Skip to main content

​
Contact Center live operations floor

The Contact Center — Live Operations page at Voice → Contact Center (/voice/contact-center) is the composite supervisor floor: it pulls the queue pulse, the agent status board, the service-level gauge, live alerts and callbacks, and the per-channel SLA rollup onto one column so the first answer a supervisor needs — “how is the floor doing right now, glance level” — never takes several tabs. The page currently renders a disclosure line, “Showing sample data for demonstration.”, beneath the header. The figures on the page are a fixed showcase dataset so the surface can be reviewed and the workflow learned before it is wired to your live queues; this guide covers the panels as shipped and, at the end, the role gates and refresh behavior to expect. The drilled destinations (queue console, callbacks list, softphone, inbox) are all live routes. This guide strings the composite view together for a first-week supervisor. The per-surface guides — wallboard, wallboard console, supervisor live monitoring (voice), and supervisor live monitoring (digital) — each go deeper on one panel; this one is the map of the whole floor.

​
1. What the page is for

The floor sits between two surfaces you will also use every day:
  • One queue, full detail — Voice → Queues (/voice/queues) is the per-queue console: staffing, overflow actions, SLA targets, escalation ladder. When an alert names one queue, that is where you fix it.
  • Every channel, for the room — the wallboard (/voice/wallboard) is the fullscreen, TV-mounted display: auto-cycling tiles, masked caller numbers, no click-throughs. It is for the common-room TV, not your desk.
  • The floor — this page — is the desk-side composite: one read of queue health, staffing, SLA, and open callbacks across channels, with every panel linking out to the surface that owns the fix.
Use the floor to notice; use the linked surface to act.

​
2. The panels in order

Six panels read top to bottom, the same order your eyes should walk them:
  • Header — the floor title with a LIVE pill, plus two actions: Export (downloads the agent status board as CSV, scoped to exactly the rows your filters currently show) and Broadcast to floor (opens the shared announcement composer; owner/admin roles only, same gate as the wallboard trigger it reuses).
  • Queue pulse — six KPI cards across the top: In Queue, Longest Wait, Service Level, Abandon Rate, Avg Handle Time, and Occupancy. Each card carries a label, the current value, a direction chip (delta vs a reference window), a thin progress bar, and a target marker so “are we above or below goal” reads at a glance.
  • Agent status board — the roster with a segmented channel filter (All / Voice / Chat / Email), a name search, a status Filter popover (Available / On Call / Offline checkboxes), and a per-row action menu with Call from softphone (/voice/dialer) and Message via inbox (/inbox) links.
  • Service-level gauge — the right rail’s top card: a ring with the current service-level percentage against the queue target (for example “Target 80/20”), a measurement window label, and a legend row splitting Calls / Met SLA / Breached.
  • Alerts & Callbacks — the rail’s middle card: flagged queue events with tone pills (Breaching, At risk, On track) and per-alert action labels that link through to the owning surface (/voice/queues for queue events, /voice/callbacks for callback due-times). The header badge counts open alerts.
  • Live pulse, SLA timeline, and per-channel rollup — the rail’s Live Pulse ticker streams newest-first queue events (New / Routed / At risk / Abandoned / Resolved, coloured per channel) on a shared clock; the footer’s Live SLA Timeline scatters the waiting heads against the per-queue target divider; and a 2×2 channel grid (Voice, Live Chat, Email, WhatsApp) breaks out each channel’s queue depth, longest/oldest wait, and SLA percentage so cross-channel imbalance is visible without opening per-channel queues.

​
3. Acting on callbacks

Callback rows surface in the alerts card as On track — callback due entries. Clicking the row’s action opens Voice → Callbacks (/voice/callbacks) on the due entry. From the callbacks console the lifecycle is:
  1. Find the due request. The list shows the queue it came from, the masked caller number, status, and when it was requested. Due entries sit in On track until handled — the green tone is a reminder, not an alarm.
  2. Claim and place it. Work the list in position order; the status moves through Dialing to Connected when the caller answers and joins an agent, and the request completes. If a callback is business-critical, re-prioritize it to the top of the pending list from the row menu while it is still waiting.
  3. Complete or cancel. Settled requests close and leave the queue, so the console always reads as the work that remains.
The full operator lifecycle — where virtual-hold requests come from, filtering the claim queue, re-prioritizing, and how completion clears the console — is documented in the voice callback console guide. The floor is where you notice the due callback; the console is where you work it.

