Updated on: .
Quick answer: Pipedream says Workflows and String shut down on 31 March 2027, with Workflows data deleted by 30 April 2027. Pipedream Connect is not affected. Export your projects first, rebuild each workflow on Zapier or n8n, and test duplicates and failures before cutting over. Zapier suits low-maintenance business flows; n8n suits webhook-heavy workflows you want to control.
Pipedream vs Zapier now needs an important update: Pipedream says its Workflows and String products will shut down on March 31, 2027. Pipedream Connect remains supported. If you need a platform for a new hosted business workflow, do not treat those two Pipedream products as the same choice.
For an existing Pipedream workflow, the useful question is how to move its triggers, code and app actions without losing records or repeating writes. Zapier is one possible destination; n8n is another when you need more control over workflow logic or hosting.
This guide compares workflow ownership, usage and migration checks. It is based on official documentation, not a claimed paid-account speed benchmark.
What changed at Pipedream?
The official shutdown notice says Workflows is in maintenance mode, existing workflows can keep running through March 31, 2027, and new signups and subscriptions are no longer accepted. It also states that Workflows data will be deleted by April 30, 2027 and provides a project-level JSON export tool.
Connect API, MCP servers and API Proxy are outside that shutdown. Connect is an integration layer for apps and agents; it is not simply a replacement workflow canvas. Confirm which product your application uses before planning a migration.
Pipedream vs Zapier: choose by what you are running
| Your situation | Useful next step | Why |
|---|---|---|
| New app-to-app business workflow | Test Zapier or another continuing workflow platform | Pipedream Workflows is being retired |
| Existing Pipedream Workflows user | Export, inventory and rebuild a representative flow | Code, credentials and trigger behaviour need separate checks |
| Application using Pipedream Connect | Review Connect’s own API and billing requirements | The shutdown notice does not apply to Connect |
| Need self-hosted workflow execution | Compare n8n and self-hosted alternatives | Zapier is a managed workflow service |
Where Zapier can fit a migration
Zapier is worth testing when your current flow is mainly supported app actions and a business team will maintain it. Check the exact trigger, action and account permissions for each app. An integration name in a catalogue is not proof that your custom field or API operation is available.
For a lead workflow, start with these steps: receive the lead → validate the required fields → search for an existing contact → create or update it → notify the team. Map fields from a test event, then check the destination record. A successful editor test is only the first check.
Custom code may still need to be rewritten. Inspect the runtime, dependencies, data size and authentication pattern supported by the destination. Do not assume a Pipedream script can be pasted into a Zapier Code step unchanged.
If your workflow uses a niche API, check Zapier’s API request options before deciding how to replace it. Choose the appropriate method for the authentication and operation, rather than putting a secret in an ordinary text field.
Compare costs using the same workflow
For existing Pipedream Workflows accounts, historical execution credits and Zapier tasks measure different things. Step count alone is not a fair price comparison.
Pipedream’s pricing documentation describes Workflows credits by compute time and memory per workflow segment: one credit per 30 seconds at 256 MB. Branches, delays, higher memory and dedicated workers can change usage. Connect has separate usage and external-user billing rules.
Zapier counts eligible successful actions as tasks. Triggers, Filters and Paths do not use tasks; some AI, code and other products have different rates. Replaying successful actions can count again. Check the current task rules and your actual run history.
For an illustrative comparison, suppose one successful lead uses two standard billable actions: one CRM write and one Slack message. A hundred such leads would produce 200 tasks under that assumption. Extra search actions, AI steps, retries or different rates change the result. This is arithmetic for a stated example, not a quote for your account.
Record the destination platform’s actual usage for a normal lead, a duplicate and a failed write. Add external API fees and migration work. A lower subscription price does not make a difficult rebuild free.
Inventory your Pipedream workflow before exporting
- Trigger: webhook, schedule or app event; note the sender and authentication.
- Code: language, packages, environment settings and returned fields.
- Connections: app accounts and required scopes; reconnect them in the destination.
- State: duplicate keys, cursors, stored IDs and progress records.
- Side effects: records created, messages sent and files changed.
- Recovery: retries, alerts, manual review and the last completed stage.
Keep an export as a reference, then verify what it contains. A workflow structure file is not proof that saved state, secrets and credentials are portable. Keep sensitive exports out of public repositories and blog attachments.
