It brings benefits to workplace teams in terms of travel requirement, space issues, and scheduling difficulties. but can also have dire impact on meeting effectiveness and inclusivity.

We explore how to make meetings more inclusive and effective through intelligent feedback dashboard.

First, we conducted a requirement analysis study collecting data from actual workplace employees.

We found that attendees might perceive a meeting to be better based on meeting behaviors, such as more participation.

Next, we categorize the challenges attendees face in remote meetings.

From there, we identified which challenges we could address through a feedback dashboard.

And finally, we explored what concerns employees might have regarding such a feedback system.

Participants expressed concerns regarding how accurate the AI system could be, and with whom the feedback data might be shared.

In the second phase, we conducted a longitudinal study with actual workplace teams.

The teams had recurrent meetings of our Microsoft Teams. A custom bot joined each meeting and recorded it.

From video, the facial signals provided feedback features, such as head nod, leading to consensus measurement.

From audio, the vocal activity provided features such as turn taking and participation.

Audio was also processed to create meeting transcript, which was used for features such as question detection.

Using the actual meeting data, we processed and visualized those features and created two dashboard prototypes for each individual meeting attendee.

Initially, there were 13 features in the wireframe prototype.

On one side, we have the summarized feedback, and the temporal feedback were shown on the right side.

For example, summarize score of talk time.

we created a list of suggestive feedback to be shown with the corresponding summarized score.

After the fourth week, participants reviewed the interactive dashboard. It had more information visibility and interactive functionalities.

For example, hovering would reveal more insights, suggests who asked a question, and what was being asked.

Clicking on an event would play the meeting recording of that timestamp helping people recollect that part of the discussion.

Participant could successfully identify when consensus was being reached and when more questions were being asked.

Suggestive feedback was useful nudging participants towards the desired behavior for the future.

Meeting history over the time was useful to identify changes in behavior patterns over a long term of meetings.

Our findings show that such a feedback dashboard can be effective in making meeting behavior more transparent for an individual.

We provide design suggestions for future systems incorporating similar kinds of effective and actionable highlights for video conferencing based meetings to make teams more effective and inclusive.