Two thirds of OVO's recorded call time is silence, because nobody was in the room.

The silence is not a Zoom fault. It is an attendance problem.

Zoom auto-records every booked meeting the moment it opens. When nobody shows, it faithfully records an empty room. That is where the 325 silent minutes came from.

Zoom's participant report tells us exactly who was in each meeting. Cross-referencing it against the audio produces a result with no exceptions: every meeting that produced a substantive transcript is a meeting where both sides attended, and every meeting that didn't, isn't. Seven for seven, both directions.

In 8 meetings the prospect showed up and nobody from OVO ever joined. One of them sat alone in the room for 86 minutes. Zoom recorded every silent second of it.

One meeting had no humans in it whatsoever — the only attendee was an AI notetaker bot, transcribing an empty room for ten minutes.

The problem concentrates in the coaching calls.

16 of the 28 meetings are booked as coaching calls under Samantha's calendar. Samantha does not attend any of them — Jaci runs every one. 6 of the 8 closer no-shows sit in this bucket.

A recording cannot tell us why a side is absent — a call cancelled late, a prospect who never got the memo, and a closer who forgot all look identical on the tape. That distinction needs a human. What the tape does establish is the rate, and the rate is 75%.

Zoom is deleting the evidence every 30 days.

The account is set to auto_delete_cmr_days = 30, with permanent purge from trash. There is no undelete and no archive.

April and May 2026 are already gone. Both calls titled “ovo x fitia” from 2 June were destroyed before anyone looked. On the day of this audit the oldest surviving recording was 28 days old — roughly 48 hours from deletion.

Everything in this report was captured to local disk before that window closed. It is a snapshot, not a fix: nothing is scheduled, so the next 30 days will be lost the same way unless the capture runs again.

An AI notetaker is sitting in your calls.

A Fireflies.ai notetaker bot joined 16 of the 28 meetings, including a call where it was the only attendee. It is recording and transcribing OVO's sales conversations to a third-party account. Whether that is sanctioned is a question for you, not for the data.

A “Fitia calls” dashboard would show two rows.

Across the entire account for 30 days, Fitia is mentioned in 2 meetings. Not thirty-four. That figure comes from a keyword count across a separate archive of pre-June recordings, and it counts mentions — a prospect saying “I already use Fitia” lands in it.

This isn't a reporting bug. With 7 real conversations in a month, there is very little to filter. The filter is easy; the calls are the constraint.

What to do about it

1. Fix the calendar, not the dashboard

   6 of 8 closer no-shows are on Samantha's coaching calendar, which Samantha doesn't attend. Either the bookings route to whoever is actually running the call, or they stop being booked. This is worth more than any reporting work.

2. Stop the 30-day bleed

   A scheduled job that copies transcripts, audio and attendance out of Zoom before deletion. Until it exists, every call older than a month is unrecoverable — and one month of calls has already been lost during this investigation's lifetime.

3. Then build the dashboard

   Attendance is the column that matters and it needs no AI to compute — Zoom reports it directly. Brand mentions come free from the transcript. Distinguishing pitched from mentioned is the only part that needs a model, and it applies to seven calls a month.

4. Decide about the notetaker

   Confirm the Fireflies bot is sanctioned, and that its transcripts live somewhere OVO controls.