How to Automate Email Triage With n8n

n8n charges per workflow run, not per step — ideal for email triage. Build a Gmail triage workflow in n8n, cloud or self-hosted, step by step.

By Galactic News Editorial · Published 2026-11-08T22:24:21+00:00 · Updated 2026-11-08T22:24:21+00:00

How to Automate Email Triage With n8n

n8n is a workflow automation tool with a key advantage for email triage. It bills per execution, not per step. One execution is one full run of a workflow, however many nodes it has. A ten-node triage flow costs the same as a two-node flow. That makes n8n cheap for long, careful triage chains.

This guide shows how to automate email triage with n8n step by step. You will connect Gmail, classify new mail, apply labels, and log results. You can run it on n8n Cloud or self-host the free Community Edition. No code is needed for the basic version.

Why n8n Fits Email Triage

n8n connects more than 400 apps through a visual node editor. Gmail has a dedicated trigger node. It can watch for new messages and start the workflow. Other nodes can call an AI model, update a sheet, or send a Slack message. You chain them on a canvas.

The billing model is the main reason to pick n8n for triage. Cloud plans bill by execution. Starter costs $20 a month billed annually, with 2,500 executions. Pro costs $50 a month billed annually, with 10,000 executions. Monthly billing costs $24 and $60. All plans include unlimited users, unlimited active workflows, and every integration. The Community Edition is free to self-host with no execution limit. You pay only for your own server.

Self-hosting also keeps your email data on your own server. That matters for teams with privacy rules. Cloud hosting is simpler. It runs in Frankfurt, Germany, and n8n handles updates and uptime. Pick the option that matches your comfort with servers.

How to Automate Email Triage With n8n: The Build

Build the workflow in this order. Test each node before you connect the next one.

1. Set up n8n. Use n8n Cloud for the fastest start, or install the Community Edition on your own server. Sign in and open a blank workflow.
2. Add the Gmail Trigger node. Search for Gmail Trigger and add it. Connect your Google account. Set it to check for new emails. Start with a five-minute interval for testing.
3. Add a filter node. Use an IF node to drop mail you never triage. Exclude newsletters and automated receipts by sender or subject. Filtering early saves executions on mail you already understand.
4. Add an AI classification node. Send the subject and a trimmed body to a language model. Give it a fixed list of categories, such as urgent, reply, newsletter, and archive. Ask for one label only. Trim the body to the first few hundred characters to cut model cost.
5. Add a Switch node. Route the workflow by the AI label. One branch handles urgent mail. One handles normal replies. One handles the rest. Each branch can have its own actions.
6. Add the Gmail label node. Use the Gmail node to add a label to the message. Create the labels in Gmail first. Match the names exactly, or the node will fail.
7. Add a Google Sheets node. Write each triaged email to a sheet. Log the date, sender, subject, and label. This log is your audit trail. It also shows you where the classifier goes wrong.
8. Add a Slack node for urgent mail only. Send a short alert with the sender and subject. Do not alert on every email. Alerts on everything get ignored fast.
9. Test with real emails. Send yourself test messages in each category. Use the Execute Workflow button. Check the labels and the sheet after each run.
10. Activate the workflow. Switch it to Active. Watch the execution log for the first day. Fix errors before they pile up.

Watch Out for the Polling Trap

The Gmail Trigger polls on a schedule. Each poll that finds mail starts an execution. A poll every five minutes means about 8,640 checks a month. If each check counts as an execution, the Starter plan allowance of 2,500 runs out in days. This is the most common cost surprise in n8n Cloud.

There are two fixes. First, lengthen the interval. Checking every 30 minutes uses far fewer runs. Second, use a trigger that only fires on new mail. Some setups use a webhook or a Gmail push setup so empty checks cost nothing. Check the current trigger docs for your version. The right choice depends on your plan and your volume.

Do the math before you pick a plan. Count your daily emails. Add your polling runs. Compare the total with the plan limit. If the total is close to the limit, move to Pro or self-host. A workflow that stops mid-month is worse than no workflow at all.

Designing Triage Categories That Hold Up

Keep four categories. Urgent means a customer or a manager needs action today. Reply means a real person asked a direct question. Newsletter means bulk mail for later reading. Archive means receipts and confirmations.

Write one rule per category in the AI prompt. Keep each rule to one sentence. Test the prompt on twenty real emails. If more than two are wrong, rewrite the rules. Short, concrete rules beat long ones. Add the sender to the prompt when a sender always belongs in one category.

Review the sheet weekly for the first month. Look for the same mistake twice. Fix the rule that caused it. Small weekly fixes make the classifier steady over time.

n8n vs Make vs Zapier for Triage

The three tools bill differently. That difference decides the cost of a long triage flow.

n8n wins when the workflow has many steps. A fifteen-node flow costs one execution on n8n. It costs fifteen credits on Make. Make wins for short flows and for users who want a simpler builder. Zapier wins for the fastest setup on a simple flow. Pick by step count and volume, not by brand.

Common Mistakes to Avoid

The first mistake is a polling interval that is too short. Five minutes sounds responsive. It burns executions fast. Start at thirty minutes. Shorten it only if the volume allows.

The second mistake is sending full email bodies to the AI. Long bodies cost more in model fees. They also distract the classifier with footers and signatures. Trim the body first.

The third mistake is auto-archiving on day one. Label first. Archive only after a week of correct labels. This gives you a safety net while the rules settle.

The fourth mistake is no error handling. When the AI call fails, the workflow should label the email for manual review. Without that branch, failed mail vanishes from view. Add an error path to every AI node.

The Bottom Line

n8n is a strong pick for email triage when your workflow has many steps or your mail volume is high. Build one workflow with a Gmail trigger, an early filter, an AI classifier, and a label step. Log every result to a sheet. Watch the polling cost, and size your plan from real counts. Start on Cloud for speed, or self-host the Community Edition for unlimited runs and full data control. Either way, test with real mail before you trust the labels.