You just inherited 100 n8n workflows. Now what.

The rule for the first 24 hours: read-only.

The single most common mistake when inheriting an n8n instance is "cleaning up" before you understand what's there. You deactivate a workflow that looks dead and break a quarterly process that fires next Tuesday. You rename a webhook trigger and silently sever a Stripe integration.

For the first 24 hours, your job is to look, not touch. Don't deactivate, don't rename, don't archive. Build the map first.

Hour 1 to 4: Inventory.

You need a list. Every workflow, with:

• Name
• Active / inactive
• Trigger type (schedule, webhook, manual, error)
• Last run timestamp
• Total executions in the last 30 days
• Error count in the last 30 days
• Apps/credentials it touches

In n8n you can pull this from the workflows API endpoint and join against execution history. If you're doing it by hand from the UI, it's an afternoon for 100 workflows — and you'll miss things. Either way, put it in a spreadsheet so you can sort and filter.

What you're looking for:the gap between "active" and "actually running." A workflow flagged active but last-run 90 days ago is either dormant or broken. Both warrant a closer look.

Hour 4 to 12: Categorize.

Sort the inventory into four buckets. Don't skip this — it determines what you do next.

• Active & healthy: ran recently, low error rate, makes sense by name. Leave alone.
• Active but failing: ran recently, high error rate. These are bleeding right now. Triage today.
• Active but idle: active flag on, no recent runs. Either dormant or the trigger broke. Investigate before disabling.
• Inactive: flag off, never runs. Probably safe but read before deleting — these are often work-in-progress or paused-on-purpose.

Hour 12 to 24: Find the dependencies.

Two kinds of dependencies will trip you up if you don't map them first.

Execute Workflow chains

One workflow that calls another via the Execute Workflow node. In a tangled instance these can be 4+ levels deep — a webhook fires "router," which calls "process_lead," which calls "notify_sales," which calls "update_crm." Deactivate any one of them and the chain breaks silently.

Map them. For each workflow, list its callers and its callees. The workflows that get called by the most other workflows are your hub workflows — they need the most protection because the blast radius is largest. More on dependency tracking.

Shared credentials

One Stripe credential might be used by 17 workflows. Revoke or rotate it and they all break at once. Pull the credential inventory — which credential each workflow uses — and you can see at a glance what depends on what.

Day 2: Set ownership.

For every active workflow you're keeping, answer two questions:

• Who at the client owns the business process this serves? (Not who built it. Who relies on it.)
• What happens if this stops running for a week? Low impact = leave alone. Medium = get on a monitoring schedule. High = needs alerting, today.

Capture ownership in n8n's tags or as notes on the workflow. The future-you who gets paged at midnight will thank you.

Day 3: Alert on the critical ones.

For the workflows you tagged high-impact, you need at minimum:

• Failure alerts — an email or Slack message within minutes of a workflow erroring, naming the workflow and the failing node.
• Staleness alerts— if a workflow that should run hourly hasn't fired in 6 hours, something is wrong (the trigger broke, the credential expired, n8n is down).
• Dedup— a flaky workflow shouldn't page you 30 times an hour. Group by workflow + type, suppress duplicates within a window.

n8n's built-in error workflow does the first one badly and the other two not at all. Why error workflows fail consultants.

Day 4 to 5: Ship a status report.

Before you start "optimizing" or "cleaning", send the client a one-pager. Include:

• Inventory count (active vs. inactive)
• Categorization breakdown (healthy / failing / idle)
• Top 5 hub workflows with dependency counts
• Top 5 broken / silently-failing workflows
• What you're going to do this month, in order

This is the audit deliverable. It earns trust, surfaces problems the client didn't know about, and sets expectations for the work ahead. More on n8n client reporting.

What to fix first, after the audit.

In rough priority order:

• Anything failing silently on a high-impact workflow. Money-losing or compliance-risking.
• Expired or expiring credentials. Future failures you can prevent.
• Workflows tagged active but idle. Decide: turn off or fix.
• Hub workflows without test coverage. A single bug in a 10-callees hub is a real outage.
• Workflows nobody owns. Either find an owner or retire them.

The shortcut

The audit playbook above is what it takes to do this by hand. We built WorkflowRadar to compress most of it into a few minutes: connect the n8n instance, get the inventory + dependency map + health scores + alerting in one go.

If you'd rather have a printable version of this entire checklist to use the next time you take over a client's stack, grab the one-page PDF below.