Working with Blueshift sandbox accounts

The sandbox environment in Blueshift is designed for safe testing and validation of integrations. No real messages are sent, and messaging quotas are not consumed.

Why use a sandbox?

Use a sandbox account to:

• Preview campaigns and safely personalize content
• Validate segmentation and trigger logic
• Explore API workflows without impacting production data
• Test CloudApp integrations and payloads

Sandbox limitations

Message delivery

• Campaign execution is disabled in sandbox accounts (account_mode = sandbox).
• No messages are sent — including Email, SMS, Push, or CloudApp.
• Test sends are allowed for all templates, allowing you to preview content and personalization. These do not consume quota.

Error codes

When attempting to trigger campaigns in sandbox accounts via API, you'll receive:

• HTTP 422 Unprocessable Entity
• Error code 101: "sandbox.account.com is a sandbox account"

This confirms campaign execution is properly blocked in sandbox mode.

Quota usage

Campaign messaging is disabled in sandbox accounts, so there is no impact on your annual messaging quota.

User uploads and segmentation

• You can upload test users and build segments exactly as you would in a production account.
• Segments and triggers can be used for previewing journeys, but will not result in message sends.

Load testing

• The sandbox is not intended for high-volume load testing.
• There is no infrastructure allocation for MTA scaling, API rate guarantees, or throughput validation.
• If you need to test at scale, you can contact Blueshift Support to coordinate a suitable test plan.

Recommendations and data-dependent features

Sandbox accounts typically have limited user engagement data and catalog information compared to production environments. This affects features that rely on substantial data:

• Recommendation engines: May not function optimally without sufficient user behavior data, purchase history, and catalog interactions
• Photon V2 recipes: Some recommendation recipes require specific data streams and user engagement patterns that may not be available in sandbox
• Predictive features: Machine learning models may have limited effectiveness without representative user data volumes

For testing recommendations, consider using simplified approaches like random catalog item selection or manually configured product sets rather than behavior-based recommendations.

Best practices

• Use test users and small segments to validate campaign setup.
• Utilize CloudApp integrations to inspect webhook payloads and validate endpoint behavior.
• Please don't use sandbox accounts for real-time load or deliverability testing.