Blog Post

Microsoft Teams Blog
2 MIN READ

Performance enhancements to Microsoft Teams lead to faster response times

Mark_Longton's avatar
Mark_Longton
Icon for Microsoft rankMicrosoft
Jun 02, 2022

Over the past year we’ve delivered enhancements to Teams that improve its overall interaction responsiveness time and creates a more fluid experience for the user. Investments have included transitioning from Angular framework to React, upgrading Electron (a framework for building desktop applications), reducing re-rendering, and making incremental improvements to the code. Our desktop, framework, and performance teams made several foundational improvements and our messaging and calling/meeting teams partnered to optimize the code for targeted user experiences we identified as important to the overall experience.


To gauge our progress, we recently looked at anonymized data from the 95th percentile of all desktop users in the world (meaning that 95 percent of the time the experience is better than this metric). We tend to focus on the 95th percentile because it includes users on low end devices, users on low bandwidth networks, and incorporate other edge cases that can impact the user experience.

 

The data showed notable improvements in messaging and meeting experiences as seen today vs. August 2021.

 

Improvements in latency and page load times for messaging

When users scroll over the chat list, latency has improved by 11.4%, and scrolling over the channel list has improved by 12.1%.


The compose message box loads 63% faster, enabling the user to type a message immediately once they switch into a chat or channel.


Page load times are much shorter as well:

  • The time to switch to a channel and to open a chat window-both were dramatically improved by 25%
  • Switching threads in the activity feed has improved by 17.4%
  • Switching between chat threads has improved by 3.1%


Greater fluidity and reduced lag in meetings

The mute and unmute audio response during a call improved by 16%.


Navigating to the ‘Pre-meeting join’ screen is 9% faster.


Opening a calling/meeting window loaded 4.5% faster. Then once a user is in a meeting switching into a chat improved by 13%, switching to the activity feed improved by 18.7%, and switching to a channel improved by 20%.

 

 

These improvements complement the power reduction improvements shared in the recent blog "Microsoft Teams performance improvements reduce power consumption in meetings by up to 50%."


We continue to focus and invest in Microsoft Teams performance optimization and value the customer and partner feedback that helps us to prioritize the most critical scenarios.

Updated Jun 07, 2022
Version 2.0
  • ediflyer's avatar
    ediflyer
    Brass Contributor

    rbrynteson that's really helpful and makes sense, thanks. I never use the personal version, only the work one so I can't wait to see more improvements in efficiency. I find when I'm in a call my GPU usage is at about 20% (CPU 9% or so, Core i7) and it really hits the battery - about 100% to 80% after a one hour call (albeit with screen brightness up a fair whack too), anything to cut that back would be amazing (and mean the fans didn't kick in as much too!). 

  • Hi Mark_Longton , the pictogram shows "Open call/meeting +63%", but the sentence says "Opening a calling/meeting window loaded 4.5% faster". Which is correct?

     

  • mikepiff's avatar
    mikepiff
    Brass Contributor

    Does anyone know from which version of Teams these improvements will be noticeable? 

  • PatrickDahlinSW's avatar
    PatrickDahlinSW
    Copper Contributor

    Moving from skype, the performance has degraded noticeably. Taking seconds to load in a channel fully and overall sluggish UI.

     

    A simple thing such as sending a text to a bulletin-board should be quick and responsive. Everything else is second to this, yet it has become an overall "meh" experience. If this app is to compete in any shape or form with something alike to Slack, the performance *needs* to go from seconds to sub-second loadtimes on medium-end hardware.

     

    Oh and btw, these improvements are a great step in the right direction, but percentage improvements tell me nothing. 11% can be anything from years to milliseconds, and as hardware always differ from oneanother, example timings and hardware is much more intuitive.

  • Fabrice_R's avatar
    Fabrice_R
    Copper Contributor

    Hi Mark_Longton 

    Thank you for sharing these performances enhancements. I would be able to measure and demonstrate them to my end users but I have no tools for that.

    Could we have nominal average delays to expect on a few test protocols including Teams version, OS version, hardware and network connection in place of percentage ? This way, we'll be able to compare ourselves and to argue for an investigation on our company context.

    And we really need timeline visibility about the next release bringing performance improvements.