Complete Guide: NPS Low Score → Automatic CS Recovery Workflow with n8n

Your customers are telling you exactly when they are about to churn — and most teams do nothing with that signal for days. A detractor submits an NPS score of 3, the response lands in a spreadsheet, and by the time a CS rep reviews it on Friday, the customer has already renewed their frustration or started evaluating a competitor. The window to recover a dissatisfied customer is measured in hours, not weeks. This guide shows you how to close that gap entirely with an n8n workflow that reads every NPS response the moment it arrives and routes it — automatically — to the right recovery, nurture, or advocacy path.

The Problem: NPS Data That Nobody Acts On

NPS surveys are cheap to send and expensive to ignore. The typical failure isn't collecting the score — it's the dead time between response and action. Three structural problems compound:

• Latency. Detractors (0–6) need contact while the frustration is fresh. A manual triage queue reviewed once a day means most at-risk customers wait 12–48 hours for a human to even notice.
• Uniform treatment. A promoter scoring 10 and a detractor scoring 2 get dropped into the same inbox. Promoters who would happily leave a public review never get asked; detractors who need de-escalation get a generic "thanks for your feedback" autoresponder.
• No closed loop. Scores rarely make it back into your CRM against the actual account, so CS has no context and leadership has no trend line. The survey becomes a vanity metric instead of an operational trigger.

The result: you pay for the insight and throw away the value. Recovering a single detractor on a $200/mo plan is worth more than most of the marketing spend that acquired them — yet the recovery motion is the one nobody automates.

The Solution: One Workflow, Three Routes

The core idea is a single n8n workflow that fires on every survey submission and branches on the score into three distinct motions:

• Detractors (0–6) → Recovery. Create a high-priority CS ticket, alert the account owner in Slack within seconds, and log the risk against the account in your CRM.
• Passives (7–8) → Nurture. Enroll the contact in an educational email sequence that pushes them toward the features that correlate with promoter scores — no human touch required.
• Promoters (9–10) → Advocacy. Send a timed review or referral request while goodwill is at its peak, routing them to G2, Capterra, or your referral program.

Because everything runs off one trigger, you get a consistent, auditable path for every response — and you can tune each branch independently without rebuilding the pipeline.

Step-by-Step Setup in n8n

Here is the node-by-node build. The whole thing takes under an hour if your survey tool can send a webhook.

1. Capture the response

Start with a Webhook node set to POST. Point your survey provider (Typeform, Delighted, SurveyMonkey, or a native form) at the production URL. If your tool can't push, swap in the provider's trigger node — Typeform and Delighted both have dedicated n8n triggers — or use a Schedule Trigger polling their API every 15 minutes as a fallback.

2. Normalize the payload

Add a Set (Edit Fields) node to extract a clean, predictable shape: email, score, comment, and account_id. Cast the score to a number explicitly — many survey webhooks send it as a string, which will silently break numeric comparisons downstream. Use an expression like {{ Number($json.body.answers.score) }}.

3. Branch on the score

Use a Switch node (not three chained IFs — the Switch is cleaner and easier to audit). Configure three rules on the score field in Number mode:

• Output 0 — Detractor: score is smaller or equal to 6
• Output 1 — Passive: score between 7 and 8
• Output 2 — Promoter: score is larger or equal to 9

Set the fallback output to route to an error branch so malformed scores never vanish silently.

4. Build the detractor recovery branch

From the Detractor output, run three actions in parallel:

• A CRM node (HubSpot, Pipedrive, or Salesforce) to update the contact/deal with a churn_risk = high property and log the score and comment as a note.
• A ticketing node (Zendesk, Freshdesk, or a generic HTTP Request) to open a priority ticket assigned to the account owner.
• A Slack node posting to the CS channel: score, verbatim comment, account name, and a deep link to the ticket. This is the alert that turns a 48-hour lag into a 2-minute response.

5. Build the passive and promoter branches

For Passives, connect an email/marketing node (Mailchimp, Customer.io, or Brevo) that enrolls the contact into a nurture list. For Promoters, add a short Wait node (1–2 hours, so the ask doesn't arrive on the same minute as their survey) followed by a Send Email or Gmail node with a personalized review-request template and a direct link to your G2 or Capterra profile.

6. Close the loop and handle errors

Wire every branch back into a final Google Sheets or database node that appends the response, score, route taken, and timestamp — your permanent NPS ledger for trend reporting. Finally, attach an Error Trigger workflow that pings Slack if any execution fails, so a broken CRM token never means a lost detractor.

The Benefits: What Changes When This Runs

• Response time collapses. Detractors get a human within minutes instead of days — the single biggest lever on recovery rate.
• Promoter reviews compound. Asking every 9–10 at peak goodwill can multiply your review volume without a single manual outreach.
• CS focuses where it matters. The team stops triaging a queue and starts working pre-qualified, pre-contextualized recovery tickets.
• Leadership gets a real signal. Because every score lands in a structured ledger tied to accounts, NPS becomes a leading churn indicator instead of a quarterly slide.
• It scales for free. Whether you get 10 responses a week or 10,000, the workflow cost and latency stay flat.

Common Pitfalls to Avoid

• Comparing strings to numbers. The number-one silent bug. If your Switch always routes to the fallback, your score is still a string — cast it in the Set node.
• No idempotency. Webhook retries can fire the same response twice, double-alerting and double-emailing. Deduplicate on a response ID using a Remove Duplicates node or a lookup against your ledger before acting.
• Instant promoter asks. Firing a review request the same second the survey submits reads as robotic. Always insert a Wait node.
• Alert fatigue on passives. Don't route 7–8s to Slack. They don't need a human; they need nurture. Reserve human alerts for detractors only.
• Silent failures. Without an Error Trigger, an expired CRM API token will drop recoveries with no warning. Test the failure path, not just the happy path.
• Missing account context. If you can't map the survey email to an account_id, CS gets an alert with no history. Enrich early — a CRM lookup right after the Set node pays for itself.

NPS stops being a report and starts being an operating system for retention the moment every score triggers action automatically. Build the three branches once, and every future response works for you — recovering the customers you were about to lose and amplifying the ones who already love you.

PRONTO PARA USAR

Ja construimos isso pra voce

Nao comece do zero. O NPS Low Score → Automatic CS Recovery Workflow e um workflow n8n pronto para instalar — conecta suas ferramentas em minutos, sem codigo.

Instalar por $49 →