Kickoff

Most projects start quietly. Someone writes a document, two engineers get assigned, a designer opens a file, and that's the start. Nobody has been in a room together, so everyone carries a slightly different picture of the work, and the differences stay invisible for a while. