The Frisconian

SF Event Index

What fills a room in San Francisco, from thousands of public Luma events. Filter the charts, or describe your event and see how events like it actually did.

When it happens, and how many come

Color shows median turnout, on the scale under the grid. The number in each cell is how many events ran then.

Formats by turnout

Median guests at finished events. Tap a format to filter by it.

When people register

Share of the final guest count already in, by days before the event. The band is the middle half of events.

Price and sell-outs

Share of finished events that sold out, by ticket price.

What the page says

Median guests for event pages that do this, against pages that don't.

Page does itPage doesn't

Co-hosts

Median guests by how many hosts the page lists.

Description length

Median guests by how long the event description is, in characters.

How far ahead it was listed

Median guests by days between first listing and the event.

Where

Events by neighborhood. Tap one to filter by it.

Events per week

Every listed event in the current filters, by the week it starts. Weeks ahead are still filling up as hosts post.

How this data was collected

Every night a small script reads the public event pages listed on Luma for San Francisco, Oakland and Berkeley, one page per second, and records what any visitor can see: the date and time, the neighborhood, the ticket price, whether approval is needed, and the guest count when the host shows it. It reads each upcoming event again every night, which is how the sign-up curves exist.

No login, no private data, no guest names. This site never shows a single event: every number comes from at least five events, and the raw records stay off the internet.

How to read it

  • Turnout is the guest count the morning of the event, for finished events whose hosts show their guest list. That is about half of all events.
  • Registrations are not attendance. Free events usually lose a large share of sign-ups at the door; paid events lose far fewer.
  • These are patterns across many events, not causes. A habit that goes with bigger rooms may simply be what experienced hosts do.
  • Formats are sorted by keywords in the title and description, so a few events will land in the wrong bucket.
  • Nightly reads began in late August 2026, so the history is short and grows every night.