How to export Pipedream workflows before the shutdown
According to Pipedream’s export documentation, exporting is done per project:
- Open the project in the Pipedream dashboard and choose Settings.
- Click Export Workflows. Pipedream builds the archive in the background.
- Download
project_export-<project_id>.zipfrom the project’s File Store. The docs say exports require a plan that includes File Stores.
The ZIP holds one JSON file per workflow plus a manifest.json. It includes triggers, steps, code, configured props and control-flow branches. It does not include everything:
| Item | In the export? | What you do |
|---|---|---|
| Workflow structure, step code, trigger settings | Yes | Use it as your rebuild specification |
| Secret props and connected-account tokens | No (redacted) | Reconnect accounts and re-enter secrets on the new platform |
| HTTP auth settings | No (redacted) | Recreate the webhook authentication |
| Environment variables | No | Re-enter them in the new tool |
| Event history, data stores, File Store files | No | Download or copy them separately before 30 April 2027 |
| Workflows never deployed | Skipped | Deploy first if you still need them |
A single export holds up to 500 workflows or 256 MB, whichever comes first; if a limit is hit, manifest.json marks the export as truncated and lists the skipped workflows. Move those into another project and export again. Each project can have only one export at a time, so delete the existing archive from the File Store before re-exporting.
Migration example: webhook lead to CRM
1. Preserve the event and record IDs
Keep the provider’s event ID and your contact lookup key. They solve different problems: the event ID helps stop duplicate delivery; the contact key helps find the existing person. Do not use a randomly generated new ID every time an event is retried.
2. Rebuild the validation before the write
Use a sample with missing contact details. The new flow should stop or request review instead of creating a partly identified contact. If the old code trims text, normalises dates or handles arrays, test each behaviour explicitly.
3. Replace the CRM action and check its output
Confirm whether the action creates, updates or upserts records. Compare the actual stored fields with the original input. Keep the returned record ID so later steps can refer to the same contact.
4. Keep notification separate from record creation
If Slack fails after the CRM write succeeds, repeat the notification stage instead of creating another contact. If a write times out, inspect the CRM before sending another request. A timeout does not tell you whether the other service completed the action.
5. Plan the trigger cutover
Test in a separate destination. Then decide when the old trigger will stop, how queued events will be handled and how you will reconcile records. Running two production triggers at once can create duplicates unless you have a shared, reliable duplicate check.
When n8n is worth considering
n8n is worth testing when the flow needs custom logic, several branches or a hosting option you control. It still requires correct data mapping and recovery design. Self-hosting also makes your team responsible for infrastructure, updates and backups.
Begin with the Cloud versus self-hosting decision, then use a concrete workflow such as email extraction to a CRM.
Migration test checklist
| Test | Expected outcome |
|---|---|
| One complete lead | One correct CRM record and one notification |
| The same event twice | No second completed write |
| Missing identity field | Review or rejection before the CRM action |
| CRM timeout | Reconciliation before another write |
| Notification failure | Retry notification without recreating the contact |
| Large payload or many records | Explicit pagination, limits and usage checks |
Keep a short results log with the input, record ID and observed outcome. That is stronger evidence for your decision than an unsourced claim that one tool is five times faster.
Frequently asked questions
Is Pipedream shutting down completely?
No. The official notice concerns Workflows and String. Connect, MCP servers and API Proxy remain supported.
Should a new workflow use Pipedream Workflows?
The official notice says new signups and subscriptions are no longer accepted. Evaluate a continuing workflow platform for a new project.
Can Zapier replace every Pipedream script?
Do not assume that. Check code runtime, packages, authentication and data limits. Some workflows need a different design or a separate service.
Is Zapier always more expensive?
No. Compare the same workload using its actual billable tasks, plan requirements and external API fees. Include the time needed to build and maintain it.
Documentation checked on 6 October 2026. Confirm the official shutdown notice and account details before your cutover.
n8n
Best for: technical automation workflows
Consider n8n when your workflow needs custom logic or control over deployment. Self-hosting also requires time for updates, backups, and monitoring.
Operant Solo may earn a commission if you purchase through this link, at no extra cost to you.
You can read this article without subscribing. Want the next tested article?

