# Statistics Metrics — Reference

The Statistics module provides detailed metrics for assessing contact center work efficiency. You can analyze operator work, FastLine and Copilot efficiency, client satisfaction, and other indicators.

---

## Metric Groups

Metrics are organized by dashboards and groups:
- **Dialogues** — basic dialogue information
- **Requests** — general request statistics
- **Operators** — operator work metrics
- **FastLine** — AI bot efficiency
- **Copilot** — AI assistant usage for operators
- **CSAT** — client satisfaction
- **FCR** — first contact resolution
- **FRT** — first response time (SLA metrics)
- **Users** — user statistics
- **Topics** — metrics by dialogue topics

---

## Dialogue Metrics

### Basic Information

**Dialog ID and channel** — unique dialogue identifier and communication channel, link to chat room.

**Channel** — communication channel through which the dialogue is conducted (Telegram, Viber, WhatsApp, etc.).

**Source platform (Web Widget channel)** — where the chat was opened: web browser or embedded chat in a mobile app (iOS/Android). The widget sets this automatically or you can pass it from your site; it appears in reports and dashboards (including Metabase). See [What is ConnectiveOne Widget](/en/channels/explanation/what-is-widget.md).

**Source site (Web Widget channel)** — hostname of the page where the client opened the widget (for example, `shop.example.com`). The widget detects it automatically from the embedding page URL; to override, pass `source_hostname` in `params_from_site` at initialization (your value takes priority). Stored in dialogue statistics so you can split traffic when the same widget is installed on multiple sites.

**Bot** — bot name to which the client wrote.

**Outgoing** — indicator that specifies whether the dialogue is outgoing (initiated by operator).

**Dialogue opening date and time** — date and time of dialogue creation.

**Dialogue closing date and time** — date and time of dialogue completion by operator, auto-closure, or closure by client.

**Dialogue duration, sec** — total dialogue duration in seconds, calculated from creation to closure.

### Statuses and Topics

**Topic** — dialogue topic, including "No topic" variant, with which the dialogue was closed.

**Status** — dialogue status with which it was closed.

**Tags** — tags assigned to the dialogue at the time of closure.

**Skill group** — list of skill groups determined by the dialogue topic assigned to the dialogue at the time of closure.

**Operator comment** — comment left by the operator when closing the dialogue.

### Time and Speed

**Operator response time, sec** — time in seconds until the operator's first response for this dialogue.

**Operator connection through, sec** — time in seconds until operator connection to the dialogue. Calculated between manual connection date-time and chat creation date. 0 — if the system auto-connected the operator.

**Average operator response speed within dialogue, sec** — average operator response time within the dialogue in seconds.

**Time in queue, sec** — time between dialogue creation and auto-connection of operator to it.

**Time in queue** — time the dialogue spent in queue.

### Additional Information

**Additional operators connected to dialogue** — indicator that specifies additional operator connection to the dialogue. Also set during dialogue transfer to another operator.

**Completed by timeout** — indicator that specifies that the dialogue was completed automatically by timeout.

**chat_id** — Chat_id of the user who wrote.

**Repeat request** — indicator that specifies that the user contacted again.

---

## Request Metrics

**Number of requests (dialogues)** — number of created dialogues.

**Number of incoming** — total number of incoming chats for the reporting period.

**Number of outgoing** — total number of outgoing chats for the reporting period.

**Total number of chats with response** — total number of chats with response for the reporting period.

**Number of dialogues closed automatically by timeout** — number of dialogues closed automatically by timeout specified in general settings or in the operator connection block.

### Duration and reaction time (Appeals tab, native statistics)

On the **Appeals** tab, **Duration indicators** shows average and total dialogue duration, **time in queue**, and **closed by inactivity**.

The **Reaction time** block has three separate KPIs:

**Average operator reaction time** — average seconds **until the operator joins** the dialogue (queue / assignment wait). Can be high over long periods when many dialogues waited in queue.

