Alphabet Soup:  Voice Quality & Your Customers

You can't be surprised.  It happens every time there's even the smallest evolutionary change in technology.  The industry is bombarded with another set of acronyms.  All the old and new acronyms start popping up all over the place -- in your favorite industry mags, at trade shows, during impromptu watercooler discussions with your favorite company techie.  Voice quality is
no different...QOS, QOE, VSQ, VQ, MOS, EMOS, PSQM, MNB, PAMS, PESQ, and the list goes on.   In the fog of acronyms and excitement of new technologies, it is easy to see why some of us temporarily lose sight of the ultimate objective of voice quality, communications and customer service efforts: making sure communications are intelligible to real people and reducing the effort these real people have to expend to get what they need.

There are many technologies and methods for measuring voice quality that help us get closer to determining how people perceive voice quality.  Your communications infrastructure may already include devices that measure some of the variables that impact voice quality.  These devices might be "talking" to other devices trying to figure out if one "heard" things more clearly than the other.  If the devices detect something isn’t working as expected, you might have to fix a gateway or tweak the network.  But what do these attempts to quantify or recreate a subjective opinion mean for your customer?  Does a voice quality rating really reflect what the human brain perceived?  If the device shows a Mean Opinion Score of 4 to 5, does that really mean your customer's experience was a good one?  If the score is 2 to 3, was your customer really annoyed or irritated?  Does she blame you or her cell phone for the dropouts?  Was she able to understand and complete her transaction?    Not every voice quality issue is 100% detectable by every technology and not every voice quality issue is a problem for the end-user.  Ultimately, it is the human brain that matters most in determining the quality and more importantly the intelligibility of a telephone call and all the implications that might entail for a business.  “It is from the perspective of the end user that voice quality is best defined” (Anderson, p 4).  

So how do you make sense of the voice quality alphabet soup when what you really care about is that your customers get good service and your company efficiently turns a decent profit? 

Focus on the Customer…Not the Scores
There are many ways to find out if your customers can access and understand the communications supported and delivered by your contact center and communications technologies. 

• You can attempt to survey some or all customers about what they’ve experienced.
• You can review data from internal monitoring mechanisms that attempt to extrapolate customer experience based on information gathered from various components of your infrastructure. 
• You can review call center logs for customer and agent complaints. 

But missing from this list is a methodology that allows you to observe and manage the customer experience as it relates to contact center and communications technologies from the critical outside-in perspective without burdening real customers – remote communications technology performance testing and monitoring that focuses on call intelligibility from the end-user perspective.  This cost-effective testing methodology allows you to leverage the efficiency of voice quality methods and technologies from the all-important end-user perspective.  In addition to the usual design, integration and time delay issues, remote performance testing and monitoring helps detect voice quality issues that can be most troublesome from the end-user intelligibility perspective including things like:

• Choppy speech
• Partial words
• Echo
• Talking over each other because of a dropout
• Too loud/too soft
• Delay and jitter
• Humming, white noise, and other burst noise in the background

There are several ways remote contact center performance testing and monitoring can be used to meet a company’s unique business requirements. Most VoIP implementations need to run on local area networks configured to support voice prioritized over data to be minimally acceptable as a voice communications tool.  But prioritization of voice in itself doesn’t guarantee high voice quality.  In addition to the basic voice quality metrics available from such a network infrastructure, many 3rd-party offerings are available to optionally increase the load on the network and monitor its performance under injected synthetic load as well as on a day-to-day basis.  Some companies prefer to implement testing and monitoring strategies that do not require any additional on-premises installation of hardware or software to measure voice quality.  These companies keep it simple by leveraging robust test call data and recordings that are collected remotely. The testing and monitoring technology pinpoints potential design, integration, time delay and/or voice clarity issues and the audio recordings leave no doubt as to the impact on the end-user.  By keeping things non-invasive, these companies avoid the potential security and overhead issues that might be associated with adding something new to their network.

For unique business reasons, other companies require the integration of voice quality measurements (e.g., PESQ, MOS, PSQM, etc.) with the robust end-user perspective data that remote contact center performance testing and monitoring can offer.  These companies install onsite voice quality appliances to collect voice quality metrics, which are later consolidated with remote calling data to give a clear picture of contact center technology performance.  The integrated data pinpoints issues and reveals potential areas for improving voice quality.   By installing onsite devices, these companies can see what if any end-user impact results from the detected voice quality issues revealed by more traditional quality of service metrics including:

• Audio and Video Jitter/Latency
• Listening/Conversational R Factor
• Listening/Conversational MOS core
• P.862 Raw MOS
• Burst Packet Loss Rate (%)
• Average Burst Length (ms)
• Average Gap Length (ms)
• Comprehensive Bandwidth Analysis
• Comprehensive Activity Analysis
• Complete Breakdown by Protocols, Codecs
• Complete Voice, Video, and Data Summaries

Conclusion
Even though our world is driven by technology and flooded with acronyms, not everyone has the time or interest for the intricate details of voice quality and related technology.  However, if you take a step back and realize that voice quality isn't about technology but rather the ability to effectively communicate, the fog of acronyms begins to lift.  It all starts to makes sense when you re-focus on the customer experience. 

References
Anderson, John, “Methods for Measuring Perceptual Speech Quality”, Agilent Technologies, 2001.

Back