A supervisor walks into their shift. 30 agents on the line, 12 calls in queue, wait time climbing. Without a real-time dashboard, they find out about the problem when customers have already hung up.
My take: I have seen supervisor dashboards with 40+ widgets, real-time stock-ticker animations, and color schemes that would make a traffic light jealous. The supervisors using them? They look at their phone instead. The best dashboards are boring. They show six things clearly, and they turn red when something breaks.
Yesterday's report will not save today's SLA. Here is exactly what should be on the screen — with Asterisk configurations to make it work.
The 6 Widgets That Actually Matter
Everything else is noise. If your dashboard has more than 8 widgets on the main screen, your supervisors are ignoring it.
1. Agent Status Board — Who Is Doing What Right Now
The single most important widget. Shows each agent and their current state:
- —On call — talking to a customer
- —Available — ready to take the next call
- —Wrapup — documenting after a call
- —Paused — break, lunch, coaching
- —Logged out — not in the system
In Asterisk, agent states come from the queue system. Here is how to configure proper state tracking:
; queues.conf — enable detailed member state reporting
[general]
shared_lastcall=yes ; share lastcall across queues
autofill=yes ; fill multiple waiting calls simultaneously
log_membername_as_agent=yes ; log readable agent names
[support-queue]
strategy=leastrecent
timeout=30
wrapuptime=30 ; 30 seconds wrapup — this is critical
autopause=yes ; auto-pause agents who miss calls
autopausebusy=yes ; pause on busy
autopauseunavail=yes ; pause on unavailable
setinterfacevar=yes ; expose MEMBERINTERFACE for trackingThe wrapuptime setting directly controls what your dashboard shows. Set it to 0 and agents jump straight from one call to the next — you lose all visibility into post-call work. Set it to 30 seconds and you can see who is documenting, who is stalling, and who is genuinely processing complex issues.
My take: autopause=yes is controversial. Some supervisors hate it because agents get paused for missed calls. I think it is essential — an agent who misses calls is not available, and your dashboard should reflect reality, not aspiration. Your operator performance tracking depends on accurate state data.
Alarm threshold: More than 20% of agents in wrapup for over 2 minutes → something is wrong. Either calls are unusually complex or agents are avoiding the queue.
2. Queue Depth and Wait Time — The Two Numbers That Decide Everything
Two metrics on one widget:
- —Queue depth — calls waiting right now
- —Longest wait — how long the oldest call has been sitting there
-- Real-time queue snapshot from Asterisk AMI or queue_log
SELECT
queuename,
COUNT(*) FILTER (WHERE event = 'ENTERQUEUE'
AND NOT EXISTS (
SELECT 1 FROM queue_log ql2
WHERE ql2.callid = queue_log.callid
AND ql2.event IN ('CONNECT', 'ABANDON', 'EXITWITHTIMEOUT')
AND ql2.time > queue_log.time
)) AS calls_waiting,
MAX(EXTRACT(EPOCH FROM (NOW() - time))) FILTER (WHERE event = 'ENTERQUEUE'
AND NOT EXISTS (
SELECT 1 FROM queue_log ql2
WHERE ql2.callid = queue_log.callid
AND ql2.event IN ('CONNECT', 'ABANDON', 'EXITWITHTIMEOUT')
AND ql2.time > queue_log.time
)) AS longest_wait_seconds
FROM queue_log
WHERE time > NOW() - INTERVAL '1 hour'
GROUP BY queuename;Thresholds (adjust to your volume):
- —🟢 Green: queue < 5, wait < 30 sec
- —🟡 Yellow: queue 5-15, wait 30-60 sec
- —🔴 Red: queue > 15 or wait > 60 sec
My take: The yellow zone is where supervisors earn their salary. Red means you already failed — customers are hanging up. Yellow means you have 30 seconds to act. If your dashboard does not have a yellow state, you are flying blind between "everything is fine" and "everything is broken."
For deeper queue analysis, see our guide to real-time Asterisk queue monitoring.
3. Real-Time SLA — Not Yesterday, Not Last Week, Right Now
Percentage of calls answered within the target time since the start of the current shift.
Standard formula: 80% of calls answered within 20 seconds (80/20 rule).
