Testing Workflows

Test and debug your workflows to ensure they work correctly before activating them for production use.

Time to complete: 15 minutes

---

Why Test Workflows?

Testing catches issues before they affect real users:

Without Testing	With Testing
Errors in production	Issues found early
Broken customer experiences	Smooth automations
Difficult debugging	Clear error visibility
Unknown failures	Confident deployments

---

Testing Methods

Method	When to Use
Manual Test	Quick validation with sample data
Node Testing	Test individual nodes
Execution Replay	Analyze past executions
Debug Mode	Step-by-step inspection

---

Manual Testing

Running a Test

1. Open your workflow in the builder
2. Click Test in the toolbar
3. Enter test input data:

{
  "email": "test@example.com",
  "name": "John Doe",
  "message": "Hello, I need help"
}

4. Click Run Test
5. Watch execution in real-time

What to Observe

Indicator	Meaning
Green highlight	Node executed successfully
Red highlight	Node failed
Blue highlight	Currently executing
Gray	Not yet executed

Viewing Node Results

Click any node during or after test to see:

• Input: Data received by node
• Output: Data produced by node
• Duration: How long it took
• Error: Any error message

---

Test Input Formats

Webhook Trigger Input

{
  "payload": {
    "customer_id": "cust_123",
    "order_id": "ord_456",
    "items": [
      {"name": "Product A", "qty": 2},
      {"name": "Product B", "qty": 1}
    ]
  },
  "headers": {
    "Content-Type": "application/json",
    "X-API-Key": "test-key"
  }
}

Database Trigger Input

{
  "entity_id": "550e8400-e29b-41d4-a716-446655440000",
  "entity_type": "contacts",
  "operation": "INSERT",
  "new_data": {
    "first_name": "Jane",
    "last_name": "Smith",
    "email": "jane@example.com",
    "lead_score": 85
  },
  "old_data": null,
  "changed_fields": ["first_name", "last_name", "email", "lead_score"]
}

User Message Trigger Input

{
  "message": "What are your business hours?",
  "user_id": "user_123",
  "conversation_id": "conv_456",
  "context": {
    "previous_messages": ["Hi", "I have a question"]
  }
}

Scheduler Trigger Input

{
  "triggered_at": "2024-03-15T09:00:00Z",
  "schedule_type": "daily",
  "iteration": 1
}

---

Node-Level Testing

Run Single Node

1. Click on a specific node
2. Click Run Node in properties panel
3. Provide input data for that node
4. See output immediately

When to Use

• Testing node configuration
• Debugging specific logic
• Validating transformations
• Checking integrations

Example: Test Transform Node

Input:

{
  "raw_data": {
    "firstName": "John",
    "lastName": "Doe",
    "email": "JOHN@EXAMPLE.COM"
  }
}

Transform Code:

return {
  full_name: `${input.raw_data.firstName} ${input.raw_data.lastName}`,
  email: input.raw_data.email.toLowerCase(),
  created_at: new Date().toISOString()
};

Expected Output:

{
  "full_name": "John Doe",
  "email": "john@example.com",
  "created_at": "2024-03-15T10:30:00.000Z"
}

---

Execution History

Viewing Past Executions

1. Go to Automation → Workflows
2. Click your workflow
3. Open Executions tab
4. See all workflow runs

Execution Details

Column	Description
ID	Unique execution identifier
Status	Completed, Failed, Running
Trigger	What started the execution
Started	When it began
Duration	How long it took

Replaying Executions

1. Click any execution
2. Workflow loads with that execution's data
3. See exact input/output for each node
4. Identify where issues occurred

---

Debug Mode

Enabling Debug

Some nodes support debug mode:

1. Click the node
2. Enable Debug Mode toggle
3. Run test
4. See detailed execution info

Debug Information

Debug mode shows:

• Variable resolution steps
• Condition evaluation details
• API request/response logs
• Timing breakdowns
• Memory usage

Example Debug Output

{
  "debug": {
    "prompt_length": 245,
    "system_prompt_length": 120,
    "input_keys": ["message", "contact", "context"],
    "variable_resolution": {
      "{{contact.name}}": "John Doe",
      "{{message}}": "Hello"
    },
    "execution_time_ms": 1234
  }
}

