> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.rydhm.ai/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.rydhm.ai/_mcp/server.

# Analytics

**Analytics** in the sidebar shows how your experts are doing across every channel. Everyone in the workspace can open it.

## Choose what to look at

* **Range**: **24h**, **7 days** (the default) or **30 days**, always ending now.
* **Expert**: **All experts**, or one expert. With one expert picked, the page adds that expert's success criteria and hides the per-expert splits.

If some conversations are still waiting to be analysed, a pill shows how many, for example "3 analyses waiting".

## Live now

The strip at the top lists calls and meetings in progress right now, and refreshes every 15 seconds. When nothing is live, it says so.

## The tiles

![Analytics for the last 7 days with tiles and a chart of conversations over time](/_fern-files/_img/6be99208f162c341c8c4d48e0cc791ef6a32e4a6715b4be88a2076cda4d2b8e2.webp)![Analytics for the last 7 days with tiles and a chart of conversations over time](/_fern-files/_img/3e9f5f48b1f24e49e835ab2cf97091345e5943186acc1f987f0ed55459ffb792.webp)

| Tile               | What it counts                                                                                                                                  |
| ------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------- |
| **Conversations**  | Chats and calls started in the range that have at least one message, plus meetings that started in the range and completed, are live, or failed |
| **Minutes**        | Time on calls and in meetings                                                                                                                   |
| **Phone answered** | Phone calls answered out of all phone calls, with the missed and failed ones                                                                    |
| **Analysed**       | Conversations that got a [post-call analysis](/quality/conversations#post-call-analysis)                                                        |
| **Average score**  | The average quality score out of 100, from the analyses                                                                                         |
| **Cost**           | An estimate of what the range's usage cost to run, and per conversation. It is not your bill                                                    |

Deleted conversations are left out of the counts. Their minutes still count, because the usage happened.

## Charts and breakdowns

* **Conversations over time**: calls, chats and meetings, by hour for short ranges and by day for longer ones.
* **Minutes**: time spent on calls and in meetings.
* **Outcomes**: how conversations ended, using the outcomes you defined in each expert's analysis settings.
* **Sentiment and score**: positive, neutral, mixed and negative, with the **Average quality score**.
* **By channel**: chats, web calls, visitors, phone calls, meetings and the API.
* **Success criteria**: with one expert picked, how often each yes/no question was answered yes, no or unclear.

Outcomes, sentiment, scores and success criteria all come from post-call analysis. If an expert doesn't have analysis switched on, these stay empty for it. See [Conversations](/quality/conversations#set-it-up).

## Response time

The **Response time** table shows, per expert:

* **First reply**: how long the expert took to answer the first time on a call;
* **Each turn**: how long it took to answer on later turns of a call;
* **Chat reply**: how long a text reply took.

Each cell shows the median, with the 95th percentile under it. Times up to 1 second are green and up to 3 seconds are amber. Rows marked **(estimated)** were timed from transcript timestamps, which include the length of the reply itself, so they read higher than real response times.

## Visitor feedback

The **Visitor feedback** card covers what visitors told you on your [share links and website widget](/channels/public-links):

* **% helpful**: the share of thumbs up among all ratings;
* **By expert**: the same split per expert, when all experts are shown;
* **Latest comments**: the newest written comments, each linking to its conversation. Select **Show all** to see more;
* **Blocked messages**: how many visitor messages [safety screening](/workspace/security#safety-screening-of-public-messages) blocked in the range.

## Export

Admins can select **Export CSV** to download the analyses in the range: one row per analysed conversation, with its outcome, score, sentiment, summary, facts and success criteria. The export holds up to 10,000 rows.

## Where else to look

* **What you're charged.** Credits spent, by channel or by expert, are on **Settings → Billing** under **Where credits went**. See [Billing](/workspace/billing).
* **API traffic.** Admins can see API requests and usage in **Developers → Usage** and **Developers → Request log**.
* **Problems as they happen.** [Alerts](/workspace/alerts) tell admins when calls fail, phones go unanswered or tools break.
* **Test results.** Test runs don't create conversations, so they never appear in Analytics. See [Test suites](/quality/test-suites).

## Related

* [Conversations](/quality/conversations)
* [Test suites](/quality/test-suites)
* [Alerts](/workspace/alerts)