Skip to content
Analytics & improvement

Read analytics without mixing scopes

Use the right period and agent scope before interpreting trends.

Where to find itAnalytics / agent analytics
In this article

Before you begin

Open Analytics or the relevant agent’s analytics/insights tab. Select a useful period and confirm which workspace or agent the report covers.

Example workspace · Demo data. These are illustrative demo values; compare the same scope and reporting period. Select the image to enlarge it.
Example analytics overview cards with usage and conversation counts

Example workspace · Demo data. These are illustrative demo values; compare the same scope and reporting period. Select the image to enlarge it.

Step by step

  1. 1

    Choose a useful scope

    Check whether the dashboard covers the workspace or one agent, then choose a period that matches your question.

  2. 2

    Read metric labels

    Distinguish conversations, messages, usage and human support. They measure different events and should not be compared as interchangeable counts.

  3. 3

    Inspect real examples

    Open representative conversations behind a spike, failure or repeated topic. Compare the answer with its source and workflow state.

  4. 4

    Change one cause at a time

    Update a source, prompt or routing condition, then compare a similar period and workload rather than attributing every change to the model.

Measure what visitors ask and where the assistant needs help.

Workspace Analytics gives a cross-bot view; each bot has an analytics page for its own activity. Use the dashboard to see volume and patterns, then inspect a conversation before changing the knowledge or prompt.

  • Start with the trend: Check overall volume and the time range so you can distinguish a genuine quality issue from a temporary traffic change.
  • Find recurring gaps: Use conversation insights and message context to identify questions the assistant could not ground well.
  • Improve the right layer: Add or update knowledge for missing facts, adjust the prompt for behaviour, or add a handoff rule for requests that should always go to a person.
  • Export when needed: Use the analytics export endpoint for reporting or deeper analysis outside Corthex.

Good to know

A short time window or a handful of conversations can be misleading. Analytics describe observed activity, not guaranteed customer satisfaction.

Check the result

Compare the same scope and period before drawing conclusions. Investigate real conversations and sources alongside aggregate metrics.

Questions & troubleshooting

Two screens show different counts

Compare period, timezone, agent/workspace scope and event definition before treating it as data loss.

Was this article helpful?

Rate this guide on this device. Ratings are not sent to Corthex.