> ## Documentation Index
> Fetch the complete documentation index at: https://docs.callivox.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Dashboards and tags
> The built-in Overview dashboard, building your own from the variables your agents collect, and the Tags page that shows what people contact you about.
Dashboards turn what happens in your inbox into numbers you can look at. There is one built-in dashboard that works from day one, a builder for making your own, and a separate **Tags** page that shows what your conversations are actually about.
All of it sits under **Dashboards** in the sidebar.
## The Overview dashboard
**Overview** is always the first item, it is maintained for you, and it needs no setting up. A range control at the top switches between the last **7d**, **30d** and **90d** — 30 days to begin with.
If nothing has happened yet you will see "No conversations or calls yet — connect a channel to start seeing activity here." with a **Connect a channel** link.
### The numbers at the top
| Card | What it tells you |
| ----------------------- | -------------------------------------------------------------------------------------------- |
| **Total conversations** | How many conversations you had across all channels in the range. |
| **Open conversations** | How many are still awaiting a reply. |
| **Avg first response** | How long, on average, a customer waited for their first reply — counting both people and AI. |
| **Total calls** | Voice calls in the range. |
**Avg first response** is the one to watch week to week. It is the number your customers actually feel, and it is the first thing that slips when volume rises.
### The charts
* **Conversations over time** — volume day by day. Use it to spot your busy days and the effect of a campaign.
* **Conversation status** — the split between open, snoozed and closed. A growing open slice means work is arriving faster than it is being finished.
* **Chats by channel** — where your customers actually contact you, which is often not where you expect.
* **Chats by agent** — the split across your team and your AI agents.
* **Recent conversations** — the latest conversations with their contact, channel, agent, status and last message.
Where your plan includes voice, a **Voice & calls** section follows with **Call answer rate**, **Avg call duration**, **Call outcomes**, **Calls over time**, **Calls by agent** and a **Recent calls** table. See [Voice agents](/voice-agents/overview).
## Building your own dashboard
Your own dashboards answer a different sort of question. Rather than counting conversations, they chart **the variables your agents collect** — the budget a customer mentioned, the product they asked about, the date they wanted, whether they qualified.
Those are the same variables you can see and edit in a conversation's **AI details** panel. Anything an agent is set up to collect can be charted here. See [Prompt, knowledge and tools](/ai-agents/prompt-knowledge-and-tools).
Name it — "Sales overview", "Support quality". Dashboards are shared with your whole workspace.
The widget builder opens with a live preview beside it.
Give the widget a **Title**, pick an **Agent** (or **All agents**), then pick the **Variable** to chart. The variable list is whatever that agent collects.
**Counter** for a single number, **Line** for a trend over time, **Bar** to compare, **Pie** for a breakdown, **Table** for the rows themselves. Then pick the **Aggregation** — **Count**, **Sum**, **Average** or **Distribution**. Line charts also take a **Bucket**: **Day**, **Hour** or **Week**.
Narrow the widget to part of the data — "budget is at least 5000", "qualified is true". All conditions must match.
Not every combination is offered: a counter or a line chart can count, sum or average; bar and pie charts show a distribution; tables count. The builder only offers what makes sense for the chart type you chose.
While you are creating a widget the preview shows illustrative sample data, because there is nothing saved to read from yet. Save it and the widget fills with your real numbers; editing it later previews the real data too.
### Arranging and filtering a dashboard
**Edit layout** lets you set each widget's **Width** (1, 2 or 3 columns) and **Height** (S, M or L), then **Save**. Each widget also has **Edit widget** and **Remove widget**.
Above the widgets is a filter bar that applies to all of them at once: a **From** and **To** date range, an **Agent** selector, and a **Channel** control (**All**, **WhatsApp**, **Web chat**, and **Voice** where your plan includes it).
Changing those filters affects only your own screen until you save them. While they differ from what is saved you will see "Unsaved view — only you see this until you save". Click **Save view** to make them the dashboard's defaults for everyone.
Dashboards can be renamed and deleted from the list. Deleting takes its widgets with it and cannot be undone.
## The Tags page
**Tags** answers the question the other dashboards cannot: *what are people contacting us about?*
Callivox reads every conversation and tags it by topic on its own — no forms for your team to fill in, no discipline to maintain. Tagging happens twice: once early, while the conversation is live, so the tags are useful in the inbox, and again when it closes, reading the whole conversation to settle on the final set. **Re-evaluate tags** in a conversation's **More actions** menu re-runs it on demand.
Tags you add yourself are kept separate and are never overwritten by the automatic pass.
Tags aren't only for reporting — a [routing rule](/settings/folders-and-rules#routing-on-ai-assigned-tags) can match on a tag and file the conversation into the right folder automatically.
The page has two tabs.
### Dashboard
Four cards at the top: **Total tags**, **Tagged conversations**, **Open & tagged**, and **AI-created tags**.
Below them, **Conversations per tag** lists every tag with its **Open**, **Snoozed**, **Closed** and **Total** counts, plus a **Status split** bar. Automatically created tags carry an **AI** badge.
Click a row and a panel opens on the right listing that tag's conversations, with a status filter of its own. Click one to read it.
Read this page top-down once a month. The biggest tag is what your business is really being asked about — often something worth a knowledge-base article, a change to your website, or an AI agent of its own.
Before anything has been tagged you will see: "**No tags yet** — Once conversations are auto-tagged (or you add tags manually), their topic breakdown shows up here."
### Manage tags
The second tab is where you keep the vocabulary clean:
> Rename, recolor, delete, or **merge** near-duplicate tags. AI often creates variants like "Refund" / "Refunds" — merging keeps every conversation and re-points it to the target tag.
The grid lists each tag with its **Origin** — **AI-created** or **Manual** — and its **Usage** as a bar and a count, most-used first. Each row offers:
* **Rename**
* **Recolor** — a colour picker with a live preview.
* **Merge into…** — "Every conversation and contact tagged with the source moves to the target, and the source tag is removed."
* **Delete** — "It will be removed from every conversation and contact it's on. This can't be undone."
**+ New tag** adds one by hand: "Give the tag a short, distinct name."
This is one shared vocabulary. The same tags are used on your [contacts](/contacts/managing-contacts) and can be targeted by [campaign audiences](/campaigns/creating-a-campaign) — so merging duplicates here tidies up all three places at once.
Merging is the action worth using regularly. Automatic tagging tends to produce near-duplicates over time, and a monthly pass of merging them keeps the breakdown readable without losing a single conversation.
## Next steps
Where tags and collected variables show up while you work.
Set up the variables you want your agents to collect.