Glossary Call quality
What is MOS?
Also called mean opinion score, call quality score
MOS, or Mean Opinion Score, rates call quality from 1 to 5. It began as a genuinely human measure, with listeners scoring recordings, and is now usually calculated automatically from latency, jitter, packet loss and the codec in use. It is the standard way to express "how did that call actually sound" as a single number.
Reading the scale
A score above 4.3 is excellent and better than a traditional phone call. Around 4.0 is good and what a well-run VoIP service should deliver. Around 3.6 is where users begin to complain unprompted. Below 3.1 is bad enough that people avoid using the phone. Note that the scale is capped by the codec: a narrowband call cannot score at the top no matter how perfect the network.
What it is useful for
A single call's MOS is not very interesting. MOS is valuable as a trend across many calls, because it turns "the phones seem worse lately" into something measurable, and it lets you correlate degradation with a time of day, a site or a particular link. Treat it as a smoke alarm rather than a diagnosis.
Where MOS misleads
Two cautions. First, an averaged MOS hides the calls that matter: a hundred perfect calls and five terrible ones average out to something that looks acceptable, while five customers had an awful experience. Look at the distribution, not just the mean. Second, MOS is calculated from network measurements and knows nothing about the room, so it will happily report an excellent score for a call ruined by a cheap headset, a speakerphone in an echoing space, or background noise.