Use Continue on Fail for Graceful Degradation

When processing a batch of items, one failure should not stop the entire batch. Enable Continue on Fail on nodes where individual item failures are acceptable. The workflow continues processing remaining items, and you can handle the failed items separately.

Real-world example: A workflow enriches 200 leads by calling a company data API for each one. 5 of the 200 companies are not found (404 error). Without Continue on Fail, the workflow stops at the first 404 and the remaining 195 leads are never processed.

# Node Settings on the "Enrich Company Data" HTTP Request node

Settings:
  Continue On Fail: true
  # When a single item fails, the node outputs the error

  # as part of the item data instead of stopping the workflow

```text
The output of a node with Continue on Fail includes error information:

```json
// Successful item output
{
  "lead_id": "L001",
  "company": "Acme Corp",
  "enriched_data": {
    "industry": "Technology",
    "employees": 500,
    "revenue": "$50M"
  }
}

// Failed item output (Continue on Fail adds error fields)
{
  "lead_id": "L002",
  "company": "Unknown LLC",
  "error": {
    "message": "Request failed with status code 404",
    "description": "Company not found"
  }
}
```text
Separate successful and failed items downstream:

```javascript
// Code node: "Split Successes and Failures"
const items = $input.all();

const successes = [];
const failures = [];

for (const item of items) {
  if (item.json.error) {
    failures.push({
      json: {
        ...item.json,
        failedAt: new Date().toISOString(),
        retryable: item.json.error.message?.includes('429')
          || item.json.error.message?.includes('500')
      }
    });
  } else {
    successes.push(item);
  }
}

// Output 0: successes, Output 1: failures
// Use this node's "Output" setting to split into two outputs
return [successes, failures];
```text
> **Note: Continue on Fail vs. Error Workflow**
>
> These serve different purposes. **Continue on Fail** handles expected, item-level failures within a running workflow. **Error Workflow** handles unexpected, workflow-level crashes. Use both together: Continue on Fail for batch processing, Error Workflow as the safety net.

This pattern ensures partial failures do not block your entire pipeline. Process what you can, log what you cannot, and review failures separately.

**Related:** [Use "Pin Data" to Freeze Node Output](../testing-and-debugging/01-use-pin-data-to-freeze-node-output.md) | [Break Large Workflows into Sub-Workflows](../workflow-architecture/01-break-large-workflows-into-sub-workflows.md)