Reranking improves search relevance by incorporating user feedback, popularity metrics, and custom weights to dynamically adjust result ordering.

​
Ranking Types

Feedback Signals

  • Click interactions
  • View duration
  • Skip actions
  • Custom events

Popularity Metrics

  • Historical engagement
  • Recent interactions
  • Time decay
  • Confidence scores

​
Configuration

​
Interaction Recording

​
Common Use Cases

await mixpeek.trackInteraction({
  feature_id: "prod_123",
  interaction_type: "click",
  position: 2,
  metadata: {
    device: "mobile",
    duration_ms: 45000
  }
});

​
Implementation Flow

​
Best Practices

1

Configure Weights

Start with balanced weights and adjust based on data

2

Monitor Quality

Filter spam and validate interaction data

3

Handle Edge Cases

Account for new items and missing signals

4

Test Changes

A/B test different reranking configurations

​
Limitations

Be aware of these technical constraints:

  • Minimum interaction threshold required
  • Historical data retention limits
  • Processing delay for new signals
  • Resource impact of complex configurations

For detailed implementation examples, see the Interactions API Reference.

Was this page helpful?

Suggest editsRaise issue
Grouping