All use cases

Retain / Save

Return-to-exchange save conversation

The moment a shopper reaches for a refund, a short conversation can turn "send it back" into "send me the right one instead."

Get aheadClose the loopBehavioral triggerSMSChatE-commerce & Retail

What it does

Most returns are a wrong fit, a wrong colour, or a changed mind — not a customer who wants their money back. But the returns portal only offers one path, so the order leaves and the revenue with it. When a shopper signals return intent, the agent opens a proactive SMS or chat conversation that asks why, then offers the exchange, the right replacement, or store credit before defaulting to a refund. The intent that would have walked out as a refund is kept as a swap.

How it works

  1. 1
    Trigger. a return_initiated signal fires — a return started in the portal, or a return-intent message in support.
  2. 2
    Decision. the agent reads the order, the stated reason, and the catalog, then decides the best save offer (size or variant swap, a better-fit alternative, or store credit) within returns policy; a judge gates the message.
  3. 3
    Action. it opens the conversation on SMS or chat — "saw you started a return on the boots; was it the fit? I can swap them for the next size up today" — and walks the chosen path end to end; when the swap is to a higher-value item, any price difference is collected through a secure payment step the shopper completes, never card details taken in the thread.
  4. 4
    Follow-through. the resolution is written back to memory; if the shopper still wants a refund, the agent processes it cleanly rather than stalling, and the outcome (exchange / credit / refund) is logged for the save-rate measure against a holdout.
Example
AgentHi — I saw you started a return on the trail runners. Quick question before it goes back: was it the size or the shoe itself?
CustomerSize, they were a bit tight.
AgentEasy fix — I can send the half-size up today and email you a free return label for the pair you have. Want me to set that up, or would you still rather a refund?

Configuration

How the agent is wired for this use case.

Triggera return_initiated event from the returns/OMS portal, or a return-intent message detected in the support channel.
Tools & actions
  • OMS / returns platform · read the order and return reason; create the exchange or replacement order and the return label.
  • Catalog / inventory · check stock for the swap variant or the best-fit alternative.
  • Store-credit / wallet system · issue credit when an exchange isn't available.
  • Payment provider · generate a secure payment step for the shopper to settle any price difference on a higher-value swap.
  • Messaging channel · run the proactive SMS or chat save conversation.
  • CRM · write back the resolution and outcome for measurement.
Autonomythe save conversation and exchange creation run unattended behind the judge, within a fixed exchange/credit policy; any price difference on a higher-value swap is taken through a secure payment step the shopper confirms, never charged automatically; a refund is always honoured on request rather than withheld — this is a save, not a block.
Channelssms · chat
Escalationa damaged or faulty item, a policy exception, or an unhappy shopper insisting on escalation hands off to a human agent.

What you need

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

Signals

return_initiated, return-reason capture, refund-request events.

Data

order contents and value, return reason, catalog and live stock for swaps, returns/exchange policy, store-credit rules, customer memory and consent state.

Guardrails

judge gating on every outbound message; the refund path stays one tap away and is never withheld; frequency caps; suppression on faulty-item and damaged-order cases; holdout assignment for honest save-rate measurement.

Metrics it moves

  • save-rateup, by converting return intent into exchanges and credit instead of refunds.
  • recovered-revenueup, the order value kept on the books rather than refunded.
  • aovup where the swap or alternative is a higher-value item, measured per saved order.

Related use cases

Prevent / Care

Returns and refunds, judge-gated end-to-end

executes the refund correctly when an exchange isn't the right outcome

Retain / Save

Damaged-order instant make-good

the defect-triggered counterpart, where a refund or replacement is owed

Prevent / Care

Post-resolution follow-up

keeps the relationship warm after the swap is settled

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