Automate Stripe Refund → HubSpot Retention Deal + Recovery Email in n8n — Step by Step
A Stripe refund is not an accounting event. It's a churn event with a countdown timer. The moment the money leaves your account, you have a narrow window — usually a few hours — where the customer sti
A Stripe refund is not an accounting event. It's a churn event with a countdown timer. The moment the money leaves your account, you have a narrow window — usually a few hours — where the customer still remembers why they bought, still has your product in their head, and is still emotionally reachable. Miss that window and the refund becomes a silent, permanent loss. Most teams miss it every single time, because the refund lands in Stripe, nobody on the revenue side sees it, and HubSpot keeps showing the customer as "active" for weeks.
This article walks through a production n8n workflow that closes that gap automatically: every Stripe refund updates the HubSpot contact, opens a retention deal, fires a personalized recovery email, and pings Slack — all in under 10 seconds, with zero manual triage.
The problem: refunds are invisible to the people who can save the account
In a typical SaaS or e-commerce stack, the refund lives entirely inside Stripe. Finance sees it on the dashboard. Support may have processed it. But the customer success and sales motion — the people whose actual job is retention — find out days later, if at all, usually through a monthly churn report that's far too late to act on.
The manual alternative is worse. Someone has to watch the Stripe dashboard, copy the customer email, search HubSpot, update the lifecycle stage, remember to create a follow-up task, and draft a "sorry to see you go" email. Under load, this never happens consistently. It gets skipped on busy days, done sloppily on others, and it depends entirely on one person remembering. Meanwhile the data drifts: HubSpot says the customer is a paying subscriber, your MRR reports are wrong, and your CS team is emailing a churned account with an upsell.
The cost compounds. Every refund you don't respond to is a lost win-back opportunity, a corrupted CRM record, and a broken feedback loop — because nobody ever asks why the customer left. You need the reaction to be automatic, instant, and structured. That's exactly what event-driven automation is for.
The solution: one webhook, four coordinated actions
The workflow listens for Stripe's charge.refunded event and orchestrates four downstream actions in parallel:
- Update HubSpot — set the contact's lifecycle stage to a recovery/at-risk state and log the refund amount and reason as properties.
- Open a retention deal — create a HubSpot deal in a dedicated "Retention" pipeline so a human owns the save attempt, with the refund value attached.
- Send a recovery email — a warm, non-defensive message that asks what went wrong and offers a path back (extended trial, discount, or a call).
- Alert Slack — post to your #retention or #cs channel so the account owner sees it in real time, not next month.
Because n8n runs these as branches off a single trigger, the whole thing completes in seconds. HubSpot is accurate before your finance export runs, and the customer gets a human-sounding email while they still remember your product.
Step-by-step setup in n8n
1. Trigger — Stripe Trigger node. Add the Stripe Trigger node and subscribe to the charge.refunded event. n8n registers the webhook endpoint with Stripe automatically once you connect your Stripe credential (use a restricted API key with read access to charges and customers). If you prefer full control, use a generic Webhook node and register the endpoint manually in Stripe's dashboard under Developers → Webhooks. Always verify the signing secret so you don't act on spoofed payloads.
2. Normalize the payload — Set node. The Stripe event is deeply nested. Drop in a Set (Edit Fields) node to extract the fields you actually need: {{ $json.data.object.customer }}, the customer email ({{ $json.data.object.billing_details.email }} or a lookup), {{ $json.data.object.amount_refunded / 100 }} for a human-readable value, and the currency. Cleaning the shape here keeps every downstream node readable.
3. Find the contact — HubSpot node (Get / Search). Use the HubSpot node in "Contact → Search" mode, filtering by email. This returns the HubSpot contact ID you'll need for the update and the deal association. Add an IF node right after: if no contact is found, branch to a "create contact" path so a refund from a customer who isn't in HubSpot yet still gets captured instead of silently failing.
4. Update the contact — HubSpot node (Update). Set lifecyclestage to a custom value like at_risk or churned_recoverable, and write custom properties: last_refund_amount, last_refund_date, and refund_reason. Create these properties in HubSpot first (Settings → Properties) so the writes don't get rejected.
5. Open the retention deal — HubSpot node (Create Deal). Create the deal in a dedicated Retention pipeline, name it something scannable like Retention – {{contact email}} – ${{amount}}, set the amount to the refunded value, and associate it to the contact you found in step 3. Assign an owner (round-robin or the account's existing CSM) so it lands in a real person's queue.
6. Send the recovery email. Use the Send Email (SMTP) node, or a Gmail/SendGrid/Postmark node if you send through a provider. Keep the copy short and human: acknowledge the refund, ask one question ("What made this not work for you?"), and offer a concrete path back. Personalize with the customer's first name and the product from the payload. For scale and deliverability, a transactional provider node beats raw SMTP.
7. Alert Slack — Slack node. Post to your CS channel with the essentials: customer, amount, reason, and a direct link to the new HubSpot deal (https://app.hubspot.com/contacts/<portalId>/deal/{{ $json.dealId }}). Use Slack Block Kit for a clean layout with a button that jumps straight to the deal.
8. Fan out, don't chain. Wire steps 4, 5, 6, and 7 as parallel branches off the normalized data so one slow node (email) doesn't delay the others. Add an Error Trigger workflow or per-node "Continue On Fail" so a single failed branch still lets the rest complete — a Slack outage should never block the HubSpot update.
Why this pays off
The obvious win is speed: sub-10-second reaction versus days of lag. But the deeper value is structural. Your CRM stays truthful, so MRR and churn reporting are correct in real time. Every refund now generates a deal, which means retention becomes a measurable pipeline instead of a vibe — you can track win-back rate, revenue recovered, and which reasons are most saveable. The recovery email running automatically means you're gathering churn reasons at scale, feeding product and pricing decisions you'd otherwise never see. And because it's all event-driven, it scales from 5 refunds a month to 500 without adding a single hour of manual work.
Common pitfalls to avoid
Confusing charge.refunded with charge.dispute.created. Chargebacks are a different event and often need a different, more urgent playbook. Subscribe deliberately.
Not handling partial refunds. A partial refund isn't churn. Check amount_refunded against amount in an IF node and route partials to a lighter branch — log it, maybe Slack it, but don't fire a "we're sad to see you go" email to someone who's still a customer.
Skipping idempotency. Stripe can deliver the same webhook more than once. Store the Stripe event ID and check it (a quick lookup, a dedupe node, or a database) so you don't create duplicate deals and double-email the customer.
Firing before HubSpot is updated. If the email or deal references stale contact data, sequence the contact update first, or read fresh data in each branch. Test with a real refund in Stripe test mode end to end before going live.
Emails that sound like a robot. The whole point is a human moment. Don't lead with policy or defend the refund — ask, listen, offer. Automation delivers the message; it shouldn't make it feel automated.
Ja construimos isso pra voce
Nao comece do zero. O Stripe Refund → HubSpot Retention Deal + Recovery Email e um workflow n8n pronto para instalar — conecta suas ferramentas em minutos, sem codigo.
Instalar por $29 →
