Build in the sandbox, deploy what is ready
A HubSpot standard sandbox (Enterprise) is a copy of your portal's setup where you build and test changes, then push supported assets to production with Deploy to Production. Test with real-shaped records, include connected assets, reconnect integrations and redo the settings that do not deploy. Deployed workflows arrive turned off, so you switch them on in production yourself.
A sandbox is the gap between "it works on my screen" and "it works in the portal the team relies on." In our CRM and RevOps work on Enterprise portals, structural changes go through a sandbox before they touch production, the same way the HubSpot implementation guide describes building the object model before turning anything on. This page is the sandbox itself: what it copies, what it skips, how to test so the test means something, and how to deploy when you are ready.
What a HubSpot sandbox is (and which plans have it)
A standard sandbox is a separate HubSpot account that copies your production portal's supported assets at creation. You build and test there, then push new assets to production with Deploy to Production. The sandbox has its own account ID, its own record IDs and its own user list. Super Admins switch into it from the account menu, and a banner shows when you are in the sandbox.
Standard sandboxes are available with Enterprise subscriptions: Marketing Hub Enterprise, Sales Hub Enterprise, Service Hub Enterprise, Data Hub Enterprise, Content Hub Enterprise, Smart CRM Enterprise, or Revenue Hub Enterprise (deploy sandbox changes to production, checked 18 Nov 2026). Current Enterprise pricing stays on the HubSpot pricing guide. If you are moving to HubSpot Enterprise or upgrading from a lower tier, the sandbox is part of the HubSpot migration conversation.
HubSpot sunset legacy standard sandboxes on 30 April 2026. The resync feature was removed because it could overwrite in-progress work and break change tracking. To get a fresh copy of production, recreate the sandbox. Deploy to Production only works from a new standard sandbox, not from a legacy one (legacy standard sandboxes sunset, announced 19 Dec 2025). If your account still has a legacy sandbox, create the new one (the sandbox limit was temporarily raised for this), then delete the legacy version.
Teams below Enterprise do not get a sandbox. The closest alternative is the workflow Test feature, available on Professional and Enterprise. It simulates a record's path through the workflow without actually enrolling the record or running any actions (test your workflow, checked 18 Nov 2026). The other fallback is building with the workflow turned off, then enrolling a single test contact manually before enabling it. We describe both in the HubSpot workflows playbook.
What copies into the sandbox and what does not
| Item | Copies at creation? | What to do |
|---|---|---|
| Object schemas, properties, pipelines, forms | Yes. View and edit access settings on properties are not copied | Reconfigure access settings after creation |
| Workflows | Yes, when every asset they depend on is supported. A workflow that relies on an unsupported asset (such as a segment with a lead scoring filter or a feedback survey) fails to copy | Check Failed assets in the Activity log |
| Webhooks and custom coded actions in workflows | Yes, but webhook URLs still point at production endpoints and secrets are not copied | Update URLs and set secrets in the sandbox |
| Records | Optional. Up to 5,000 most recently updated contacts plus up to 100 associated deals, companies and tickets per contact. Event data is not copied | Select the checkbox at creation. Import up to 200,000 records per object afterwards |
| Users and teams | Super Admins, Partner admins with elevated access and users with Developer tools access. Team hierarchy copies, team members do not | Add others manually |
| Emails | Automated marketing emails copy, regular marketing emails do not. Sends only reach contacts whose address matches a sandbox user, over production's shared IPs | Add a sandbox user with the email you want to test |
| Reports and dashboards | No | Recreate the ones a test needs |
| Integrations and connected apps | No | Connect the sandbox version |
| Record IDs | IDs differ between sandbox and production | Use an External record ID property (see testing section below) |
| Notetaker | No | Not supported in sandbox accounts |
Source: production assets eligible to copy to a sandbox and deploy sandbox changes to production, checked 18 Nov 2026.
Creation can take a few hours. When it finishes, check the Activity log under Settings, then Sandboxes. The Failed assets tab shows what broke and why. The Copied assets tab shows what landed.
How to test a change so the test means something
A sandbox test proves nothing if the data inside it looks nothing like production. Three pristine demo contacts with every field filled in will not catch the workflow that breaks on a null lifecycle stage or a contact with no owner.
Start with real-shaped records. If you selected the 5,000 contact option at creation, those are production records with production data quality, which is exactly what you want. If you imported records instead, include the messy ones: contacts with no owner, deals stuck in a stage for months, companies with no domain. Those are the records that trip a workflow.
Use the External record ID pattern. Record IDs differ between sandbox and production, so a workflow that references a specific record by ID will not behave the same way. HubSpot's recommended workaround: create a custom property called something like "External record ID" in production, use a workflow to copy each record's ID into that field, then use that property to upsert records in the sandbox (deploy sandbox changes to production; legacy sandbox sunset FAQ).
