# Calls and call-backs

> Find any phone call, see why it happened and how it ended, play the recording, read the transcript and the call outcome, and manage scheduled call-backs.

Source: https://docs.amomic.in/console/calls

---

**Calls** lists every phone call in your workspace, newest first. That covers calls people make to your numbers, calls your agent places (one at a time through the API, from a campaign, or as a call-back), and calls your team places from a ticket. Open any call to see how it ended, what was learned, how fast the agent answered, the recording and the transcript.

Everyone in the workspace can open Calls, and anyone can cancel a scheduled call-back or ask for a call to be read again. To change how calls are answered, use **Phone settings** at the top of the page, which opens [the phone workspace](/channels/phone-workspace).

![The Calls page: four tiles (50 loaded calls, 122:03 talk time, 90% completion, 1.2s median response), three scheduled call-backs, the search and filters, the call list, and an inbound call open on the right](https://docs.amomic.in/screens/voice/calls.light.webp)

*The tiles and the list cover the calls loaded so far. Select a call to open it on the right.*

## Who sees Calls

Calls comes with any plan that includes the Voice channel, which is **Starter**, **Growth** and **Enterprise**. On **Free**, the page is locked and reads **See every call your agent handles**; upgrading opens it. Workspaces that already had phone numbers before plans existed keep it. See [Plans and limits](/operate/limits#plans).

![Calls on the Free plan: a locked panel titled See every call your agent handles, explaining that transcripts, recordings, outcomes and latency arrive once the plan includes the Voice channel](https://docs.amomic.in/screens/voice/calls-locked.light.webp)

## The summary tiles

The four tiles at the top describe the **calls loaded on the page**, not your whole history. The page loads 50 calls at a time; **Load older calls** at the bottom of the list adds the next 50, and the tiles change to include them.

| Tile | What it counts |
|---|---|
| **Loaded calls** | How many calls the page has loaded. |
| **Talk time** | Answered time across those calls, in minutes and seconds. Ringing is not included. |
| **Completion** | The share of those calls that connected and ended normally, whoever hung up. Calls nobody answered, busy lines and failures count against it. |
| **Median response** | The middle of the agent's response times across those calls: how long, typically, between a caller finishing and the agent starting to speak. |

The tiles ignore the search and filters. For totals over a date range, use [Analytics](/console/analytics#voice-performance). Analytics works its completion figure out a little differently, so the two percentages can differ.

## Finding a call

:::steps

:::step Search
Type in **Search caller, number, or call ID…**. It matches the caller's number, your business number, the call ID and the campaign name. Search only looks through calls already loaded, so for an older call choose **Load older calls** first, or open it from a link (below).
:::

:::step Narrow by direction, outcome or agent
Use the three menus next to the search.
:::

:::step Open the call
Select a row. Its details open on the right.
:::

:::

| Filter | Options | Notes |
|---|---|---|
| **Direction** | **All directions**, **Inbound**, **Outbound (agent)**, **From a ticket** | **Outbound (agent)** is every call the agent placed: API, campaign and call-back. **From a ticket** is calls your team placed. |
| **Outcome** | **All outcomes**, **Completed**, **No answer**, **Busy**, **Idle hangup**, **Time cap**, **Failed** | Matches how the line closed, not the call outcome badge. **Completed** here only catches calls where the caller hung up; to see every connected call, leave it on **All outcomes**. |
| **Agent** | **All agents**, then each agent by name | Reloads the list with only that agent's calls. Calls placed from tickets have no agent, so they drop out. |

When nothing matches you see **No calls match these filters**; **Clear filters** resets all four.

**Links that open a call.** `https://console.amomic.in/calls?call=<call_id>` opens Calls with that call selected, even if it is older than the first 50. `…/calls?search=<text>` opens with the search filled in. A contact's activity in [Contacts](/console/contact-memory) links to its calls this way.

