The voice quality measurement that no longer makes sense

Mean Opinion Score has been used for years as a way of gauging whether or not communications technologies are providing “good enough” service. But is “good enough” still good enough?

Share this article:

“I’m sorry, we can’t quite hear you.” Groan. It was one of those conference calls.

Even though it wasn’t me they were having a hard time making out, I felt immediately embarrassed, somehow, for the person being singled out. And the person doing the singling out was coming through pretty faintly himself, at least on my end. Overall, it was a miracle this discussion was happening at all. Mean Opinion Score doesn’t have a number which can capture this kind of frustration, even though it happens to almost everyone almost all the time.

Most enterprise users have probably never heard of Mean Opinion Score (MOS), unless they have been on the inside of a SIP trunking deployment or similarly transformative unified communications project. If they did, they would realize how absolutely ridiculous and irrelevant it has become. Perhaps the ITU Telecommunication Standardization Sector (ITU-T) was just trying to be helpful when it first suggested have someone sit in a quiet room and subjectively measure call quality, but the rise of mobility and the consumerization of IT have raised expectations far higher than anything on the MOS scale:

Mean opinion score (MOS)

MOS Quality Impairment
5 Excellent Imperceptible
4 Good Perceptible but not annoying
3 Fair Slightly annoying
2 Poor Annoying
1 Bad Very annoying


I recently came across MOS again when I was looking through a white paper on how to make the business case for SIP trunking. It included the following comment:

Voice quality is subjective to the caller. Some people have grown accustomed to repeating themselves and working hard to listen to what a caller is saying. Others, who make their living by talking to people, require that voice quality be as good as if the communication were face-to-face.

Well, I agree with the “subjective” part. As for the rest, who has grown accustomed to repeating themselves and working hard to listen to what a caller is saying? Only people who are probably fed up enough with their organization (or their telecom provider) to think about jumping ship.

As for the second point, I think you could argue that today, we all make our living by talking to people. This is no longer confined to call centre employees, or even writers like me, but anyone who needs to connect with a customer, a project team or a supplier. In other words, everyone. To suggest that we aim for less than face-to-face quality in voice is to set the bar too low.

Of course, there will be some who argue that shooting for 100 percent quality isn’t feasible, particularly in SIP trunking environments where voice codecs and compression sometimes have an impact on call quality. When technology is brand new, I think the tolerance for some quality issues is higher. In the early days of cell phones, for instance, the tradeoff between a slightly jittery call and being able to make that call outside the office was an acceptable tradeoff. Probably not any longer.

Instead of MOS, I think maybe IT departments should consider the Wong-Baker FACES pain rating scale used by health professionals. It’s equally subjective, but the visuals are more helpful than mere numbers:


See No. 4, where the mouth goes flat? That’s about how low voice quality can go. Anything closer to a grimace than that and you can be sure your communications infrastructure is negatively contributing to employee satisfaction, engagement and productivity. And believe me, no matter how bad the connection, users will be loud and clear when they tell you about it.

Learn a new way to manage voice by downloading SIP Trunking eBook: Complete Buyer’s Guide (Expanded 2013 Edition) from Allstream

Image courtesy of artur84 at

Share this article:
Comments are closed.