AivahHelp CenterGo to Aivah
Insights3 min read

Understand Insights

Set employee and date filters, distinguish usage metrics from sampled records, and interpret review signals.

By the end of this guideConfirm the visible scope and use the overview caveats before drawing conclusions from Insights.

Who this guide is for

Workspace owners, team leads, and reviewers responsible for improving an AI employee

Last reviewed 13 September 2026Some workflows still need verification
On this page
  1. Before you begin
  2. Steps
  3. 1. Open Insights
  4. 2. Choose the employee and date range
  5. 3. Read Activity overview with its scope
  6. 4. Inspect the review signals
  7. 5. Check the relevant records
  8. Expected result
  9. Status meanings
  10. Usage credit impact
  11. Information stored or shared
  12. Limits and permanent actions
  13. Common problems and recovery
  14. Next step

Insights combines usage information with summaries of recent records. Start by checking the employee, date range, and metric explanation before using a number to make a decision.

Before you begin

Choose the employee or all employees, the period, and the question you want to investigate. Use the same scope when comparing periods. Confirm that you may review any conversation or personal information involved.

Steps

1. Open Insights

Choose Insights from the main navigation. The tabs are Insights, Chat history, Leads, Quiz, and Call history.

2. Choose the employee and date range

Use the employee selector and Today, Yesterday, 7 days, 30 days, or Custom. Custom opens From and To fields with an apply action.

After any selection, read the range actually displayed on the page. If it does not match the dates you entered, do not use the results as that custom period. Record the mismatch and use only a range you can confirm.

3. Read Activity overview with its scope

The page explains two different sources for the overview:

  • Unique users and Credits used come from usage analytics for the selected range.
  • Other overview signals use up to 50 recent conversations, leads, quiz attempts, and calls in that range.

Sentiment and intent are keyword-based estimates intended to help you choose what to review. They are not direct measures of a person's feelings or verified business outcomes.

Insights with employee and date filters, five category tabs, zero activity and a disabled CSV export.View full screenshot ↗
Read the selected range and metric definitions before interpreting results. This example has no reported activity.

4. Inspect the review signals

Use Conversation topics, Activity summary, Review queue, and Suggested review to choose examples to investigate. “Nothing needs review in this sample” describes the loaded sample; it does not mean all activity in the account is problem-free.

5. Check the relevant records

Open Chat history, Leads, Quiz, or Call history. Compare any available records with the question you are investigating. Empty detail tabs do not establish that all activity or every private Chat conversation is included in the selected view.

Expected result

You can state the selected employee and visible date range, distinguish usage metrics from sampled signals, and identify which underlying records are needed to support a conclusion.

Status meanings

An empty view applies to its current scope. A review suggestion is a prompt to inspect examples. Keyword-based estimates and sampled counts should not be described as complete account totals.

Usage credit impact

Credits used shows usage analytics for the selected range. This guide does not establish per-action rates, refresh timing, or the exact inclusion of every activity type.

Information stored or shared

Insights can display activity and personal information from conversations, leads, quizzes, and calls. Keep review focused on the question and protect any records or screenshots you share.

Limits and permanent actions

The up-to-50-record limit applies to the sampled overview signals, not a published definition of every detail table or export. Time-zone behavior, ingestion delay, private Chat inclusion, and export completeness must not be inferred from an empty screen. Overview labels are not guarantees of a business outcome.

Common problems and recovery

ProblemWhat to check
A total looks unexpectedConfirm the employee and displayed date range, then read the metric's usage or sample basis.
Custom dates do not applyRecord the entered and displayed dates; do not describe the data as the requested period.
A private Chat conversation is missingPreserve its time and employee details for support; do not infer whether it is included or delayed.
A sentiment or intent label seems wrongReview the underlying text; the label is an estimate.

Next step

Review result tabs and export availability before using records outside Aivah.