Wholesale buyers compare routes on four measurements: answer seizure ratio (ASR), average call duration (ACD), post-dial delay (PDD), and mean opinion score (MOS). Each one measures something different, and no single figure tells you whether a route is good. This guide defines each metric, gives the formula, explains what it can and cannot show, and lists what to ask a provider when they quote one.
ASR: answer seizure ratio
ASR is the share of call attempts that are answered. It is calculated as:
ASR = answered calls ÷ call attempts × 100
Every attempt counts, including calls that ring out, reach a busy line, or hit a disconnected number. ASR therefore reflects both the network and the people being called. A low ASR can mean a congested route, or it can mean a list with many dead numbers. Read ASR together with the destination and the traffic type before judging the route.
Published guides disagree on what a healthy ASR looks like. Sippy Software, a switch vendor, gives roughly 50 percent as a typical target for international wholesale SIP traffic. Softtop, a wholesale provider, cites 60 to 80 percent for premium CLI routes. Both are vendor statements, not regulatory standards. The useful comparison is your own ASR on one destination against another route to the same destination.
ACD: average call duration
ACD is the average length of answered calls:
ACD = total duration of answered calls ÷ number of answered calls
ACD is measured in seconds. It is shaped by the traffic more than the route. Call-center traffic usually runs short, while a conversation-based customer may run long, so a route should be judged against the ACD its traffic normally produces. Sippy describes wholesale SIP traffic as expecting around 300 seconds, and call-center traffic as commonly 60 to 120 seconds. These are rough guides, not targets.
ASR and ACD together show more than either alone. A healthy ASR with a very low ACD suggests calls that connect and then drop almost at once. A low ASR with a normal ACD suggests the route is connecting few calls, but the ones it connects last as expected.
NER and why it differs from ASR
Network effectiveness ratio (NER) counts a busy line or a ring-no-answer as a success, because the network did deliver the call to the terminal. ASR does not. When a provider quotes one of these figures, confirm which one it is. The earlier guide on network effectiveness ratio explains the difference in detail.
PDD: post-dial delay
PDD is the time between dialing and the caller hearing ringing or an answer. Long PDD is noticeable to people on the call, and it can also hurt dialer performance, because agents or automated systems wait longer per attempt. The post-dial delay guide covers what drives PDD and how to reduce it. As a rough reference, delays above about 5 to 6 seconds are commonly described as noticeable, which is a rule of thumb, not a standard.
MOS: mean opinion score
MOS is a 1 to 5 rating of perceived voice quality. The ITU-T P.800 scale defines five categories:
- 5: excellent
- 4: good
- 3: fair
- 2: poor
- 1: bad
Subjective MOS comes from listeners rating calls. Most carriers cannot run listening tests on live traffic, so they use an estimated MOS calculated from network measurements: packet loss, jitter, latency, and the codec in use. An estimated score is only as good as the measurements behind it, so ask how the figure is produced. A route that reports only a high MOS figure, without loss, jitter, and latency numbers, gives you less to check.
Codec choice changes the estimate. G.711 and G.729 trade bandwidth for quality, and an estimate should state which codec it assumes. The codec guide on VoIP codec mismatch covers the failure mode where the two ends disagree.
Measurement details that change the answer
- Sample size: an ASR calculated from 20 calls is not reliable. Our route quality testing guide uses a minimum of 100 calls per destination prefix before trusting a figure.
- Per prefix or blended: a blended ASR across many countries hides a weak destination. Ask for figures per destination.
- Time window: a figure for one week can differ from a figure for one month. Ask what period the figure covers.
- Route class: CLI, NCLI, and CC routes carry different traffic, so their ASR and ACD are not directly comparable.
Questions to ask a provider
- Which ASR and ACD do you report for each route class, and over what period?
- Is the figure per destination prefix or blended?
- What PDD do you measure, and how is it calculated?
- Is the MOS measured on live traffic or estimated, and from which network inputs?
- Do you report NER alongside ASR, and how do you define a successful attempt?
- Can I see the raw CDRs for my own test calls?
Where to go next
Quality metrics only matter next to price, which depends on billing increments and payment terms. For a test plan that measures a route before you move traffic, see the wholesale route quality guide.