**Average first reaction time** — average time from operator **join** to their **first reply** to the client. Use for SLA “first response after taking the chat” when connection time skews the picture.

**Dialogs with reaction time anomaly (>1 day)** — count of dialogues where time until join exceeded one day. Helps separate rare outliers from typical load.

See [How to read duration and reaction metrics](/en/statistics/how-to/view-appeals-duration-and-reaction-metrics.md).

---

## Operator Metrics

### Dialogue Acceptance

**Number of accepted dialogues** — operator connected first to the dialogue themselves or through auto-distribution, or by any other method. On the **Operators** tab this metric appears in the **Accepted** column.

> **Cross-tab consistency:** if you sum **Accepted** across all operators for the same period and filters, the result matches **Accepted by operators** in the **Summary for the period (all operators)** block on the **Service quality** tab. Each closed dialogue is counted once.

**Number of unaccepted (missed) dialogues** — operator was online when the dialogue without operators closed (through timeout, closed by client themselves, mass closure, etc.).

**Percentage of accepted dialogues** — percentage of dialogues the operator accepted from total available.

**Number of accepted dialogues in target 5min** — number of dialogues accepted within 5 minutes.

**Percentage of accepted dialogues in target 5 min** — percentage of dialogues accepted within 5 minutes.

### Transfer and Addition

**Number of dialogues transferred by operator** — number of dialogues where the operator was first, but overall there was more than one operator in the dialogue (this indicator also includes operator addition).

**Number of dialogues to which operator was added** — number of dialogues where the operator was connected not first, but second, third, and so on.

**Number of dialogues to which second operator was connected** — number of dialogues to which a second operator was connected.

**Total number of dialogues in which operator participated** — either dialogue initially taken by operator, or one transferred to them, or dialogue to which operator was added through joining, or one to which they joined themselves.

### Time and Productivity

**Online time** — number of operator hours worked (when operator was online) for the reporting period.

**Chats per hour** — average number of processed chats per operator per hour. Closed chats for the entire day divided by number of working hours.

**Average operator response (connection) time in seconds** — time it took operator to join the dialogue.

**Average dialogue duration min:sec** — this is the time a specific operator spent in chat. Calculated from connection date-time to operator disconnection within each dialogue to which the operator connected.

**Average operator response speed within dialogue** — time between client and operator messages, recorded when dialogue is completed. Calculated as average of (operator response date-time to last client message minus last client message date-time) across the entire dialogue.

**Total dialogue duration** — total time of operator participation in all dialogues. Participation — difference between connection and disconnection, divided by total sum of chats.

---

## FastLine Metrics

### FastLine Efficiency

**Number of requests (dialogues)** — number of created dialogues with FastLine.

**Closed FastLine without operator** — number of dialogues closed by FastLine without operator connection (including dialogues closed by timeout).

**Percentage of dialogues closed FastLine without operator** — percentage of dialogues closed by FastLine without operator connection (including dialogues closed by timeout).

**Number of dialogues to which operator was connected** — number of FastLine dialogues to which an operator was connected.

**Percentage of dialogues to which operator was connected** — percentage of FastLine dialogues to which an operator was connected.

### Time Saved

**Time saved thanks to FastLine** — time operators saved thanks to FastLine independently completing dialogues without their participation (including dialogues closed by timeout).

**Average dialogue closure time by operator without FastLine** — time that on average goes to operator for closing a dialogue without FastLine help.

**Time saved until operator connection** — time during which FastLine worked with dialogues before transfer to operator.

### Token Usage

**Total LLM tokens used** — total number of tokens used by FastLine for the reporting period.

**Average number of LLM tokens per dialogue** — average number of tokens used by FastLine per dialogue for the reporting period.

### Dialogue Duration

**Average dialogue duration** — this is the time FastLine spent in chat. Calculated from FastLine connection to its disconnection by timeout or dialogue redirection to operator.