Use the workflow Test feature. Even inside a sandbox, Test simulates a record's path without enrolling the record or running any actions. Test criteria checks whether a specific record meets the enrollment triggers and shows a tick or cross per criterion. The full Test simulates the enrollment path and highlights the branch the record would follow. Branch outcomes are predicted from the record's current state, and the record follows the first matching branch. Branches that depend on a previous action's output must be selected manually. Send preview delivers every email on the preview path to your own address, not to the contact's (test your workflow, checked 18 Nov 2026).
Respect the enrollment cap. The sandbox allows 100,000 workflow enrollments per day across all workflows in the account. Enrollments past the cap are dropped (deploy sandbox changes to production). For a large batch test, stagger it or use the Test feature first.
Connect integrations to their sandbox or test instances. A sandbox that calls your production Slack channel, your live n8n webhook or your production API key is not a test. Connect to sandbox versions, staging endpoints or test keys. If you run automations through n8n, point the n8n workflow at the sandbox portal ID and a sandbox private app token.
Before moving to deploy, run through a short checklist:
- The intended enrollment trigger fires on the records it should.
- Records that should be excluded stay out (re-enrollment settings, suppression lists, segment filters).
- Null and missing property values do not cause unexpected branching.
- Re-enrollment behaves as expected when a contact loops back.
- Downstream actions (emails, task creation, integration calls) land where you expect.
Deploy to Production, step by step
When the sandbox build is tested:
- In production, go to Settings, then Sandboxes.
- Click Set up a deploy in the top right.
- Select the asset types you want to deploy. For objects, click Select configuration.
- Select the eligible changes. For objects, click Save selection in the right panel.
- Click Next.
- Review connected assets. These are assets your selected assets depend on. Check "Add connections to my deploy to continue" to include them.
- Resolve any conflicts. HubSpot detects them automatically. If none are found, click Next.
- Enter a deploy name and description, then click Next.
- Review the summary, click Deploy to Production, and confirm.
Source: deploy sandbox changes to production, checked 18 Nov 2026.
A few constraints to know before you start:
- Only new assets deploy. Edits to assets that were copied from production into the sandbox cannot be deployed back. If you need to change an existing workflow, the change still happens by hand in production (assets eligible to be deployed, checked 18 Nov 2026).
- Up to 300 changes per deploy. A large project may need several deploys.
- A started deploy cannot be cancelled. Review the summary carefully.
- Deployed workflows arrive turned off. You switch each one on in production yourself (assets eligible to be deployed).
After the deploy, check the error log under Settings, then Sandboxes. If a deploy fails, the log shows which asset failed and why. Then walk through production:
- Confirm each workflow is present and still turned off.
- Reconfigure the settings that do not deploy (see the next section).
- Turn workflows on one at a time, enrolling a controlled record first.
What you still do by hand
Deploy to Production covers supported asset types, but several settings and categories land in production without their full configuration or do not deploy at all.
Settings that do not deploy:
- Manage access and Sensitive Data settings on custom properties
- Pipeline rules, the Automate tab and the Deal tags tab on pipelines
- Form settings
- Segment settings
- Workflow organizational settings: folders, access restrictions, brand and campaign labels
Source: assets eligible to be deployed, checked 18 Nov 2026.
Edits to production-copied assets. If you changed a workflow that was copied from production into the sandbox, that change cannot be deployed. Only assets created new in the sandbox are eligible.
Integrations, webhooks and custom-coded actions. The sandbox has a different account ID, so private app tokens, webhook URLs and authentication in custom-coded workflow actions all need updating. Reconnect private apps and generate new tokens. Reauthorize public apps. Review every webhook URL and every custom-coded action for hardcoded account IDs or production endpoints (legacy sandbox sunset FAQ). If your automations run through n8n or another external tool, the same applies: point those flows at the right portal. Workflows with webhooks are copied with their production URLs, and secrets are not copied. Update URLs and set secrets in the sandbox before testing (production assets eligible to copy to a sandbox).
Permissions and users. Only Super Admins, Partner admins with elevated access, and users with Developer tools access are copied into the sandbox. Everyone else must be added manually. Permission sets and teams are not among the asset types Deploy to Production supports (assets eligible to be deployed). Team members are not copied into a sandbox (production assets eligible to copy to a sandbox). Manage access and Sensitive Data settings on new custom properties must be reconfigured in production (assets eligible to be deployed). Set those up after the deploy, following the same role design we describe in the user permissions guide.
AI connectors. An AI assistant or MCP connector runs against the account it is connected to. If you connected it in the sandbox, it reads sandbox data. Reconnect it in production after deploy, with the same narrow permission set. We cover setup and the audit check in the HubSpot MCP server guide.
Run it as a release process
A sandbox that sits open for six months with half-finished experiments just adds a second mess next to the first one.