---

Common Testing Scenarios

Testing Conditions

Test True Path:

{
  "lead_score": 90,
  "company_size": "Enterprise"
}

Test False Path:

{
  "lead_score": 30,
  "company_size": "Startup"
}

Testing Loops

{
  "items": [
    {"id": 1, "name": "Item A"},
    {"id": 2, "name": "Item B"},
    {"id": 3, "name": "Item C"}
  ]
}

Verify:

• All items processed
• Results collected correctly
• Error handling works (add invalid item)

Testing Error Paths

{
  "email": null,
  "required_field": ""
}

Verify:

• Error conditions trigger
• Error handlers execute
• Appropriate responses sent

Testing AI Nodes

{
  "message": "What is your return policy?",
  "kb_id": "your-kb-id"
}

Check:

• KB search returns results
• AI generates relevant answer
• Response format is correct

---

Test Data Best Practices

Use Realistic Data

✅ Real-looking names and emails
✅ Valid data formats
✅ Typical data volumes
✅ Edge cases included

❌ Don't use "test" everywhere
❌ Don't skip optional fields
❌ Don't ignore special characters

Test Edge Cases

Edge Case	Test Data
Empty values	null, "", []
Long strings	1000+ character text
Special characters	Unicode, emojis, HTML
Large numbers	Integers, decimals, negatives
Date boundaries	Past, future, midnight

Document Test Cases

Create a test plan:

## Test Cases for Welcome Email Workflow

1. **Happy Path**
   - Valid email and name
   - Expected: Email sent successfully

2. **Missing Email**
   - No email provided
   - Expected: Error response

3. **Invalid Email Format**
   - Email: "not-an-email"
   - Expected: Validation error

4. **Special Characters in Name**
   - Name: "José García-López"
   - Expected: Handles correctly

---

Troubleshooting Common Issues

Variable Not Found

Symptom: {{variable}} not found in output

Causes:

1. Typo in variable name
2. Node hasn't executed yet
3. Path is incorrect

Solutions:

1. Check node alias matches
2. Verify execution order
3. Use debug mode to see available variables

Condition Always False

Symptom: Always takes false path

Causes:

1. Type mismatch (string vs number)
2. Case sensitivity
3. Empty value comparison

Solutions:

1. Check actual value in debug
2. Use correct operator
3. Handle null/undefined explicitly

API Node Fails

Symptom: External API call fails

Causes:

1. Invalid credentials
2. Rate limiting
3. Incorrect URL/parameters

Solutions:

1. Test integration separately
2. Check API logs
3. Verify request format

Loop Not Executing

Symptom: Loop shows 0 iterations

Causes:

1. Array is empty
2. Array path is wrong
3. Data is not an array

Solutions:

1. Print array value before loop
2. Check array_source path
3. Wrap in array if single item

---

Performance Testing

Monitor Execution Time

Watch for slow nodes:

• LLM calls: Expect 1-5 seconds
• API requests: Depends on external service
• Database queries: Should be < 1 second
• Transforms: Should be milliseconds

Identify Bottlenecks

1. Run workflow multiple times
2. Note duration per node
3. Optimize slowest nodes

Optimization Tips

Issue	Solution
Slow LLM	Use faster model
Many API calls	Batch if possible
Large loops	Add concurrency
Complex transforms	Simplify logic

---

Pre-Activation Checklist

Before activating a workflow, verify:

☐ Happy path works correctly
☐ Error paths handle failures gracefully
☐ All conditions tested (true and false)
☐ Edge cases don't break workflow
☐ Performance is acceptable
☐ Outputs are correctly formatted
☐ Integrations are connected
☐ Variables resolve properly
☐ No sensitive data in logs
☐ Proper error messages for users

---

Next Steps

With testing mastered, continue with:

1. Error Handling - Build resilient workflows

---

Quick Reference

Test Input Structure

{
  "trigger_field": "value",
  "nested": {
    "field": "value"
  },
  "array": [{"item": 1}]
}

Execution Status

Status	Meaning
Pending	Waiting to start
Running	Currently executing
Completed	Finished successfully
Failed	Error occurred

Debug Checklist

☐ Check input data format
☐ Verify variable paths
☐ Review node configuration
☐ Enable debug mode
☐ Check execution logs
☐ Test each node individually