-- Shift SLA calculation from queue_log
WITH shift_calls AS (
SELECT
callid,
queuename,
MIN(CASE WHEN event = 'ENTERQUEUE' THEN time END) AS entered,
MIN(CASE WHEN event = 'CONNECT' THEN time END) AS connected,
MIN(CASE WHEN event = 'ABANDON' THEN time END) AS abandoned
FROM queue_log
WHERE time > DATE_TRUNC('day', NOW()) + INTERVAL '8 hours' -- shift start
AND time < DATE_TRUNC('day', NOW()) + INTERVAL '20 hours' -- shift end
GROUP BY callid, queuename
)
SELECT
queuename,
COUNT(*) AS total_calls,
COUNT(CASE WHEN connected IS NOT NULL
AND EXTRACT(EPOCH FROM (connected - entered)) <= 20 THEN 1 END) AS answered_in_target,
ROUND(100.0 * COUNT(CASE WHEN connected IS NOT NULL
AND EXTRACT(EPOCH FROM (connected - entered)) <= 20 THEN 1 END)
/ NULLIF(COUNT(*), 0), 1) AS sla_pct
FROM shift_calls
GROUP BY queuename;Thresholds:
- —🟢 Green: SLA > 80%
- —🟡 Yellow: SLA 70-80%
- —🔴 Red: SLA < 70%
If SLA drops to 70% by mid-shift, the remaining hours need 90%+ performance to recover. Your dashboard should show both the current SLA and the required SLA to hit the daily target. That gap is the supervisor's action signal.
4. Abandon Rate — Every Lost Call Is a Lost Customer
Percentage of callers who hung up before reaching an agent.
; queues.conf — configure announcement to reduce perceived wait
[support-queue]
announce-frequency=45 ; announce position every 45 seconds
announce-holdtime=once ; tell them estimated wait
announce-position=yes ; "You are caller number 3"
queue-thankyou=custom/thankyou ; play after connectThese settings directly affect your abandon rate. Callers who know their position and estimated wait time are significantly less patient — but less likely to rage-quit. Counter-intuitive: position announcements at 30-second intervals reduce abandonment by 15-20% compared to silence.
Thresholds:
- —🟢 Normal: < 5%
- —🟡 Warning: 5-8%
- —🔴 Critical: > 8%
My take: The 5% benchmark is lazy — we wrote about why. What matters is your abandon rate by queue and by time-of-day. A 3% overall rate might hide a 15% rate during lunch hours. Segment before you react.
5. AHT and ASA — Speed Metrics With Context
Two numbers side by side:
- —AHT (Average Handle Time) — average seconds per call including wrapup
- —ASA (Average Speed of Answer) — average seconds to pick up
-- Real-time AHT and ASA per queue, current shift
SELECT
queuename,
ROUND(AVG(CASE WHEN event = 'COMPLETEAGENT' OR event = 'COMPLETECALLER'
THEN data2::int + data3::int END)) AS aht_seconds,
ROUND(AVG(CASE WHEN event = 'CONNECT'
THEN data1::int END)) AS asa_seconds,
COUNT(CASE WHEN event IN ('COMPLETEAGENT','COMPLETECALLER') THEN 1 END) AS calls_completed
FROM queue_log
WHERE time > DATE_TRUNC('day', NOW()) + INTERVAL '8 hours'
GROUP BY queuename;AHT alarm: ±15% from baseline is normal. 15-30% deviation is a warning. Over 30% means something changed — new agents, system issue, or a complex call type flooding the queue.
My take: AHT obsession kills call centers. I have watched supervisors pressure agents to cut calls short, which tanks first call resolution and increases repeat contacts. Show AHT for awareness, not as a target. The metric you should actually care about is whether AHT is lying to you.
6. Last-Hour Trend Chart — Direction Matters More Than Position
A simple line chart showing queue depth, SLA, and abandon rate over the past 60 minutes. Not absolute numbers — direction.
One bad minute is not a problem. Fifteen minutes of steady decline is a trend that needs intervention.
My take: This is the widget most dashboards get wrong. They show a 24-hour view or a week view. Useless for a supervisor making decisions right now. The window should be exactly one hour, updating every 5 seconds. Anything longer and you are looking at history, not making decisions.
The Dashboard Mistakes I See Everywhere
Tired of guessing what's happening in your queues?
Astervis gives you 30+ real-time charts, operator KPIs, and CRM integration for your Asterisk PBX. Self-hosted. Install in 5 minutes. From $119/mo flat — unlimited operators.
Too Many Widgets
A dashboard with 20 widgets is not a dashboard — it is a report someone put on a screen. The "command center" fantasy looks impressive in vendor demos. In practice, supervisors glance at the screen for 3 seconds between conversations. If they cannot assess the situation in those 3 seconds, the dashboard has failed.
Rule: 6-8 widgets on the primary view. Everything else goes on drill-down screens.
No Color Coding
"Queue: 12" — is that good or bad? Depends on time of day, staffing level, and call type. Numbers without context are decoration.
Rule: Every metric gets green/yellow/red thresholds. Configurable per queue, per shift, per day-of-week.
Minute-Interval Updates
A minute is an eternity in a call center. In 60 seconds, queue depth can jump from 3 to 20.
Rule: 3-5 second refresh for critical metrics (queue, agent states). 15-30 seconds for aggregates (SLA, AHT). If your monitoring tool refreshes once per minute, it is not real-time.
