A participation score is not a voice

Read AI's support documentation describes a Read Score that combines two measures. Sentiment classifies reactions as positive, neutral or negative, using facial expressions, head movements, body language and the words people use. Engagement rates involvement as high, medium or low, using facial and verbal cues, vocal pitch, volume, intonation and proportional talk time. Scores above 80 are graded good and below 70 poor. The documentation also says that for users in the EU and UK the score leaves out video and facial elements, and that all users can switch these scores off.

No. Talk time counts seconds. Equity of voice is whether each person had a real chance to affect the outcome. A person who speaks for ten seconds and changes the decision had more voice than someone who spoke for ten minutes and was ignored. A facilitator who optimises for even airtime gets a room that takes turns politely and decides nothing. Use the number as a prompt to look, not as the definition of success.

Plenty. A quiet face may be concentration, a bad connection, a cultural norm about speaking or a person reading the chat. Flat vocal tone may be how that person speaks. A neurodivergent participant may look disengaged and be fully tracking. Reading a face or a voice for an emotion is an inference, and the inference will be wrong in a predictable subset of people. I wrote about the same trap in camera off is not disengagement.

California now has rules on how some workplace monitoring tools may be used, and I cover what they do and do not reach in AB 1883 and your engagement dashboard. The short version for a facilitator is to find out what your platform computes about participants before you turn it on, tell them, and keep it away from performance decisions.

Use structure. Start with a round where everyone answers a short question in turn. Use chat and silent writing before discussion so the first ideas are not the loudest. Call on people by name only after you have given them a way to pass. Split into pairs when the group is large. These moves change who speaks, which is what the dashboard can only observe. I covered the chat side in the chat box is not a participation strategy.

After the session, at the level of the meeting and not the person, as one input to a conversation with the group. "The summary says three people covered most of the time. Does that match how it felt?" is a fair question. "Your engagement score was low" is not. I would not show a live score to the room at all. Being measured in real time changes how people behave, and it is not the behavior you were hoping to measure.