HubSpot lists four sandbox purposes: active development and deployment, integration testing, new user training, and testing new features. Each comes with a different access footprint and a different recreate cadence. Active development sandboxes should be deleted and recreated regularly, such as after a major project. Integration testing sandboxes can last longer but still need periodic recreation. Training and feature-testing sandboxes are recreated as needed (deploy sandbox changes to production).
The process that works in the HubSpot portals we work in:
- One owner per sandbox. That person decides what goes in and what gets deployed. The owner does not need to be a Super Admin, but the person who clicks Deploy to Production does.
- Limited access. Only the people building and testing get sandbox access. Broad access leads to conflicting changes and stale experiments.
- A short change log per deploy. Write down what you deployed, when, and what still needs manual setup in production. This is the same discipline we describe in the data governance guide.
- Deploy small batches. The 300-change limit encourages this anyway. Smaller deploys are easier to verify and easier to troubleshoot.
Sandboxes are especially useful during migrations and structural changes. When you are moving from another CRM, build the HubSpot object model in a sandbox first, test the import, then deploy the structure to the production portal. We cover the migration sequence in the CRM migration playbook and the Salesforce-specific path in the Salesforce to HubSpot guide. When you are adding custom objects, the sandbox is where you test whether the associations, properties and pipeline stages behave as expected before they exist in production. The custom objects guide walks through that decision. For the broader integration architecture, including what connects to what, the CRM integration guide covers the patterns.
Mistakes we see
These come up regularly in the HubSpot portals we work in. None of them are complicated. All of them cost time.
Testing with three clean demo contacts. A sandbox full of perfect data proves the happy path. The workflow that breaks on a null owner, a contact with no lifecycle stage, or a deal with a close date in the past is the one that causes trouble in production. Import records that look like your real data, edge cases included.
Not reading the connected assets list. The deploy requires you to add every form, segment, property and email a workflow uses before it will proceed (deploy sandbox changes to production). Teams tick the checkbox without reading the list and push properties or segments into production they did not plan to. Read the list and note each connected asset in the deploy description.
Leaving integrations pointed at production. If the sandbox webhook URL still calls your production n8n instance or your live Slack channel, the sandbox is writing to production. Copied webhook workflows keep their production URLs until you change them (production assets eligible to copy to a sandbox). Connect sandbox versions, staging endpoints or test keys.
Turning workflows on without a controlled first record. After a deploy, workflows arrive turned off. Turning one on and walking away means the first enrollment is uncontrolled. Turn it on, enroll a single known record, verify the outcome, then leave it running.
Editing a copied asset and expecting to deploy it. Only assets created new in the sandbox can be deployed. If you changed a workflow that was copied from production, that change stays in the sandbox. The production edit has to be done by hand.
Letting a sandbox drift. A sandbox that has not been recreated since the original build no longer matches production. Properties, pipelines and assets change in production over time, and the sandbox falls behind. Recreate on a schedule, tied to the purpose HubSpot recommends for that sandbox type. If you are unsure whether your current sandbox or release process is sound, that is a good question for a second opinion.
FAQ
Which HubSpot plan includes a sandbox?
Enterprise subscriptions on Marketing Hub, Sales Hub, Service Hub, Data Hub, Content Hub, Smart CRM or Revenue Hub. No other tier includes a standard sandbox (deploy sandbox changes to production). For what each Enterprise tier costs, see the HubSpot pricing guide. Teams below Enterprise can use the workflow Test feature on Professional and above.
What happened to legacy sandboxes?
HubSpot sunset legacy standard sandboxes on 30 April 2026 and removed the resync feature. To get a fresh copy of production, create a new standard sandbox. Deploy to Production only works from the new version (legacy sandbox sunset FAQ).
Does a sandbox copy my contacts and deals?
Optionally. At creation you can select a checkbox to copy the 5,000 most recently updated contacts plus up to 100 associated deals, companies and tickets per contact. For more records, import up to 200,000 per object type after creation (deploy sandbox changes to production).
Can I undo a deploy to production?
No. Once a deploy starts, it cannot be cancelled. HubSpot's guide does not describe a way to reverse a deploy (deploy sandbox changes to production). Review the deploy summary carefully before confirming. If something goes wrong, check the error log and fix the affected assets in production by hand.
Do integrations work in a HubSpot sandbox?
Integrations are not connected automatically. You need to connect the sandbox version of each integration, generate new private app tokens, reauthorize public apps, and update any webhook URLs or custom-coded actions that reference the old account ID (deploy sandbox changes to production; legacy sandbox sunset FAQ). Point integrations at sandbox or test endpoints, not at production. For the HubSpot and Salesforce integration specifically, the integration guide covers the data sync layer.
Who can create a sandbox and deploy changes?
A Super Admin. Super Admin permissions are required both to create a standard sandbox and to run Deploy to Production (deploy sandbox changes to production). We cover how to decide who gets Super Admin, and why to keep the list short, in the user permissions guide.
If your HubSpot portal runs live changes without a sandbox or a release process, that is the first thing to fix.