960 registered, 237 watched live, 32 booked a call, and one deal has already closed at $19,800 — from a lead that had been sitting cold since June. The single biggest lever available isn't the content. It's when the link appears.
Provisional — 21 Aug 2026. Twelve of the 32 booked calls run through 25 Aug and haven't happened yet. Revenue and show-up figures will move. This page updates in place; re-run scheduled for 1 September.
Two days after the event, with twelve of the thirty-two booked calls still to run. One closed deal is the floor, not the result.
The content held. Average watch time was 54.5 minutes of a 79-minute session, and 155 people — 65% of the room — stayed to the final second. Attendance barely moved between minute 6 and minute 79.
The offer didn't reach everyone. A single link drop at minute 50 meant 54 attendees left without ever being asked. Of the 183 still in the room, 15.3% booked. Of the 54 already gone, 1.9% did.
Sales capacity is leaking through D-tier. Show-up on the resulting calls is 50%. Eight of nine cancellations were D-tier prospects, and no D-tier prospect has completed a call.
Watch time looks like it drives conversion. It doesn't. There was one moment of offer exposure in the whole session, and watch time mattered only because it decided who was still there for it.
How to read it: the attendance band is anchored on four measured points — 185 peak concurrent, 183 present at the link, 155 at the final minute, and the join ramp — so the shape between anchors is interpolated, not observed. The booking line is fully measured. The minute-50 timestamp is approximate: the Zoom participant export carries join and leave times but no chat events, so the drop can't be timed from the data. Retrieving the saved chat transcript would fix it exactly.
They watched a median of 13 minutes and left before any offer existed — 12 of them watched 30+ minutes. They aren't disinterested; they were never asked. Right now they sit undifferentiated inside “Attended Did Not Book” alongside 129 people who saw the link and passed. Those are two different follow-up cases in one list. The names are attached as webinar_2026-08-19_pre-link-leavers.csv.
Worth walking through, because it shows the whole machine working end to end — and because it's the pattern everything upstream should be aimed at producing more of.
Martin Gelman arrives as a cold paid-social lead on the Land A Talk funnel. An SDR deal opens the same day.
A handful of site visits. No call. Eleven weeks of outbound produce nothing.
Right through to the final second. When the offer link goes up he doesn't move on it — he watches to the end.
No SDR, no chasing, no follow-up email. Scores 5 at check-in — top of the scale. The deal record opens automatically 79 seconds after he books.
His Marketing Call A. Completed. The only appointment he has ever had in JIT.
Roughly nineteen hours from joining a webinar to signed.
HubSpot still records the source as paid social, because that's where he came from in June. The webinar didn't generate this lead — it woke one up. Both are worth having; they aren't the same thing. Whether the other 31 bookers are also pre-existing leads hasn't been checked. One query on contact createdate across the cohort would settle it, and it changes how the 13.5% attendee-to-booking rate should be read.
Fourteen of the thirty-two slots have passed. Seven were held. Every cancellation but one came from a tier that has never completed a call.
| Tier | Completed | No-show | Cancelled | Read |
|---|
Read it as: D-tier books, holds a slot, and drops — 8 cancellations, 0 completions. D-tier is already excluded from auto-matching by the min_score_pref floor set on 13 Aug, but they are clearly still reaching the booking page and consuming calendar. Whether that is intended is a routing question, not a marketing one.
Blake Hernandez and Michael Williams both have no_show_alerted_at set and both took the call — they simply arrived after the alert fired. If those alerts drive rep behaviour or reporting, the threshold is too tight.
Four things, in the order they pay off.
Move the offer link earlier, or post it twice. A single drop at minute 50 left 23% of attendees never seeing an offer. It's a cheap change and a precisely measurable one — this event is the baseline. One deal here was worth $19,800; converting even a handful of the 54 pays for it many times over.
Taylor / MarketingWork the 54 pre-link leavers as their own follow-up segment, separate from the people who saw the link and passed. Different message, different case.
MarketingDecide whether D-tier prospects should reach the booking page at all. They are consuming slots the sales team is working from and completing none of them.
Jordan · decisionRe-run the revenue and show-up view on 1 September, once every booked call has had its slot and normal close lag has played out.
Darren · 1 SeptFour data problems found while building this brief. The first one will produce a wrong figure in any report built today.
Sampled membership includes a contact whose last conversion is a post-call CSAT survey, one whose only qualifying event is an application form, one with an application and no meeting at all, and one whose last conversion belongs to a different funnel entirely. The criteria appear to capture attendance plus any historical booking. Do not report from this list until it is rebuilt.
One group, all conditions true:
1 Segment membership is member of
10. MKTG: Webinar: House: Live: CC: Taylor Conroy: 08-19-2026: Attended
2 Form submission — has submitted any of:
Meetings Link: leadr/strategy-session-a-content
leadr/strategy-session-b-content
leadr/strategy-session-c-content
leadr/strategy-session-a
leadr/triage
after 19/08/2026 7:00pm ET
3 Email does not contain shahmir.aashir+ (strips test records)
Expected size ~32. If it returns near 76 the date bound hasn't applied; if it returns near 22, a meeting-property condition has been used instead of a form condition. Use form submissions, not Last meeting booked date — the obvious filter returns 22 and misses roughly a third of genuine bookers, because their meeting property is blank despite live JIT appointments. Rebuild the complement 24276 the same day: member of Attended AND NOT member of the corrected 24275, expected ~205 rather than the 161 it shows now.
engagements_last_meeting_booked is unreliable. Roughly a third of confirmed bookers have live JIT appointments and no meeting engagement on the HubSpot contact. This is a writeback gap and looks like the same class of problem as the outbox split-brain issue. It should not be used as a booking indicator until it's understood — and it wants its own investigation, not a paragraph in this brief.The contact-level webinar properties (last_registered_webinar_status, webinar_status_last) are blank for 918 of 960 registrants and are legacy artefacts. Any report built on those properties will be wrong. Build on lists. And because both lists hardcode the event date, every webinar needs a fresh pair built by hand — which is how criteria drift in. Worth templating rather than rebuilding.
| Question | Owner |
|---|
Every figure above traces to one of four sources. If you rebuild from these, you get these numbers.
| Source | Used for |
|---|
Zoom reports 281 participants. That counts join events: 240 unique emails produced 281 rows because 28 people reconnected. Removing three internal accounts gives 237 unique external attendees — which matches HubSpot list 24100 exactly. Two independent sources agreeing is strong verification. The 41 reconnections across 28 people is worth a passing look; it could be normal churn or it could indicate stream instability.
strategy-session-a/b/c-content) does not correspond to the scored tier in JIT — Sunni Knowlton booked the A-tier page and scores C. Worth confirming that's by design.The Zoom for Claude connector queries the zoom_meeting datasource and is scoped to meetings the authenticated user hosted or attended. Zoom Webinars are a separate object type — it returned nothing for this event and will return nothing for any future one. The fix is GET /report/webinars/{webinarId}/participants, which needs report:read:admin scope on a server-to-server OAuth app. Wiring that into a job that writes attendance and watch duration back to the contact record removes the manual CSV export permanently.