**Total dialogue duration** — total time of FastLine participation in all dialogues. Participation — difference between connection and disconnection (including closure by timeout).

### AI Answers (Fast Line Pro)

The **Fastline** tab includes a section that counts **successful AI generations** via Fast Line Pro (Dify integration), Copilot, AI QA, and **knowledge base integration plugin** requests (for example, AI Vision during automated KB updates). One successful call = one AI Answer. Failed attempts are not counted.

| Metric | Meaning |
|--------|---------|
| **Total AI Answers provided** | All sources combined: bot responses, Copilot hints, AI QA, and KB integration plugin |
| **Answers provided to clients** | Fast Line Pro AI responses in client dialogues (excluding Copilot and KB integration plugin) |
| **Copilot hints** | Successful Copilot hints for operators in Operator Line |
| **Integration plugin** | AI requests from the knowledge base integration plugin (AI Vision, chunkify AI, etc.) |

**How to read the report:** the three breakdown metrics (**Answers provided to clients** + **Copilot hints** + **Integration plugin**) sum to **Total AI Answers provided**.

**Same total elsewhere:** **Settings → Instance statistics** (root only) — **Total AI Answers provided** card for the selected period, aligned with the Fastline report.

> Classic Fast Line (knowledge base without Dify Pro) is **not** included in these metrics — see FastLine efficiency metrics above.

---

## Copilot Metrics

### Copilot Usage

**Total requests** — total number of unique requests from clients recorded in the system. Operator-initiated requests are not included in the calculation.

**Requests with Copilot initiated** — number of requests where Copilot was connected to the dialogue and opened for interaction.

**Messages from client** — total number of messages sent by client within the request.

**Copilot found answer** — number of client messages for which Copilot was able to find an answer in the knowledge base or suggest relevant information.

### Copilot Efficiency

**% usage** — percentage ratio of requests where operator used Copilot answers to total number of requests with connected Copilot.

**Number of clicks "Confirm and send"** — number of cases when operator clicked "Confirm and send" button, accepting suggested answer from Copilot without changes.

**Number of clicks "Edit"** — number of cases when operator chose "Edit" option, changing answer suggested by Copilot before sending.

**Copilot hint was not used** — number of client messages in response to which Copilot provided answer options, but none were used by operator (there was no click on "Confirm and send" or "Edit" buttons).

### Impact on Quality

**Average dialogue duration** — average time of active interaction in dialogues with client (in seconds). Metric compares dialogues where Copilot was used and those where it was not used.

**Copilot CSAT** — client satisfaction indicator in dialogues where Copilot was used, compared to those where it was not used. Determined based on average rating left by clients after interaction completion.

---

## CSAT Metrics

**Average CSAT rating for period** — average client satisfaction rating (CSAT) for the reporting period.

**CDSAT** — percentage of chats for the reporting period with ratings "1" and "2" (low satisfaction).

**Number of surveys** — number of dialogues with surveys.

**Survey percentage** — what percentage of all dialogues had satisfaction surveys (CSAT).

**Percentage of incoming chats with ratings** — percentage of chats with ratings from all incoming chats for the reporting period.

### Summary on the Service Quality tab

In native statistics (**Statistics** module, **Service quality** tab), the **Summary for the period (all operators)** block shows period aggregates:

| Metric | Description |
|--------|-------------|
| **Accepted by operators** | Dialogs accepted by operators in the period (matches the sum of **Accepted** on the **Operators** tab for the same filters) |
| **Rated chats** | Dialogs with a client rating in the selected scale |
| **Positive ratings** | Count of positive ratings (1–5 scale: 4 and 5; 0/5/10 scale: 10) |
| **Negative ratings** | Count of negative ratings (1–5: 1 and 2; 0/5/10: 0) |
| **Unrated** | Dialogs without a client rating |

The **Rating scale** filter limits CSAT metrics to **1–5 (standard)** or **0 / 5 / 10 (poor / average / excellent)**. See [How to view the quality period summary](/en/statistics/how-to/view-quality-tab-summary.md).