No Audio Alerts
A dashboard is useful when someone looks at it. Supervisors cannot stare at screens continuously — they are coaching agents, handling escalations, answering questions.
; Configure Asterisk to trigger alerts via AMI events
; extensions.conf — send manager event when queue exceeds threshold
[queue-monitor]
exten => s,1,Set(QUEUE_WAITING=${QUEUE_WAITING_COUNT(support-queue)})
exten => s,n,GotoIf($[${QUEUE_WAITING} > 15]?alert)
exten => s,n,Hangup()
exten => s,n(alert),UserEvent(QueueAlert,Queue:support-queue,Waiting:${QUEUE_WAITING})
exten => s,n,Hangup()Rule: Audio and visual alerts when red thresholds are crossed. Configurable per supervisor so night shift does not get daytime thresholds.
The Red-Screen Playbook
When the dashboard goes red, here is the decision tree every supervisor should follow:
| Signal | Check First | Action |
|---|---|---|
| Queue growing | Agent statuses | Too many on break? Recall. All on calls? Pull outbound agents. |
| AHT spiking | Recent call topics | Complex issue trending? Prepare a script or escalation path. |
| Abandon spike | Technical systems | Lines down? Routing broken? Check trunk status. |
| SLA dropping | All of the above | Multiple causes compound. Fix the biggest contributor first. |
-- Quick diagnosis: what happened in the last 15 minutes?
SELECT
event,
COUNT(*) as event_count,
queuename
FROM queue_log
WHERE time > NOW() - INTERVAL '15 minutes'
AND event IN ('ABANDON', 'CONNECT', 'ENTERQUEUE', 'EXITWITHTIMEOUT',
'RINGNOANSWER', 'COMPLETECALLER', 'COMPLETEAGENT')
GROUP BY event, queuename
ORDER BY event_count DESC;This query is your 15-second diagnostic. Run it when the dashboard turns red and you immediately see: are calls entering faster than connecting? Are agents timing out on ring? Are customers abandoning or exiting with timeout?
For the complete troubleshooting playbook including call drop diagnosis and trunk monitoring, keep those guides bookmarked.
What a Proper Wallboard Setup Looks Like
The supervisor dashboard and the team wallboard serve different purposes:
| Supervisor Dashboard | Team Wallboard | |
|---|---|---|
| Audience | Supervisor at their desk | Entire team on a large screen |
| Detail | Agent-level, per-queue drill-down | Team-level, motivational |
| Refresh | 3-5 seconds | 5-10 seconds |
| Widgets | 6-8 detailed | 3-4 large, visible from 20 feet |
| Alerts | Audio + visual | Visual only (color changes) |
The wallboard shows: current queue depth, team SLA, and a leaderboard. The supervisor dashboard shows everything else. Do not put 6 widgets on a wallboard — nobody can read them from across the room.
Building This With Astervis
Astervis gives supervisors all six widgets out of the box:
- —Agent status board with real-time state changes (3-second refresh)
- —Queue depth and wait time with configurable color thresholds
- —Live SLA tracking per queue, per shift
- —Abandon rate monitoring with trend overlay
- —AHT and ASA with baseline deviation alerts
- —60-minute trend charts for all key metrics
Plus: audio alerts on threshold breaches, drill-down from dashboard to individual agent metrics, and automatic KPI tracking without manual SQL queries.
No Grafana configuration. No custom queries. No Java dependencies like QueueMetrics.
The Bottom Line
A supervisor dashboard should answer one question: do I need to act right now?
Six widgets. Color coding. 3-second refresh. Audio alerts. That is it.
Everything else — historical reports, agent scorecards, weekly trends — belongs on separate screens for separate conversations. The main dashboard is a cockpit instrument panel, not a business intelligence tool.
If your supervisors are ignoring the dashboard, the dashboard is wrong. Simplify until they look at it.
Related reading:
- —How to Build an Asterisk Call Center Wallboard — the team-facing companion to the supervisor dashboard
- —How to Monitor Asterisk Queues in Real-Time — the queue data that feeds your dashboard
- —5 Call Center KPIs That Actually Drive Revenue — choosing which metrics deserve dashboard space
- —Call Abandonment Rate: Why the 5% Benchmark is a Lie — rethinking your abandon rate thresholds
- —Your Average Handle Time Is Lying — why AHT on your dashboard might be misleading
- —First Call Resolution: Why Your 70% Target is Wrong — the metric your dashboard is probably missing
- —Asterisk Operator Performance Tracking — agent-level metrics behind the status board
- —Fix Asterisk Call Drops — when your dashboard shows sudden abandon spikes
Stop guessing. Start monitoring.
See your Asterisk call center's real performance — queue wait times, agent activity, trunk usage, and 30+ charts. Self-hosted on your server. Install in 5 minutes. No credit card required.
From $119/mo flat. Unlimited operators. 14-day free trial.