onboarding → engaged → at-risk → dormant → churned. Behaviors: signed up, activated, logged in, subscription canceled.
- E-commerce — lead → first purchase → repeat customer → VIP → lapsing → churned. Behaviors: purchased, viewed product, added to cart.
- Marketplace — visitor → buyer → active trader → power user → lapsing → churned. Behaviors: signed up, first transaction, transacted, contacted seller.
- Subscription / Media — trial → subscriber → engaged viewer → superfan → at-risk → churned. Behaviors: started trial, subscribed, content consumed, subscription canceled.
- Fintech / Neobank — registered → onboarding → funded → active → power user → dormant. Behaviors: registered, funded account, transacted.
- On-demand / Delivery — new → onboarding → first order → repeat → loyal → lapsing. Behaviors: signed up, placed first order, placed order.
Running it
Classification runs automatically on the autopilot sweep (roughly every fifteen minutes for enabled workspaces). You can also trigger a pass on demand from the autopilot dashboard with Classify now, and the lifecycle distribution chart shows how your contacts spread across stages.
Next steps
- See how stages become action in opportunities.
- Use a lifecycle stage as a condition in segments.
- Make sure the right events are flowing so classification has signal.
Related
Engagement Autopilot
An AI engagement brain that watches each contact’s lifecycle, spots opportunities, and proactively drafts (or ships) tailored campaigns to grow retention.
Events
Send behavioral events about your contacts to trigger campaigns, gate waits, and power segments.
Segments
Group contacts by attribute and behavior rules to target campaigns and manual enrollments.
Opportunities & the review queue
The micro-segments the autopilot detects, how it drafts a campaign for each, and how you approve or dismiss them.