Why Voice Metrics Matter
ASR, ACD, PDD, and MOS are the four core metrics for evaluating voice service quality. Whether you're selecting a SIP trunk provider, troubleshooting call quality issues, or optimizing call center efficiency, these metrics form the foundation for data-driven decisions. This article systematically explains each metric's meaning, calculation method, industry benchmarks, and optimization strategies.
ASR (Answer Seizure Ratio)
Definition & Formula
ASR is the core metric for measuring voice route efficiency — the percentage of initiated calls that are actually answered.
ASR = Answered calls / Total call attempts × 100%
For example, if you initiate 100 calls and 60 are answered, ASR is 60%. Note that 100% ASR is unrealistic — users may not answer, be offline, or be busy.
Factors Affecting ASR
- Route quality: Premium direct routes have significantly higher ASR than multi-hop transit routes
- Caller ID: Using a local caller ID can increase answer rates by 20%-40%
- Time of day: ASR is typically lower outside business hours
- Destination network: Landline ASR is usually higher than mobile (mobile users often reject unknown numbers)
- Route congestion: Insufficient capacity during peak hours reduces ASR
ASR Industry Benchmarks
| Region | Premium Route | Standard Route | Economy Route |
|---|---|---|---|
| North America (US, Canada) | 60%-70% | 50%-60% | 35%-50% |
| Western Europe (UK, Germany, France) | 55%-65% | 45%-55% | 30%-45% |
| East Asia (Japan, South Korea) | 55%-65% | 45%-55% | 30%-45% |
| Southeast Asia (Philippines, Vietnam, Indonesia) | 45%-60% | 35%-50% | 25%-40% |
| Middle East (UAE, Saudi Arabia) | 40%-55% | 30%-45% | 20%-35% |
| Africa (Nigeria, Kenya) | 35%-50% | 25%-40% | 15%-30% |
ASR Optimization Tips
- Use local DID numbers from the destination country as caller ID
- Configure multi-route redundancy with automatic failover when primary route ASR drops
- Avoid bulk calling during off-hours in the destination country
- Regularly clean number databases to remove invalid/deactivated numbers
ACD (Average Call Duration)
Definition
ACD reflects the average duration of all successfully connected calls. ACD is not inherently "higher is better" or "lower is better" — it must be interpreted within the context of your specific business scenario.
ACD = Total duration of all successful calls / Number of successful calls
ACD Reference Ranges by Use Case
| Business Scenario | Normal ACD Range | Warning Signs |
|---|---|---|
| Call Center (customer service/outbound) | 3-8 minutes | Below 2 min may indicate call quality issues causing drops |
| Voice OTP (verification code) | 15-30 seconds | Below 10 sec may mean TTS playback is incomplete |
| Voice notification | 30-60 seconds | Below 15 sec may mean user didn't hear full message |
| Enterprise internal calls | 2-10 minutes | Below 1 min requires route stability investigation |
ACD Anomaly Investigation
If ACD is abnormally low (excluding business-specific characteristics), possible causes include: poor voice quality causing users to hang up voluntarily, route instability causing call drops, or one-way audio issues. We recommend analyzing in conjunction with MOS and ASR for a comprehensive diagnosis.
PDD (Post Dial Delay)
Definition
PDD is the waiting time from dialing the last digit to hearing ringback tone (or the recipient's phone starting to ring). PDD directly impacts user experience — in voice OTP scenarios where users are already waiting on the interface, excessive PDD leads to abandonment.
PDD Tier Standards
| Tier | PDD Range | Best Suited For |
|---|---|---|
| Premium | < 2 seconds | Financial verification, high-value transactions, call centers |
| Standard | 2-5 seconds | General verification, login authentication, notification calls |
| Economy | > 5 seconds | Batch notifications, marketing calls (non-urgent) |
Causes and Solutions for High PDD
- Too many routing hops: Each intermediate gateway adds 50-200ms → Choose direct or fewer-hop routes
- Cross-carrier transfers: Signaling negotiation takes time → Choose providers with direct carrier connections
- DNS resolution delay: Slow SIP domain resolution → Configure local DNS cache or use IP addresses directly
- Network congestion: Insufficient bandwidth causes signaling queue → Upgrade bandwidth or implement QoS
MOS (Mean Opinion Score)
Definition
MOS is the standard metric for voice quality using a 1-5 scale (5 being highest): 5=Excellent, 4=Good, 3=Fair, 2=Poor, 1=Bad. MOS can be obtained through subjective testing or calculated algorithmically using ITU-T G.107 E-Model based on network parameters.
