App usage
The App usage tab provides operational metrics across configurable time ranges (Last Week, Last Month, Last 3 Months). Each chart supports channel filtering so you can view data for specific integration channels (App, MCP, Widget, Slack) independently or combined.
- Usage trends — query volume over time broken down by channel. Use this to understand adoption patterns and peak usage periods.
- Top users — the most active users ranked by query count, filterable by channel.
- LLM latency — average LLM response time (in milliseconds) over time, helping you spot performance regressions or improvements after model changes.
- Answer success rate — a stacked chart showing answered vs. unanswered queries over time. A high unanswered rate signals gaps in your indexed content.
- User ratings — thumbs-up vs. thumbs-down trends over time, giving you a quick pulse on end-user satisfaction.
- Cache hit rate — cache hits vs. misses over time. A rising hit rate means the semantic cache is serving repeat or similar questions efficiently.
FAQ
What does the App usage tab show?
The App usage tab provides operational metrics across configurable time ranges: Last Week, Last Month, and Last 3 Months. Each chart supports channel filtering so you can view data for specific integration channels (App, MCP, Widget, Slack) independently or combined.
How do I use Usage trends in the App usage tab?
Usage trends shows query volume over time broken down by channel. Use it to understand adoption patterns and identify peak usage periods.
What does the Top users chart measure?
Top users lists the most active users ranked by query count. It can be filtered by channel to see activity for specific integration channels.
What does the Answer success rate chart tell me, and what does a high unanswered rate mean?
Answer success rate is a stacked chart showing answered versus unanswered queries over time. A high unanswered rate signals gaps in your indexed content.
What does the Cache hit rate chart indicate?
Cache hit rate shows cache hits versus misses over time. A rising hit rate means the semantic cache is serving repeat or similar questions efficiently.