5 Metrics Every Outbound Call Center Manager Should Track to Diagnose AMD Problems in 2026

Your ViciDial AMD is hanging up on live customers right now. Not occasionally. Every single day. Legacy silence-based detection produces false positive rates between 15 and 20 percent — up to 1 in 5 real people never reaches an agent. At a 50-agent center, that's roughly $120,000 in lost revenue every month.

Most managers never see it clearly in their reporting. They see low contact rates, frustrated agents, flat conversion numbers. They blame the list or the time of day. The actual cause is sitting in their AMD configuration, and the data to prove it is already in their dialer logs.

These five metrics will show you exactly where your AMD is bleeding revenue.

1. AMD False Positive Rate

This is the number that matters most — and most ViciDial installs never surface it directly.

A false positive happens when your AMD classifies a live human as a voicemail. The call gets dropped or routed to a voicemail handler. The prospect never hears an agent. You never know it happened unless you dig into your disposition logs.

How to find it in ViciDial: Pull your call dispositions and filter for AMD-terminated calls. Cross-reference those against calls that connected for more than 4 seconds before AMD fired. Any call where AMD triggered after a human-length greeting pattern is a candidate false positive.

Legacy silence-based AMD measures silence duration and initial audio length. It cannot tell the difference between a human saying "Hello?" and a voicemail system playing a brief greeting. That ambiguity is exactly where the 15 to 20 percent error rate lives.

A healthy AMD false positive rate is under 2 percent. Above 5 percent, you are actively destroying your contact rate.

2. Live Connect Rate vs. Dial Attempt Ratio

Your live connect rate is the percentage of dial attempts that result in a real conversation. It captures AMD accuracy indirectly, but reliably.

When dial volume holds steady and live connect rate drops, AMD false positives are the most common cause. Other culprits — carrier spam flagging, bad list data — show different patterns. Spam flagging reduces answer rates. Bad data increases disconnects before AMD fires. A rising false positive rate kills connections after the call is already answered.

What to track: Divide live connects by total dial attempts for each campaign. Compare week over week. A drop of 2 to 3 percentage points without any change in list quality or dial time is a strong AMD signal.

For a 50-agent team running 2,500 dials per day, a 3-point drop in live connect rate means roughly 75 fewer real conversations daily. Over a month, that compounds fast.

3. Average Handle Time on AMD-Passed Calls

This one sounds unrelated to AMD, but it's a reliable diagnostic for false negatives — cases where a voicemail gets passed to an agent as a live call.

When an agent picks up a call AMD incorrectly classified as human, they typically hear a voicemail greeting mid-sentence or dead air. Handle time on those calls is short, usually under 15 seconds, and the agent marks it as a no-contact or drops it manually.

How to use it: Segment your handle time distribution. A spike of calls with handle times between 5 and 20 seconds marked as no-contact means your AMD is passing voicemails through to agents. That wastes agent time and inflates your abandon rate.

The goal is a clean bimodal distribution: very short calls AMD correctly routed away, and substantive calls AMD correctly passed through. A messy middle band is a red flag.

4. Ghost Call and Silent Call Rate

Ghost calls and silent calls are distinct problems, but both trace back to AMD configuration in ViciDial and Asterisk environments.

A ghost call happens when a call connects, AMD fires too slowly, and the agent hears silence before the call drops. The prospect experienced dead air and hung up. Silent calls follow the same pattern — the call connects, but no audio reaches the agent in time to start a conversation.

Why this matters for TCPA and Ofcom compliance: In the US, TCPA rules restrict abandoned call rates. In the UK, Ofcom requires that no more than 3 percent of calls connected to a predictive dialer result in silence or abandoned calls. An elevated ghost call rate isn't just a revenue problem. It's regulatory exposure.

Track ghost calls as a separate disposition in ViciDial. If your rate exceeds 2 percent, your AMD latency or CPD timing is off. The fix is either tightening your AMD thresholds or replacing the detection engine entirely.

5. Revenue Per Dial Attempt

This is the metric that ties everything together. It converts AMD accuracy from a technical problem into a dollar figure your entire team understands.

How to calculate it: Take total revenue from outbound calls in a given period. Divide by total dial attempts. Track it weekly.

If your AMD false positive rate is 15 percent and your average deal value is $200, the math gets uncomfortable fast. At 2,500 dials per day, 75 of those are live humans your AMD is terminating. If even 10 percent of those would have converted, that's 7.5 deals per day you never made.

This metric makes the business case for fixing your AMD impossible to ignore. It also gives you a clean baseline to measure improvement against after you make a change.

What to Do When the Numbers Are Bad

False positive rate above 5 percent. Live connect rate trending down. Revenue per dial flat or falling. If that's what your data shows, the root cause is almost certainly your AMD engine.

ViciDial's built-in AMD uses silence thresholds. It was designed when voicemail systems were simpler and audio patterns were more predictable. In 2026, carrier audio processing, voicemail system diversity, and call quality variation have made silence-based detection unreliable. Silence detection is a 2005 solution to a 2025 problem.

ML-based audio fingerprint analysis solves this at the detection layer. Instead of measuring silence, it analyzes the acoustic signature of the audio stream in real time and classifies it as human or machine with 99% accuracy. That closes the gap from a 15 to 20 percent false positive rate down to under 1 percent.

AMDY.IO integrates directly into ViciDial, Asterisk, GoAutoDial, FreePBX, Issabel, and 3CX with a one-line install. No platform migration. No new dialer. Your existing stack stays intact. The WebSocket API connects to any SIP-based dialer outside that list as well.

The math on the fix is straightforward. A 50-agent center recovering $120,000 per month in false positive losses pays back the cost of a new AMD engine within days, not quarters.

Run the numbers for your specific team size and deal value using the ROI calculator at amdy.io. Then start a 14-day free trial with zero setup fees and see the accuracy difference in your own call data.

FAQs

What is an AMD false positive in ViciDial? A false positive occurs when ViciDial's AMD classifies a live human as a voicemail. The call is terminated or routed away from agents before any conversation happens. Legacy silence-based AMD produces false positive rates between 15 and 20 percent.

How do I find my AMD false positive rate in ViciDial? Pull your call disposition logs and filter for AMD-terminated calls. Cross-reference those against calls that connected for more than 4 seconds before AMD fired. Calls where AMD triggered after a human-length audio pattern are likely false positives. Tracking this weekly gives you a reliable trend line.

What causes ghost calls in ViciDial? Ghost calls occur when AMD fires too slowly after a call connects, leaving the agent with silence before the call drops. This is a latency and CPD timing issue. It also creates Ofcom and TCPA compliance risk if your abandoned call rate exceeds regulatory thresholds.

What is a healthy live connect rate for an outbound call center? It varies by vertical and list quality, but a drop of 2 to 3 percentage points without any change in list or dial time is a strong signal that AMD false positives are rising. Track it weekly against a stable baseline.

How does ML-based AMD differ from ViciDial's built-in AMD? ViciDial's built-in AMD measures silence duration and initial audio length to guess whether a call reached a human or a voicemail. ML-based AMD analyzes the acoustic fingerprint of the audio stream in real time. The result is 99% accuracy versus the 15 to 20 percent false positive rate common with silence-based detection.

Can I add better AMD to ViciDial without replacing my dialer? Yes. AMDY.IO integrates into ViciDial with a one-line install. No platform migration, no changes to your existing dialer configuration beyond adding the AMD engine.

How quickly can I see ROI after improving AMD accuracy? At a 50-agent center losing $120,000 per month to false positives, revenue recovery starts immediately after deployment. Most teams see measurable improvement in live connect rate and revenue per dial within the first week.

Fix the Metric, Recover the Revenue

The five metrics above give you a complete diagnostic picture of your AMD health. False positive rate tells you the size of the problem. Live connect rate shows its trend. Handle time distribution reveals false negatives. Ghost call rate flags compliance risk. Revenue per dial converts all of it into a number leadership actually cares about.

If the numbers are bad, the fix is a one-line install away. Start your free 14-day AMD trial at amdy.io and run your own numbers against the baseline you built today.