> ## Documentation Index
> Fetch the complete documentation index at: https://docs.burn0.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Feature Attribution

> Track costs per feature with track()

# track()

Attribute API costs to specific features in your app. This lets you see "my chat feature costs \$8/day" instead of just "OpenAI costs \$8/day".

## Usage

```typescript theme={null}
import { track } from '@burn0/burn0'

// Wrap any async function that makes API calls
await track('chat', { userId: 'user123' }, async () => {
  const response = await openai.chat.completions.create({
    model: 'gpt-4o-mini',
    messages: [{ role: 'user', content: 'Hello' }],
  })
  return response
})
```

All API calls made inside the `track()` callback are tagged with the feature name and metadata.

## Parameters

```typescript theme={null}
track(feature: string, metadata: Record<string, string | number | boolean>, fn: () => Promise<void>): Promise<void>
```

| Parameter  | Type             | Description                                               |
| ---------- | ---------------- | --------------------------------------------------------- |
| `feature`  | `string`         | Feature name (e.g., `'chat'`, `'summarize'`, `'search'`)  |
| `metadata` | `object`         | Key-value pairs (e.g., `{ userId: 'u123', tier: 'pro' }`) |
| `fn`       | `async function` | The function that makes API calls                         |

## Alternative: startSpan

For cases where a callback wrapper doesn't fit, use `startSpan`:

```typescript theme={null}
import { startSpan } from '@burn0/burn0'

const span = startSpan('chat', { userId: 'user123' })

// ... make API calls ...

span.end()
```

<Warning>
  Always call `span.end()` when done. Forgetting to end a span means subsequent API calls may be attributed to the wrong feature.
</Warning>

## Examples

### Per-user cost tracking

```typescript theme={null}
app.post('/api/chat', async (req, res) => {
  const { userId, message } = req.body

  await track('chat', { userId }, async () => {
    const response = await openai.chat.completions.create({
      model: 'gpt-4o-mini',
      messages: [{ role: 'user', content: message }],
    })
    res.json({ response: response.choices[0].message.content })
  })
})
```

### Per-feature breakdown

```typescript theme={null}
// Summarize feature
await track('summarize', { documentId: doc.id }, async () => {
  await anthropic.messages.create({ ... })
})

// Search feature
await track('search', { query: q }, async () => {
  await openai.embeddings.create({ ... })
  await pinecone.query({ ... })
})
```

## Viewing tracked data

Feature and metadata are stored with each event and visible in:

* The dashboard event detail view
* `burn0 report` (coming soon: per-feature breakdown)