## Reading the list

Each row is one call:

- **The icon** shows the kind of call: an arrow pointing in (green) for inbound, an arrow pointing out for a call the agent placed, and a headset (amber) for a call placed from a ticket.
- **The number** is the other party: the caller on inbound calls, the person called on outbound ones. Next to it is the duration. A `~` in front, such as `~3:12`, means the duration was reconstructed for an older call rather than measured.
- **Who and when**: the agent's name, or for a ticket call the teammate who placed it, and the start time in your workspace's time zone.
- **The badge** says how the call went. Once a call has been read it shows the outcome, such as **Completed**, **Partly done**, **Asked to call back**, **Not interested**, **Wrong person**, **Hung up early** or **No conversation**. Until then, and for calls nobody answered, it shows how the line closed, such as **Completed by agent**, **No answer**, **Busy**, **Declined**, **Time cap** or **Could not dial**.
- **The origin line** says why an outbound call happened: **Placed through the API**, **Campaign · *name*** (with **· follow-up** on a campaign's call-back or second attempt), **Scheduled call-back**, or **From a ticket · *teammate* · phone** (or **web call**). Inbound calls have none: the customer rang you.
- **The summary**, one line from the call outcome, and the call's **Median response**.

Campaign names stay on their calls, so the history still makes sense after a campaign is paused or finished.

## Inside a call

The header repeats the number, badge, start time, direction and agent, with the origin line beneath. The chip on the right shows the last 8 characters of the call ID; select it to copy the full ID, which is what support and the [API](/api/calls) ask for.

| Tile | Shows |
|---|---|
| **Business number** | Your number the call used. |
| **Direction** | **Inbound** or **Outbound**. |
| **Duration** | Talk time, from answer to hang-up. |
| **Conversation turns** | How many times the caller and the agent took turns. |

Below the tiles, in order:

- **Why we called**, on outbound calls that were given a purpose. It is the brief the agent was given.
- **Call outcome**: see [the card](#the-call-outcome-card) below.
- **Response performance**: the agent's **Median**, **p95** and **Slowest** response on this call. Green means 1.5 seconds or less (the target), amber up to 2.5 seconds, red above that. A single slow turn is often a lookup or tool call; a red **Median** means most answers kept the caller waiting.
- **Call recording**, when recording was on for the call. See [Recordings and transcripts](#recordings-and-transcripts).
- **Transcript**: every turn, caller and agent, with the time into the call.

![Inside one call: the Call outcome card (Completed, Goal met, Positive, a summary, What we learned 3 of 3 answered), Response performance with median 1.2s, p95 1.9s and slowest 2.6s, a recording player and the start of the transcript](https://docs.amomic.in/screens/voice/call-detail.light.webp)

## The call outcome card

Every finished call the agent handled is read once it ends, usually within seconds, and **Call outcome** shows the result:

- **How it ended**: **Completed**, **Partly done**, **Asked to call back**, **Not interested**, **Wrong person**, **Hung up early**, **No conversation** or **Didn't pick up**, with a one-line explanation.
- **Goal met**, **Goal not met** or **Goal unclear**, and the caller's mood: **Positive**, **Neutral** or **Negative**.
- **The summary** of the call.
- **What we learned · *x* of *y* answered**: the answer to each question set for the call, or **Not answered** where the caller never said. An important question that went unanswered is marked **· important**. Inbound calls use the number's [Outcomes questions](/console/call-outcomes); campaign and API calls use their own [capture](/campaigns/outcomes).
- **The next step owed**: **Call them back**, **Send a WhatsApp** or **A person should follow up**, with the reason, or **Nothing owed to this person.**

**How this works** opens the [full explanation of how a call is read](/campaigns/outcomes#one-call-s-outcome). If a reading looks wrong, choose **Re-read**. A call that has no reading yet shows **Read this call** instead, with one of these notes:

| Note | Meaning |
|---|---|
| *Being read now. This usually takes a few seconds after a call ends.* | The call just ended. Wait a few seconds. |
| *This call hasn't been read for its outcome.* | Usually a call from before call outcomes existed. **Read this call** reads it now. |
| *We couldn't read this call's outcome. Try again.* | Reading failed. Missed readings are also retried automatically every few minutes. |

The same reading is sent to your systems on the [`call.outcome`](/webhooks/events#call-outcome) webhook. Calls your team places from tickets are not read this way; they get an **AI summary** instead (below).

