Operata Dashboards display voice quality measured using the Mean Opinion Score (MOS).
What is MOS?
MOS is a measure of voice quality that originates from subjective measurement, where listeners would sit in a "quiet room" and score a telephone call quality as they perceived it - a score of 5 being excellent and 1 being poor.
Operata continuously measures data network performance for every second, of every call, for every agent; with the aim of calculating MOS score using the ITU-T G.107 standard, which gives the following call quality ratings:
The Amazon Connect Softphone (CCP) uses the OPUS Codec to convert speech to data, which is more tolerant of poor network conditions than when the MOS standard was formed, with this in mind Operata classifies voice quality vs the MOS score, as below:
WHAT'S GOING ON
The WHAT'S GOING ON? dashboard displays a timeline of calls and their voice quality as well as the number of calls with an Average of Good, Average or Poor MOS call quality score.
The timeline graph at the top shows several important pieces of information:
The stacked bar shows the number of calls and their quality, with data updated every time a call is completed.
The higher the stacked line the more calls completed in the period, useful to see how the number of calls impacts voice quality - are there network limitations?
The donut chart at the bottom half of the page shows the percentage of Good, Average and Poor Quality calls over the period and the dials show the number of Good, Average and Poor Quality calls over the period.
Note: The Operata platform disregards calls with a MOS score of 0 as this can only be caused where network performance data is not being captured, usually where two Softphones are active on the same device.
FIND OUT WHERE
Use the FIND OUT WHERE tab to drill down into the detail of voice quality and the associated network performance statistics.
Use the filters to view the voice quality for a specific Agent, Network Type, Location, Country, Agent IP address or even the Agent WAN IP address.
The donut charts in the middle will display Good, Average and Poor call quality calls by location. This is an easy way to see where there are issues.
The time series graph is identical to the one shown in WHAT'S GOING ON.
Below that is a table showing the Agents with the poorest MOS scores in the period, their browser type/version, and their IP address - all really useful to focus improvements.
Now scroll down the page!
Below the time series graph are graphs showing the Network Performance for the filters you have chosen above - you can now see exactly why the MOS score was so low.
The first graph shows the MOS score, below that RTT.
RTT is the Round Trip Time, the time it takes voice data to make the journey from the Agent softphone to your Amazon Connect instance and back, displayed in milliseconds. (sometimes called Latency) - the impact of high RTT is delay between speaking and being heard.
Keep scrolling down and you will see Packet Loss (the percentage of voice data lost when sent to/from your Amazon Connect instance and finally Jitter (Voice data packets are transmitted continuously, but may have differing delays before they arrive at their destination, this known as jitter - measured in milliseconds) - high packet loss and jitter impact actual voice quality.
GET THE DETAILS
If there are issues, having the call by call data will be invaluable to the IT Teams to discover root cause and fix more quickly.
GET THE DETAILS lists each call with all the data needed to identify the cause of issues.
Use the Time Range filter and Agent, Network Type, Location, MOS, Jitter, RTT and Packet Loss filters to get to the data you need.
All the data you have filtered on is shown, with the AWS Connect contactId that AWS will need to progress any issues.
The Agent External IP Address (agentExternalIPAddress) and Agent ISP (agentISP) shows where the agent connects to the public internet. Commonly, the public internet is only used for Softphone signalling (not Voice media)
Go to the History module to drill down into the second by second detail and export to CSV.
Selecting the + Icon next to the callContactId shows the call timeline (Operata Conductor is needed to see all this detail)
This dashboard is dedicated to NETWORK PERFORMANCE and shows the network performance data collected for every agent, for every call, everywhere.
The four dials show the Average MOS, RTT, Packet Loss and Jitter for the period.
Scroll down to see the network performance over time - this is the same data shown as you scroll down the FIND OUT WHERE tab.
MOS is a measure of voice quality calculated from the network performance details below - score of 5 being excellent and 1 being poor
RTT is the Round Trip Time, the time it takes voice data to make the journey from the Agent softphone to your Amazon Connect instance and back - measured in milliseconds. (often called Network Latency)
Packet Loss is the percentage of voice data lost when sent to/from your Amazon Connect instance
Jitter is the difference in delays before Voice Data packets arrive at their destination - measured in milliseconds.
This graph shows MOS over time, with an Agent and Site filter.
Use this data to track voice quality (MOS) over time and challenge the IT teams to explain long term differences and changes - a great view to drive improvement.
Operata collects agent events and softphone errors to provide valuable insights into whats happening at an agent level
Select the AGENT STATUS button below for more detail.
For the programmers using this tool - there is the Edit as Query DSL option, here you can create more complex queries - find out here..