---
title: "What each metric actually counts"
description: "Exact definitions for the report numbers that are easy to misread, using the names the Reports pages show."
url: https://help.gleap.ai/fr/articles/308-what-each-metric-actually-counts
collection: "Reports"
updated: 2026-11-19
language: fr
alternates: [en]
---

# What each metric actually counts

Exact definitions for the report numbers that are easy to misread, using the names the Reports pages show.

Most report numbers mean what their name suggests. These are the ones that don't, or where the definition changes how you should read them. The names below are exactly what the report tiles show, and hovering a tile's info icon in the dashboard shows the same definition.

### Volume

-   **New tickets**: tickets created in the period, each counted once.
    
-   **Closed tickets**: tickets closed in the period. A ticket closed, reopened and closed again counts once.
    
-   **Open tickets** (on **Live**): tickets open right now, across all boards. It's a live count, not a total for the period.
    
-   **New tickets by tag**: new tickets split by tag (the 25 most used). Tickets without a tag aren't shown in that chart.
    

### Work done

-   **Replies sent**: replies your team sent to customers across all tickets. One long conversation counts several times; internal notes don't count.
    
-   **Tickets worked on**: tickets that saw any activity from your team in the period, whether or not they were created in it. This is the fairer measure of a person's day.
    
-   **Tickets without a reply**: the share of tickets created in the period that never received a reply. Lower is better.
    

> **Warning:** Tickets without a reply is the number most worth watching. It counts people who wrote to you and got nothing at all, which no satisfaction score will ever tell you, because those customers were never asked.

### Response times

-   **First response time**: from the customer's last message (or the hand-off to your team) to the first reply from a teammate. Kai Support and workflow messages don't count.
    
-   **Reply time**: how long customers waited for any reply, not only the first one.
    
-   **Time to first assignment**: from creation until a teammate was assigned.
    
-   **Time to close**: from creation to the first time the ticket was closed.
    

The **Display** menu switches between median and average. The median is the better default: one ticket left for a week drags the average up for everyone. **Respect working hours** measures times inside office hours only.

### Satisfaction

-   **Average rating**: the average rating customers gave conversations in the period, as a percentage of the best possible rating.
    
-   **Ratings received**: conversations that got a rating. A conversation rated twice counts once.
    
-   **Rating coverage**: the share of conversations created in the period that ended up rated.
    

Always read the average against the coverage. A high score from six percent of customers describes the six percent who felt strongly enough to answer, and those are rarely the quietly disappointed ones.

### Service levels

**SLA compliance** is the share of tickets with an SLA that got their first response within it. **Breaches** counts tickets whose first response came after the SLA ran out.

Because the target is yours to choose, compliance says as much about the target as about the team. Comparing it across periods only means something if the target didn't move.

### Kai Support

-   **Questions asked**: questions customers asked Kai Support.
    
-   **Deflection rate**: the share of conversations Kai Support resolved without a teammate stepping in.
    
-   **Handed to the team**: the share of conversations with Kai Support that ended up with a teammate. It's the same population counted the other way round: if Kai Support took part in 100 conversations and 20 ended up with your team, deflection is 80%.
    
-   **Answer rate**: the share of questions Kai Support could answer. The rest usually had nothing matching in your knowledge base.
    
-   **Positive feedback**: the share of thumbs-up among all feedback on Kai Support's answers.
    

None of these should be read without the others. A high deflection rate with a low answer rate can mean customers gave up instead of being helped. The AI quality review judges each conversation's outcome directly; see [How the AI quality review scores a conversation](https://help.gleap.ai/en/articles/16751958-how-the-ai-quality-review-scores-a-conversation.md).

### Related

-   [What Gleap measures](https://help.gleap.ai/en/articles/307-what-gleap-measures.md)
    
-   [How the AI quality review scores a conversation](https://help.gleap.ai/en/articles/16751958-how-the-ai-quality-review-scores-a-conversation.md)
    
-   [Get the raw numbers out](https://help.gleap.ai/en/articles/309-get-the-raw-numbers-out.md)
    
-   [Measure how well Kai is doing](https://help.gleap.ai/en/articles/244-measure-how-well-kai-is-doing.md)
