Analytics > Latency

The Latency report is crucial for monitoring your bot's responsiveness. It measures the time it takes for the bot to process a user's request and deliver a response. Keeping this time low is essential for maintaining a positive and engaging user experience.

Key Metrics

To measure the performance of our system, we focus on the following key metrics:

• Average Response Time: This metric captures the time elapsed from the moment the user sends a question to the delivery of the last token in the response. It’s worth noting that the time to first token is typically much shorter. For example, even if the total response time is 5 seconds, the first token might appear in just 1–2 seconds, ensuring a responsive experience.

• Shortest Response Time: This represents the quickest response recorded during the period, often observed with simple interactions like greetings.

• Longest Response Time: This metric highlights the slowest response during the period. It is particularly valuable for identifying and addressing edge cases that cause delays, ensuring consistent performance across all interactions.

Average Response Time Evolution

The main graph on the page visualizes the daily fluctuations in the bot's average response time.

The data is broken down by channel (e.g., webchat, simulator), which allows you to easily compare performance across different platforms. This helps you determine if a slowdown is a general issue or specific to one channel.