Codec MOS Comparison
| Codec | Bitrate | Theoretical MOS | Bandwidth (with overhead) | Best For |
|---|---|---|---|---|
| G.711 (PCMA/PCMU) | 64 kbps | 4.1 | 87.2 kbps | Maximum compatibility, quality-first |
| G.729 (AB) | 8 kbps | 3.7 | 31.2 kbps | Bandwidth-constrained, cost-sensitive |
| G.722 | 64 kbps | 4.1 | 87.2 kbps | HD voice |
| Opus | 6-510 kbps | 4.5+ | Variable | Next-gen communications, WebRTC |
Network Factors Affecting MOS
- Jitter: Recommended below 30ms; above 50ms noticeably impacts MOS
- Packet Loss: Recommended below 1%; above 3% significantly reduces MOS (approximately 0.5-1.0 points)
- Latency: One-way below 150ms has minimal impact; above 300ms noticeably degrades call experience
Practical Advice: For international voice routes, we recommend a minimum MOS of 3.8. If MOS drops below 3.5, user experience degrades noticeably. Cainiao Voice premium routes typically achieve MOS of 4.0+, standard routes 3.8+.
NER (Network Efficiency Ratio)
NER measures call completion quality at the network level, excluding user behavior factors (no answer, busy, etc.) and focusing only on network-caused call failures.
NER = (Answered + User Busy + No Answer + Unreachable) / Total Attempts × 100%
NER approaching 100% indicates high network transport quality (call failures are mainly due to user behavior). The gap between NER and ASR reflects the proportion of network-caused failures. If NER is much higher than ASR, the route itself is fine — low ASR is due to user behavior. If NER is also low, there's a network quality problem.
Regional Comprehensive Benchmarks
| Region | ASR | PDD | MOS | Reference Rate/min |
|---|---|---|---|---|
| North America | 60%-70% | < 2s | 4.0+ | $0.005-0.015 |
| Western Europe | 55%-65% | 1-3s | 4.0+ | $0.008-0.025 |
| East Asia | 55%-65% | 1-3s | 4.0+ | $0.010-0.035 |
| Southeast Asia | 40%-60% | 2-5s | 3.7+ | $0.012-0.050 |
| South Asia | 40%-55% | 2-5s | 3.5+ | $0.008-0.030 |
| Middle East | 35%-55% | 2-6s | 3.6+ | $0.025-0.080 |
| Africa | 30%-50% | 3-8s | 3.3+ | $0.030-0.120 |
| South America | 40%-55% | 2-5s | 3.6+ | $0.020-0.060 |
Monitoring & Alert Recommendations
Building a real-time voice quality monitoring system is essential for maintaining service levels. We recommend focusing on these dimensions:
| Metric | Warning Threshold | Critical Threshold |
|---|---|---|
| ASR | Below 50% | Below 40% |
| PDD | Over 5 seconds | Over 8 seconds |
| MOS | Below 3.5 | Below 3.0 |
| ACD anomaly | 20% drop vs 7-day avg | 40%+ drop |
| NER | Below 80% | Below 70% |
| Concurrency | Above 80% capacity | Above 95% capacity |
Cainiao Voice provides detailed CDR (Call Detail Records) and real-time API interfaces, supporting quality metrics by country, route, and carrier dimensions for easy integration into your own monitoring systems.
Frequently Asked Questions
Q: What is a good ASR for voice routes?
A: ASR varies by region. North America and Western Europe premium routes typically achieve 60%-70%, Southeast Asia 40%-60%, Middle East 35%-55%, Africa 30%-50%. Below 40% generally indicates route quality issues.
Q: What PDD is considered premium?
A: Premium routes should have PDD under 2 seconds. Standard routes 2-5 seconds is acceptable. Over 5 seconds noticeably degrades user experience. Voice OTP scenarios require PDD under 3 seconds.
Q: What does MOS 4.0 mean?
A: MOS uses a 1-5 scale where 4.0 indicates good voice quality comparable to traditional phone calls. G.711 codec achieves approximately 4.1, G.729 about 3.7, and Opus can reach 4.5+.
Q: How should I monitor voice quality?
A: Build a real-time dashboard monitoring ASR, PDD, MOS, and ACD. Set tiered alerts: ASR below 50% warning, below 40% critical; PDD over 5 seconds warning. Cainiao Voice provides CDR and real-time API access.
Q: What is the difference between ASR and NER?
A: ASR only counts answered calls, while NER excludes user behavior (no answer, busy) and reflects only network-level issues. High NER with low ASR means low answer rates are due to user behavior, not network problems.
Where to Apply These Metrics
Global Voice Architecture
Design route failover and regional capacity using ASR, PDD, MOS, and concurrent-call utilization.
Voice OTP and Notifications
Use PDD, completion rate, and short-call delivery metrics to protect verification user experience.
SIP TLS and SRTP
Separate quality drops from security negotiation issues by tracking TLS handshakes and SRTP status.
Voice Codec Comparison
Compare G.711, G.729, and Opus tradeoffs when MOS, bandwidth, and compatibility targets conflict.
Need Professional International Voice Routes?
Cainiao Voice provides premium SIP trunk services covering 200+ countries with ASR up to 60%-70%
Free Consultation