Codifying meeting culture: using AI video analytics to assess your performance
For many amongst us, it’s been a summer of Zoom and Teams meetings; followed by much remote work procrastination and getting the laundry done. The way we work and the way we communicate has changed. Irrevocably.
Just three-days ago (5 Oct 2020), Microsoft announced that it would allow Teams meeting recordings to be shared externally. This offers myself and fellow data scientists a rich new source of content to analyse. With multiple person conference bridge calls, it was impossible to use any AI tools to analyse the audio recording. But with a video stream, the use of facial recognition analytics and speaker separation, we can now discern between the different parties. Soon, your engagement, conduct and performance during the virtual meeting will be codified by an AI engine and presented on a dashboard for all management to see.
Don’t think so? The desire for corporate surveillance by Global500 companies is huge as they become desperate to get a grip on the new reality of the home working environment. Zoom has been pilloried for its “attendee attention tracking” feature but very few administrators disable this function.
Dazzled by the dashboard
A select few organisations already assess meeting culture by feeding the Zoom meeting output through an analytics engine or using a dedicated video meeting platform (like team.video). The problem is, they are measuring the wrong things.
As a conversation scientist, I get excited about measuring the behaviours that contribute to better outcomes. What I’ve seen so far are companies measuring speaking time, meeting start time and use of emoticons. Trite behaviours that do not truly reflect on culture; nor do they tackle the real underlying meeting issues.
The real behavioural issues
Harvard research, based on face to face meeting observation, suggests women are twice as likely to be interrupted as men. This is known as the “Shut Out” a term coined by Professor Neil Rackham when he codified the science of effective meetings. A shut out is rarely beneficial in a meeting; except when done by the appointed chairperson to intentionally close down a conversation. You would expect a strong Chair to shut out dominant individuals and bring in people that are struggling to get their voice heard.
Another meeting behaviour to measure is the volume of “Counter Proposals” This is observed when a creative idea is deliberately shot down and met with strong resistance. For example, “What we should do is X and not your idea Dave.” (poor Dave). I once observed a nine-hour long bid strategy meeting at an $18bn aerospace/defence company. There were 187 counter-proposals during the day as the room of 18 people couldn’t agree on a single thing. They didn’t win the $1bn contract. The meeting behaviours were an observable predictor of failure.
Rather than sitting through a tortuous full-day meeting, what’s really exciting is the ability to codify and assess the culture of the meeting using AI Video Analytics.
There aren’t many providers out there right now. I’d suggest looking at XDroid who will shortly offer video analytics in their AI platform. Zoom has a built in feature to export to otter.ai; another analytics provider. TMAC is another firm with this capability. I suspect as working from home dominates our future, you’ll be seeing a lot more of Video analytics and many fancy dashboards too! If there are other providers out there, drop me a message and I’ll post them up too.