## Calls placed from tickets

When someone on your team calls a customer from a [ticket](/console/tickets), the call appears here with a headset icon and **From a ticket · *teammate* · phone** (or **web call**). No agent takes part, so the page shows different things:

- The tiles read **Business number**, **Placed over** (**Phone** or **Web call**), **Talk time** and **Placed by**.
- *This talk time counts towards your plan's voice minutes.* Ticket calls use the same voice minutes as agent calls, metered once per call, even in phone mode where the call uses two connections.
- An **AI summary** replaces the call outcome card, and there is no response performance.
- The transcript reads *This call connected two people directly, so there is no agent transcript.*

![A call placed from a ticket: Placed over Phone, Talk time 4:23, Placed by Priya Nair, the note that this talk time counts towards the plan's voice minutes, an AI summary, and a transcript note that the call connected two people directly](https://docs.amomic.in/screens/voice/call-ticket.light.webp)

## Recordings and transcripts

When recording is on for a number and the audio is available, **Call recording** has a player. Switch it on per number with **Record calls** in [the phone workspace](/channels/phone-workspace). If recording was on but the audio is missing, the card says *Recording was enabled, but the audio is unavailable.* A call with no recording section was never set to record.

The transcript does not depend on recording: it is kept for every agent call. A call with no turns at all, such as one nobody answered, reads *No transcript was captured for this call.*

Before switching recording on, read [Security and data](/operate/security#call-recording) for notice and consent.

## Talk time versus ringing

Duration starts when the call is answered, not when dialling begins, so an unanswered call shows `0:00` and adds nothing to **Talk time**. A `~` before a duration marks an older call whose length was reconstructed rather than measured; on a ticket call that figure includes ringing and was not billed.

## Scheduled call-backs

Calls the agent has been asked to make later sit in a panel above the filters, headed ***n* calls scheduled · placed automatically at the agreed time**. The panel only appears when there is at least one. Each row shows the number, a badge for why the call exists, the time (and campaign, if any) and the reason when one was given.

| Badge | Why it exists |
|---|---|
| **Customer asked** | The person asked to be called at another time. See below. |
| **Second attempt** | A campaign call nobody answered, set to try once more. See [Unanswered campaign calls](#unanswered-campaign-calls). |
| **Finish the call** | A campaign call that ended partway with answers still missing, in a campaign set to call back once to finish. See [Campaign outcomes](/campaigns/outcomes). |

The list is soonest first. Call-backs are placed within a few minutes of their time. **Cancel** stops one before it is placed (*Call-back cancelled.*). Once placed, it becomes an ordinary call in the list with the origin **Scheduled call-back**, or **Campaign · *name* · follow-up**.

### Customer-requested call-backs

On a call, an agent with the call-back tool can handle "I'm busy, call me in two hours." It books a time, tells the caller the actual local time it booked, and the call-back appears in the panel with **Customer asked** and the reason. Set the tool up under [Agent tools](/build/tools).

### Unanswered campaign calls

A call campaign can make one more attempt for people who don't answer. The retry is booked inside the campaign's calling window and shows as **Second attempt**. It is exactly one retry, not a ladder. A time the customer chose always wins: the campaign does not overwrite it.

Campaign reports separate answered, being retried and finally unanswered recipients. See [Campaigns](/campaigns/overview#following-a-running-campaign).

## Troubleshooting

| What you see | What to do |
|---|---|
| **No calls yet** | No phone number has taken or placed a call. **Set up the phone channel** opens the phone workspace. See [Get a phone number](/channels/phone-numbers). |
| A call you expect isn't in the list | Search and the tiles only cover loaded calls. Choose **Load older calls**, or open the call by link with `?call=<call_id>`. Check the **Agent** filter too. |
| Calls the agent ended disappear under **Outcome → Completed** | That filter only matches calls the caller ended. Use **All outcomes** and read the badges. |
| The **Completion** tile and Analytics disagree | They cover different calls (loaded calls versus a date range) and count completion slightly differently. |
| *Recording was enabled, but the audio is unavailable.* | The call was set to record but no audio was saved. The transcript is still there. If it keeps happening, contact support with the call ID. |
| A red **Median** response | Callers waited. See [How voice works](/channels/voice) for what slows an agent down, such as long tool calls. |
| **Call outcome** stays on *Being read now* | Choose **Read this call**. Readings that fail are also retried every few minutes. |
| A call-back didn't happen | Check it wasn't cancelled. A campaign call-back is placed only inside the campaign's calling window. |

## Related

::: cards

::: card How voice works
[The inbound route, opening line, recording and outcomes](/channels/voice).
:::

::: card Outcomes on incoming calls
[Ask callers a few questions](/console/call-outcomes) and see the answers added up.
:::

::: card Outbound calls
[Place an AI call](/channels/outbound-calls) from your backend.
:::

:::