---

## FCR (First Contact Resolution) Metrics

**FCR 24h** — percentage of repeat requests from users within 24 hours.

**FCR 48h** — percentage of repeat requests from users within 48 hours.

**FCR 72h** — percentage of repeat requests from users within 72 hours.

> 💡 **Note**: FCR (First Contact Resolution) — metric that tracks repeat requests. A request is considered resolved on first contact if the client did not contact again with the same question.

### Calculation logic (technical)

Metrics are based on `chat_close` events in the KwStatistics collection. Fields `data.fcr24`, `data.fcr48`, `data.fcr72` store the **count** of previous client requests within the respective time window: `0` — no repeat, `1+` — repeat contact. FCR is calculated as the share of dialogues where `data.fcrX > 0` out of the total number of closed dialogues.


---

## FRT (First Response Time / SLA) Metrics

**SLA 1m** — percentage of dialogues in which first response was sent within 1 minute.

**SLA 2m** — percentage of dialogues in which first response was sent within 2 minutes (including 1 min).

**SLA 3m** — percentage of dialogues in which first response was sent within 3 minutes (including 1 min and 2 min).

**SLA 5m** — percentage of dialogues in which first response was sent within 5 minutes (including 1 min, 2 min, and 3 min).

**SLA 10m** — percentage of dialogues in which first response was sent within 10 minutes (including 1 min, 2 min, 3 min, and 5 min).

> 💡 **Note**: SLA metrics (Service Level Agreement) — key efficiency indicators applied to dialogues for controlling time of certain actions.

---

## User Metrics

**Number of unique users** — number of unique users for the reporting period.

**Number of active users** — number of bot users per day who performed any actions. Action — scenario transition, even if transition was made through broadcast.

**Number of registered users** — number of users entered in the Clients section.

**Number of users blocked by operators** — users blocked by an operator or administrator through moderation in ConnectiveOne (system ban). Does not include users who blocked the bot in Telegram.

**Users who blocked the bot in Telegram** — Telegram users who blocked the bot in a private chat. The counter does not decrease when the user unblocks the bot. Available for the Telegram channel only.

> On the **Blocked users** tab, these metrics are shown alongside the number of registered users.

---

## Metrics by Topics

**Topic** — list of topics, including "No topic".

Metrics by topics allow analyzing efficiency of work with different types of requests. They include:
- Number of dialogues by topic
- Average processing time by topic
- CSAT by topic
- Number of operators who worked with the topic

---

## Metabase Dashboards

Metrics are organized by dashboards for convenient navigation:

- **Dialogues** — detailed information about each dialogue
- **Requests** — general request statistics
- **Operators** — operator work metrics
- **Operator Workload** — operator workload analysis
- **Topics** — metrics by dialogue topics
- **FastLine** — AI bot efficiency
- **Service Quality** — CSAT and other quality metrics
- **Service Metrics** — FCR, FRT, and other SLA metrics
- **Users** — user statistics
- **Custom Boards** — custom dashboards created specifically for the company

---

## Related Materials

- [How to pass parameters from the website (widget metadata)](/en/channels/how-to/pass-parameters-from-widget.md#automatic-metadata-addition)
- [How to view the quality period summary](/en/statistics/how-to/view-quality-tab-summary.md)
- [How to read duration and reaction metrics on Appeals](/en/statistics/how-to/view-appeals-duration-and-reaction-metrics.md)
- [View dashboards](/en/statistics/how-to/view-dashboards.md)
- [Create and configure funnel](/en/statistics/how-to/create-and-configure-funnel.md)
- [Configure CSAT](/en/statistics/how-to/configure-csat.md)
- [View operator schedule](/en/statistics/how-to/view-operator-schedule.md)
- [Export statistics data](/en/statistics/how-to/export-statistics-data.md)