Every row in a map traces to a public source: the event's own site, speaker and sponsor pages, session agendas, public posts and announcements, press coverage, and attendance patterns from prior editions of the same event.
What we never use: anything behind a login, purchased broker lists, private databases, or data a client shares about anyone other than their own records. Read-only CRM access is used for the coverage diff and nothing else.
The two lists are never blended. Blending is the most common trust failure in event data, and it's the one thing this format makes impossible.
The map freezes before the event starts. What we predicted is on record and can't be edited after the fact.
We reconcile the frozen map against who actually showed: post-event signals, client badge scans where shared, and on-the-ground confirmation from engagements we run.
The diff lands in the log below, dated: hits and misses both. Maps that grade poorly stay in the log.
Every map lands here as its event closes out, checked against reality. The format below is what each entry will look like.
| Event | Predicted | Verified after | Diff published |
|---|---|---|---|
| INBOUND Example | 341 | 289 | Sep 22 |
How many names the frozen map expected in the room before day one.
How many of those were confirmed present after the event, from post-event signals and shared records.
The difference, published as-is. A grade that can't show misses isn't a grade.
See a full deliverable: the sample coverage report.