How to read the room in virtual meetings (when you can't see anyone)

You're 22 minutes into a Zoom call, sharing your screen, and half the cameras are off. Chat scrolled past something three minutes ago that you'd want to come back to. You ask "any questions?" and get nine seconds of silence, then "no, that all made sense." You don't know if it did.

The cues you'd pick up in person (a confused frown, a sideways glance, a restless back row) don't survive the medium. They happen on cameras you can't see, or not at all because the attendee is in another window. This guide covers the signals that get lost online, four ways to surface them and which fits which session.

Why it's harder online

You can't see faces. Most cameras are off, and when they're on you're presenting, so you're looking at slides rather than the participant strip.

Chat and audio are two separate conversations. A question typed at 18:32 is gone from view by 18:34 if anyone else types. A spoken question you half-answered stays in nobody's memory unless you wrote it down.

People are multitasking. Calls compete with email and Slack. Silence in a room usually means attention; silence online can mean agreement, confusion or someone reading something else, and you can't tell which.

The signals that get missed

All five are recoverable if you notice them in time. Usually you don't.

Four approaches

These aren't exclusive, but each fits some sessions better than others.

1. Read it manually

Glance at chat every minute, tab between slides and the participant grid, and jot down questions in a notepad. This works for a five-person call. Above about 20 people, you can't present, watch chat, scan tiles and take notes at once without the presenting getting worse.

2. Ask the room directly

If the audience won't volunteer a signal, ask for one with polls, votes, structured Q&A or QR-code activities. Slido and Mentimeter are built around this: you pause, run a poll, see the results and respond.

This is a good fit when participation is meant to be visible, such as conference Q&A, a classroom comprehension check, an all-hands pulse or a workshop where the activity is the content. It's awkward when stopping would break the meeting's shape, such as a small sales demo, a founder pitch or a renewal call, where "scan this QR code" reads as friction. A poll answered by typing a letter in the meeting chat is less disruptive in those calls; see how to run a poll in Zoom, Google Meet and Teams.

3. Recap it after

Accept that you can't catch everything live and record the meeting. A bot joins, transcribes and sends a summary with action items to Slack, your CRM or a shared drive. Otter, Fireflies and Read.ai all do this and differ on integrations, languages and automation.

The trade-off is timing. You find the missed question in tomorrow's summary email, not when you could have fixed it. That's fine for sales follow-up, CS handoffs and lecture archives, but not for demos, pitches or live coaching.

4. Have a co-host watch for you

An AI co-host joins your call as a participant and gives you a private view of the room while you talk. Claryoo does this: it shows you who is engaged, who has gone quiet and which question has waited longest, with every question from chat and voice on one board. We built it for live sessions where you can't rebuild the meeting around activities, can't wait for the recap and can't scan everything yourself.

Which approach fits which session

Most teams running mixed sessions end up with two tools: one for the live moment and one for the record. The mistake is expecting one tool to do both jobs.

The tools fall into three groups that don't substitute for each other. Audience-input tools (Slido, Mentimeter, Poll Everywhere, Mural) let the room speak during the meeting. A co-host (Claryoo) helps the host watch the room during the meeting. Recap tools (Otter, Fireflies, Read.ai, Fathom, Grain) capture the meeting afterwards. Within a group, tools mostly differ on integrations, languages and price.

A working pattern

Before the call, pick the tool for the session. Interactive event: Slido or Mentimeter. Live demo with no room for activities: Claryoo. Internal sync where action items matter most: a note-taker.

During the call, decide which tool's output you'll act on and treat the others as background. Three AI tools in one meeting, with no clear owner of your attention, is a common failure.

After the call, read the output. Plenty of teams pay for a note-taker and never open the summaries. If you won't act on the recap, it's a backup transcript, not a reason to pay for a seat.

FAQ

What's the easiest first step?

Pick one session type you run often and one approach. For an internal status meeting, install a note-taker (Otter has a free tier). For a customer demo or pitch, try Claryoo. Don't add three tools at once.

Do I need to tell participants there's an AI in the meeting?

Generally yes. Many jurisdictions require consent when audio is recorded, and some apply the same rules to live transcription. Check the rules that apply to you before using any AI tool on external calls, and tell people anyway; it's good practice. Claryoo doesn't record video, and call audio is deleted automatically after 7 days.

Can I just learn to read chat better and skip the tools?

On small calls, yes. Above 15 to 20 people while you're presenting, there's a limit to what one person can track.

Won't Zoom and Teams build this in?

Zoom AI Companion and Microsoft Teams Premium are moving into the recap category. Audience-input and host-side tools have fewer built-in equivalents so far.

Which approach is best?

None of them on its own. They solve different problems, so pick the one or two that fit the sessions you actually run.

Want a co-host in your next meeting? Start free, no card needed.