​

Reading the Voice → Calls console

​

Filter and segment the log

• View group (All / Active / Failed / Completed) — segments in one tap. Each group chip shows a count computed over your current date range and direction, so you can scan “how many failures today” without opening a filter. Click a chip again to go back to All.
• Direction — All, Inbound, or Outbound.
• Status chips — granular control over the exact statuses you care about: completed, in progress, ringing, queued, initiated, transferred, busy, no answer, failed, canceled. Click more than one chip to OR them; the View group and the chips are two lenses on the same column, so they never disagree.
• Date range — presets or a custom window. The log, the group-chip counts, and the analytics panel all honor it.
• Search — matches against caller/callee numbers and other string fields on the call envelope.

​

Open the call detail sheet

• Fraud banner — when the carrier-side attestation or scoring flagged the call, a warning banner says so above everything else.
• AI agent handoff summary — for calls that passed through an AI agent, a card carrying the escalation reason, key topics, action items, and the slots the agent collected, so a human reviewer can pick up the thread.
• Key fields — from, to, direction, status, start/end timestamps, and duration.
• Recording — an in-sheet player when a recording exists, with a quality-check verdict when QC has run, and a re-run control to score it again.
• Timeline — the diagnostic view described below.
• Transcript and screen recordings — when the call produced them.

​

Read the timeline as a diagnostic view

• No answer — rang, and fell to voicemail. The steps go initiated → ringing → no answer, with the ringing step carrying its duration against your ring-timeout setting. The callee’s phone rang (say, three rings’ worth of seconds before the timeout) and nobody picked up within the timeout. The destination is reachable; the failure is human-side. Redial or hand the row to the callback window — there is no infrastructure to chase.
• No answer — never left the queue. The steps stop at initiated → queued, and nothing rings. The call spent its life waiting for a dialer resource or an available agent and expired there. The move is capacity-side: widen the dialer’s concurrent-call cap or the agent pool for the campaign’s window, because retrying into the same queue reproduces the same result.

​

Worked example: one failed call, start to finish

1. Isolate the failure. Click the Failed view-group chip, set Direction to Outbound, and pick Today in the date range — three taps. The Failed chip counter already told you today’s failure count before you clicked anything.
2. Open the row. Click the failed row; the detail sheet slides over. Check for a fraud banner first — a carrier-side attestation or scoring flag changes who you escalate to. For a call that passed through an AI agent, the AI handoff summary carries the escalation reason and the slots the agent collected, so a human reviewer picks up the thread without a replay.
3. Link it. Use Open full detail in the sheet header and copy the bookmarkable URL into the ticket or chat thread. Everyone who opens the link lands on the same call, with no instructions to dictate.
4. Read the timeline. A no answer row whose timeline stops at queued means the call never rang, while one that shows ringing means the destination rang and timed out — the two readings above are the difference between a capacity fix and a redial.
5. Back off to the aggregate. If this failure is not alone, the Analytics tab over the same date range shows the failure rate by day and the top failing destinations — one bad trunk route or destination cluster surfaces there before it surfaces in another ticket.

​

From a failed call to the rest of operations

• Cost. When billing rated the call, the detail-sheet header shows its cost. Voice charges resolve through the same shared per-event pricing step SMS charges use — the per-operator precedence of a verbatim per-carrier override over any markup rule. The MCCMNC overrides guide is written against SMS sends; the worked check for a voice row is the same in shape: find the destination’s network on the Billing → SMS rates card, and confirm the per-operator rate labeled absolute_override matches the per-carrier terms you negotiated — an override must be copied verbatim, never re-derived from a markup. A charge that disputes the card is a support signal, not a guessing game — the pricing and rate resolution concept page states the display-matches-charge invariant support applies.
• Request id. When an event for this call surfaces on the Insights → Logs page — failed-call events appear there alongside webhook and agent events — expand the row and copy its request id. Paste that id into Developer → Request Logs or Developer → Delivery Logs to land on the exact API call or delivery behind it, payloads included. That is the anchor support traces by.
• Messaging equivalent. When the same investigation crosses a channel — a voice call that followed up on an SMS, or an SMS that preceded a call — the Delivery Log runs the same loop for the messaging side: one search across SMS, WhatsApp, email, and voice, then the row’s timeline.

​

The analytics panel

• Scope: analytics aggregate only the data the console has loaded for the current window — they operate on the cursor-walked result set up to a cap. The log below has the ground truth for anything past that cap.
• Success-rate denominator: the daily success-rate divides answered-plus-completed over terminal statuses. In-flight calls (ringing, in progress) don’t drag the rate down while they’re still live. This matches the status badges in the log — the chart and the table classify the same way.

​

Reading call logs over the API

curl -X GET "https://api.orbit.devotel.io/api/v1/voice/calls?status=failed&direction=outbound&since=2026-09-21T00:00:00Z&until=2026-09-28T00:00:00Z&limit=20" \
  -H "X-API-Key: dv_live_sk_your_key_here"

import { Orbit } from '@devotel-orbit/node'

const orbit = new Orbit({ apiKey: process.env.ORBIT_API_KEY! })

const page = await orbit.voice.calls.list({
  status: 'failed',
  direction: 'outbound',
  since: '2026-09-21T00:00:00Z',
  until: '2026-09-28T00:00:00Z',
  limit: 20,
})
for (const call of page.data) console.log(call.id, call.status)

{
  "data": [
    {
      "id": "call_abc123",
      "direction": "outbound",
      "from": "+14155552671",
      "to": "+12125554567",
      "status": "failed",
      "duration_seconds": 0,
      "created_at": "2026-09-24T14:32:11Z"
    }
  ],
  "meta": {
    "request_id": "req_list_001",
    "timestamp": "2026-09-28T12:00:00Z",
    "pagination": {
      "has_more": true,
      "cursor": "2026-09-24T14:32:11.000Z|call_abc123"
    }
  }
}

curl -X GET "https://api.orbit.devotel.io/api/v1/voice/calls/call_abc123" \
  -H "X-API-Key: dv_live_sk_your_key_here"

const call = await orbit.voice.calls.get('call_abc123')
console.log(call.data.status, call.data.duration_seconds)

​

See also

• Recordings guide — recording lifecycle, QC, and retention
• Voice quality reading — interpret call-quality signals
• Delivery log — the messaging-side counterpart to the call log