All use cases

Retain / Save

Cancel-flow save conversation

The moment a customer hits cancel, the flow becomes a conversation that finds the why and fixes it — not a survey and a coupon on the way out.

Next best thingClose the loopBehavioral triggerRisk triggerChatIn-appEmailVoiceSubscriptionFintech

What it does

A customer who clicks cancel still has a reason, and most cancel flows never learn it: a radio-button survey, a generic discount, and the account is gone. The agent turns that moment into a diagnostic conversation. It asks why, listens, and offers the save that actually matches the reason — a pause for the traveler, a downgrade for the over-paying, a fix for the feature that broke — executing the action inline rather than promising a follow-up. And when the customer genuinely wants out, it processes the cancellation gracefully, because a respectful exit is the first step of the next win-back. Save offers are A/B-tested, and the save rate is measured against a holdout so the number is real.

How it works

  1. 1
    Trigger. cancel intent appears — the cancel button is clicked, "I want to cancel" lands in chat, or churn language surfaces mid-conversation.
  2. 2
    Decision. the agent diagnoses the real reason from the conversation plus account memory — usage pattern, billing history, past issues — and selects from a policy-bounded ladder of save offers, A/B-tested over time. A judge gates the exchange for tone and policy.
  3. 3
    Action. the flow opens as an in-app or chat conversation; the agent makes the matched offer and executes it on the spot — pauses the plan, changes the tier, fixes the underlying problem — or completes the cancellation cleanly with no extra steps.
  4. 4
    Follow-through. the outcome and the stated reason are written back to memory for win-back and product teams; accounts close to expiry or beyond the agent's policy hand off to a human in time; saves are reported as lift over the holdout, not gross saves.
Example
AgentBefore I process this — can I ask what's driving the cancellation? If it's something I can fix, I'd rather fix it.
CustomerI'm traveling for a couple of months and just won't use it.
AgentThen don't pay for it. I can pause your plan until you're back — no charges in between, and everything resumes automatically. Want me to set that up, or would you still prefer to cancel?

Configuration

How the agent is wired for this use case.

Triggercancel intent — a cancel-button click, "I want to cancel" in chat, or churn language surfacing mid-conversation
Tools & actions
  • Messaging channel · open the diagnostic conversation in the cancel flow as in-app or chat
  • Billing platform · execute the matched save inline: pause the plan, downgrade the tier, or complete the cancellation cleanly
  • CRM · write the stated reason and outcome back for win-back and product teams
  • Memory store · read past issues and offers already made; record the diagnosed reason
Autonomythe diagnostic conversation and the policy-bounded offer ladder run unattended behind judge gating; plan pauses, downgrades, and cancellations are money-moving and execute within a bounded billing-action policy; cancellation always stays one step away
Channelschat · in-app · email · voice
Escalationaccounts close to expiry or beyond the agent's policy hand off to a human in time

What you need

The inputs this use case runs on. Your channels stay yours; the agent supplies the judgment.

Signals

cancel-button events, cancel intent and churn language detected in conversation, auto-renewal status

Data

plan and billing history, usage pattern, support history, customer memory of past issues and offers already made

Guardrails

cancellation always stays one step away — no dark patterns, no loops; offer ladder capped by policy; judge gating on the conversation; human hand-off for accounts near expiry or outside policy; randomized holdout for save-rate measurement

Metrics it moves

  • save-rateup, by matching the offer to the diagnosed reason instead of leading with a blanket discount
  • churndown, as fixable cancellations get fixed at the moment of intent
  • ltvup, and even processed cancellations leave a clean reason and a warm path to win-back

Related use cases

Retain / Save

Churn-risk early-warning outreach

catches the wobble before the cancel button is ever clicked

Retain / Save

Renewal-window save

the same save conversation anchored to the renewal date

Win back

Memory-personalized win-back

picks up the relationship when the save doesn't land

See it on your own customer journey

Bring one drop-off, one churn cliff, or one silent segment. We will show you what a proactive agent with memory and judgment does with it.

Book a demo