Why one day, one track, and a room we cap at 250
ProgrammeProgramme Team
The format argument took longer than it probably deserved. Three shortish days or one long one? Parallel tracks or a single thread? Open registration or something tighter?
We landed on one day, one track, and roughly 250 reviewed seats. All three are deliberate, and all three cost us something.
Online events have no venue constraint — that is the trap
A physical conference runs three days because people flew in. The travel cost is sunk, so you amortise it across as much programme as you can fit. None of that reasoning applies online. There is no reason to inherit the shape of an event whose shape was dictated by airports.
Once you drop it, the real constraint is attention, and the real failure mode is the multi-day online event that quietly loses most of its audience after the first morning.
One track, so the room stays one room
Parallel tracks are how you fit more speakers into a day. They are also how you turn one conference into four weaker ones running simultaneously, each with a thinner audience and a chat nobody is reading.
With a single track, everyone sees the same talks. The panel can refer back to a session from two hours earlier and the room knows what it means. By the afternoon there is shared context to build on, which is the thing that actually distinguishes a conference from a playlist.
The cost is obvious: we run far fewer sessions, and good proposals get declined that a four-stage event would have accepted. We think that is the right trade for a first edition. If the line-up grows enough to need stages, we will add them — but we will not add them just to inflate the speaker count.
The window is about coverage, not duration
The day runs 08:00 to 15:00 UTC. Nobody is expected to be present for all of it.
The point of the window is not length, it is overlap: it puts some meaningful chunk of the programme inside working hours in Bangalore, Nairobi, Berlin, London and the US East Coast. A three-hour event cannot do that — it picks a timezone and tells everyone else to stay up. Recordings cover the rest.
The cap is the whole point
Around 250 seats, every one reviewed by hand.
This is the decision people push back on hardest, usually with a reasonable question: why turn away someone who wants to attend a free event?
Two answers. The first is that a few hundred reviewed practitioners behave like a room. Questions get answered. The chat is readable. You can tell who you are talking to. At five thousand, the chat is a scrolling wall and the Q&A is a lottery.
The second matters more. Recordings never leave that room. Nothing is published, nothing goes to YouTube, nothing goes to sponsors. That restriction is not branding — it is the specific mechanism that lets a speaker name the vendor, show the configuration that failed, and describe an incident the way it actually happened rather than the way legal would prefer it written up.
You cannot get that talk at an event that livestreams to the open internet. We have all sat through the sanitised version.
What we are giving up
Reach, obviously. A public, uncapped, publicly-recorded event would be seen by far more people, and the good talks would outlive the day.
We are also making it harder to attend, which will exclude people who would have been great to have and who do not hear about us in time. That is a real cost and we do not want to pretend otherwise.
And we are asking sponsors to value a few hundred reviewed people over a large anonymous number — a harder sell, and the right one for who this is aimed at.
What we will probably get wrong
The breaks. Four of them, 15–20 minutes, expo open. In an online event, a break is an invitation to go back to Slack and not return. We will see it in the retention numbers.
And possibly the cap itself. If we turn away two hundred good applications, that is a signal, and we will say so publicly rather than quietly pretending demand matched supply.