​
4. Acting on SLA breaches

Two panels tell you a breach is coming or has landed; each leads to a different fix.
  • The service-level gauge reads against its own target. When the ring drops under the configured target (for example 80/20), the legend row quantifies how far: calls offered vs met vs breached. A queue with no configured SLA target will not populate the gauge or the footer timeline — set the answer-within target per queue in Voice → Queues first.
  • A red — Breaching alert row means a queue has already missed its window. Its action opens the queue console (/voice/queues) on that queue: rebalance agents into it from the roster, or trigger the mitigation rung the queue’s escalation ladder defines. The full policy — objective, forecast gate, breach ladder — is the queue SLA forecast-callback runbook; do not re-derive it from the floor page.
  • An amber — At risk row is a spike the forecast is still absorbing. Two fixes apply, by cause:
    • A short spike — open the queue console and shift one agent early rather than wait for the breach alert to fire.
    • A structural miss (the same queue at risk every day at the same hour) — a rebalancing will not hold it. Open intraday adherence and agent requests and fix the staffing: the intraday view shows where scheduled coverage drifts from the forecast, and agent requests explain the absences behind it.
  • Per-channel SLA cards (footer). Each card colours its SLA bar by threshold (90%+ good / 80%+ watch / below that critical) so the channel with the worst number pulls the eye. Click through to the channel’s own queue surface to act; do not rebalance voice agents into a chat spike — the channel cards exist precisely to keep that imbalance visible.
The floor page is read-only apart from Export and Broadcast. Rebalancing, SLA-target edits, callback placement, and the breach-policy ladder all happen on the destination surfaces the alerts link to — never inline on the floor.

​
5. A worked first-week supervisor loop

Three passes a day covers a new supervisor’s whole first week; the floor is the anchor of each. 09:00 — the stand-up glance. Open the floor before stand-up. Queue pulse for overnight drift — In Queue and Longest Wait tell you what the off-shift left behind; the direction chips sort good-news from caution in one read. Alerts card for anything that fired overnight. Agent status board filtered to Available to confirm staffing before the first spike lands. Thirty honest seconds, and the stand-up agenda reads itself. Midday — the SLA triage. When an amber or red alert lands mid-peak, do not reach for a spreadsheet. Read the gauge against its target, click the alert row through to the queue console, shift agents into the hot queue, and watch the live SLA timeline’s scatter drift back to the left of the target divider. Due callbacks found in the same pass get claimed from /voice/callbacks. The wire never leaves the floor’s window; you lose the room’s attention if you hop tools mid-incident. If the same queue flags at the same hour two days running, stop rebalancing and fix the staffing in intraday adherence instead. End of day — the export. Set the board’s filters to the team and statuses you report on, then use the header’s Export to download exactly those rows as CSV — the file is scoped to what you see, so a filtered view never leaks the whole roster into a report. A name search that matches nothing exports nothing (the page tells you rather than handing back a header-only file), so confirm your filter before you export. Use Broadcast to floor (owner/admin) for the floor-wide note that closes the shift.

​
6. When to switch to the dedicated surfaces

The floor is a dashboard, not a replacement for the specialist surfaces. Switch when:
  • One queue needs real work — open the queue console from the alert’s action (or Voice → Queues directly). Staffing, overflow actions, SLA targets, and the escalation ladder all live there; the floor’s gauge and timeline read whatever the queue configured. See Set up and run voice queues.
  • The room needs to see it — put the wallboard on the TV: fullscreen, auto-cycling, masked numbers, omnichannel tiles. The wallboard console guide covers TV mode and its SLA telemetry. Use the floor at your desk, where the rail and click-throughs live.
  • A caller needs you in the conversation — supervisor live monitoring (voice) for listen / whisper / barge, and supervisor live monitoring (digital) for chat read-over and join. The floor never bridges a conversation itself; it routes you to the surface that does.
  • Digital team performance is the question — the floor shows cross-channel balance, not per-agent digital performance. For chat and messaging team performance over a window, open Inbox → Analytics (/inbox/analytics).
  • The pattern is a person, not a queue — when one agent’s name keeps surfacing on the wrong rows (long On Call while the queue backs up, a handle-time chip that keeps flagging), hand the pattern to the supervisor coaching surface. The floor spots the pattern; the coaching surface works it.
  • The threshold itself is wrong — tenant-wide ad-hoc thresholds (a service-level floor, a longest-wait guard) live in wallboard alarm rules and backstop the per-queue breach policy the runbook wires.
When a drill-down on the floor takes you into one of those surfaces, that surface’s guide is the authority — this page stays the map.

​
7. Role gating and refresh behavior

  • Who sees what. Every signed-in role on your account can read the floor. Broadcast to floor renders for owner/admin roles only — the same gate the wallboard’s announcement trigger uses — and is hidden rather than disabled for everyone else: a viewer who can never broadcast gets nothing to click, not a button that fails.
  • Your controls are tenant-owned. The export contains only your own roster rows; broadcasts announce only to your own logged-in agents; the queues, callbacks, and SLA targets the floor links to are configured inside your account. Nothing on the page reads or acts on another tenant.
  • Refresh. The shipped page renders a fixed showcase dataset — the live ticker animates and the shared clock advances time-ago labels, but the figures do not poll your queues. Until the live feed lands, treat the floor as the workflow rehearsal space the disclosure line describes, and the drilled destinations (queue console, callbacks, wallboard) as the live surfaces they already are.