> ## Documentation Index
> Fetch the complete documentation index at: https://braintrust.dev/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Scorer overwriting human review data with identical names

export const plans_0 = "Any"

export const deployments_0 = "Braintrust-hosted"

export const data_plane_version_0 = undefined

export const use_case_0 = "Use case - Online scoring with human review scores configured on the same project"

<Note>
  **Applies to:**

  * Plan - {plans_0}
  * Deployment - {deployments_0}
  * {data_plane_version_0}
  * {use_case_0}
</Note>

## Summary

**Issue:** When an online scorer and a human review column share the same name, the automated scorer overwrites human review data with its own scores.

**Cause:** Scorers and human review columns share the same namespace when writing to scores, causing name collisions.

**Resolution:** Use distinct names for scorers and human review columns to prevent overwriting.

## Resolution Steps

### Immediate fix

#### Step 1: Rename the conflicting score

Change either the scorer name or the human review column name to make them distinct.

#### Step 2: Verify the change

Check that human review data is no longer being overwritten by the automated scorer.

### Prevent future conflicts

#### Step 1: Use naming conventions

Adopt prefixes to distinguish score types (e.g., `auto_accuracy` for automated scorers, `human_quality` for reviews).

#### Step 2: Document naming standards

Create team guidelines for scorer and human review naming to avoid namespace collisions.

## Technical Details

### Why this happens

Both automated scorers and human review columns write to the same score storage namespace. When names match, the scorer output overwrites human review data because the system uses the score name as the storage key without distinguishing between automated and manual sources.

### Affected scenarios

* Online scoring rules that write to scores
* Human review columns configured with the same name as an active scorer
* Production logs where both automated and human scoring are enabled

### Detection difficulty

This issue provides no warning or error message. Teams often assume someone manually scored the data, making the root cause difficult to diagnose without comparing scorer names to human review column names.

## Best Practices

### Naming conventions

* **Automated scorers:** Use prefixes like `auto_`, `llm_`, or `scorer_`
* **Human reviews:** Use prefixes like `human_`, `review_`, or `manual_`
* **Specific criteria:** Include the evaluation criterion in the name (e.g., `auto_factuality`, `human_helpfulness`)