The webinar is working.
The economics are not.

Across the eight sessions the first sign-up link appears between minute 39 and minute 114, averaging around minute 80. By that point only 39% to 70% of the live audience is still watching. On 5 July the link went up at minute 114, with 42% present. On 30 July, the largest audience of the whole period at 195 people, the first link came at minute 95 with 39% present.

Roughly 442 people who showed up live never saw the offer at all. At $11.20 per live attendee that is about $4,950 spent getting people into a room and then not telling them what to do next.

The sign-up link does appear in every session. What changed is how and when it arrives. In June it was posted by hand as a host message, three to five times per session. From July it appears as an automated system entry, and in most sessions only twice.

20 June is the clearest evidence in the whole dataset. The link went up at minute 39 while 70% of the room was still there, and it was repeated five times. That session also produced the strongest retention of any session in the period. Compare 5 July, where the link first appeared at minute 114, with 42% present, and was shown twice.

The in-platform offer button was only switched on in August, and it recorded 11 clicks in total across both sessions. For June and July there is no offer click tracking at all.

The show up rate is 21.9%. For a free, warm, ad driven webinar in a high emotion category, 30% to 40% is a normal target.

One thing worth checking: across all 2,091 attendee records the add to calendar field reads No, without a single exception. A field that uniform usually means it is switched off or not recording, rather than 2,091 people each declining. Either way, a working calendar invitation and a same day reminder are the cheapest show-up levers available, and closing the gap from 21.9% to 30% is worth roughly 100 extra live attendees every month at no additional ad cost.

Sessions run between 81 and 148 minutes. Median watch time sits at 46 to 77 minutes regardless of how long the session is. People give this webinar roughly an hour, and that number barely moves when the session gets longer.

30 July proves it. At 148 minutes it was the longest session and drew the largest audience, and it produced the worst median watch time of the entire period at 46 minutes, with only 23.6% staying past three quarters. The extra length did not add depth. It pushed the offer further out of reach.

20 July has multiple attendee reports of frozen and black screens with audio glitches around 2:09pm to 2:18pm. Only 22 people attended live out of 586 registrations, a 3.8% show up.

22 August has the host asking whether anybody can see the video playing, alongside attendees reporting they can neither see nor hear it, plus a cluster of impossible 400 to 480 minute session durations in the export.

A 25% technical failure rate on the single event the whole funnel depends on is an operational risk rather than a content problem, and it deserves to be tracked as its own measure.

No email addresses. All 2,091 attendance records have a blank email field, so buyers can only be matched to attendees by name. Nobody can currently prove which webinar produced which sale.

August landing page tracking failed. 3,409 link clicks produced 682 recorded page views, against 86% to 90% in every earlier month. Leads still came through so the traffic was fine, but Meta spent a full month optimising against a broken signal.

July dates do not line up. The registration sheet lists July webinars on the 12th and 19th. The platform recorded sessions on the 20th and 30th. Both sources show three July sessions, so the monthly totals remain comparable, but the two records should be reconciled before anyone builds per session reporting on them.