This article originally appeared in the August 2010 issue of Unified Communications magazine.
Every engineer, product manager, salesman, project manager, tester, and really – everyone – knows that quality matters.
Quality is a differentiator between products in almost every market. It brings customers in, and more importantly, it keeps them from leaving. Indeed, quality is so central that companies often use it as part of their core branding. Ford’s branding is: Quality is Job 1. And Zenith’s is: The quality goes in before the name goes on.
In the telecommunications industry, quality can be measured in two ways. The first is uptime – whether or not a user is able to place a call. The famous five nines of reliability for telecom gear is directly related to the desire to provide continuous uptime. For landline phones, this uptime manifests directly to the consumer as dial-tone, giving them immediate feedback that the system is up and ready for a call. In the wireless world, coverage has replaced continuous dialtone as the way users think about availability. Instead of picking up the phone and listening for dialtone, users look to see if they have enough bars to make a call. The mobile operators have worked aggressively to improve their coverage and promote it when their coverage is better than the competition.
However, uptime is only one measure of quality, indicating only that a call can be placed. Once the call is placed, the intelligibility and fidelity of the speech is another way to measure quality. If the voice is unintelligible, it is nearly as bad as not being able to place the call in the first place. Yet, despite its importance, voice quality has seen little improvement over the years.
Landline networks have been operating with the same voice quality for decades. The public switched telephone network is locked to running narrowband, mono voice using the G.711 codec, making improvement extremely difficult. On wireless networks, the scarcity of access bandwidth has caused speech quality to take a step backwards. Even a perfect cellular call has noticeably worse quality than a landline PSTN call, and cellular calls are often far from perfect. Users have become used to fade-outs, drops, and robotic voice on cellular calls.
Yet, these quality issues have not stopped users from using mobile phones. Quite to the contrary, mobile phones have increasingly replaced landline phone usage, despite the decrease in voice quality. Perhaps, when it comes to voice, quality doesn’t matter?
As it turns out, it does matter. Skype (News - Alert) recently did a study on its user base to examine the impact of speech quality on user behavior. We selected a random subset of our users to participate in a test. The participants were completely unaware that the test was being run. For users in the test, we modified behavior of their client so that it always picked a certain codec, independent of available bandwidth. We then measured the MOS (mean opinion score) that these users provided to us in a post-call splash screen. We also measured the amount of time users spent on the call itself.
Different codecs were used in the calls. SILK is Skype’s codec, which we have recently contributed to the IETF and made available to the public. It supports four different modes of operation, ranging from narrowband (8 kHz sampling), to medium band (12 kHz sampling), to wideband (16 kHz sampling), to super-wideband (24 kHz sampling). G.729 is an industry standard narrowband codec, using heavy compression with 8 kHz sampling. In general, as you move from left to right on the vertical axes, fidelity of the speech improves.
The users included in the survey talked on a call for approximately 40 percent more time when the codec was SILK-SWB, compared to G.729. The increased talk times show that call quality directly impacts the way users interact with the system. The higher the quality, the longer they are likely to talk. The longer they talk, the more they use the service. Because these results were obtained without user awareness, it demonstrates that this effect is not just perceived, but real.
Voice quality does matter.
Jonathan Rosenberg is chief marketing strategist at Skype (www.skype.com).
TMCnet publishes expert commentary on various telecommunications, IT, call center, CRM and other technology-related topics. Are you an expert in one of these fields, and interested in having your perspective published on a site that gets several million unique visitors each month? Get in touch.
Edited by Stefania Viscusi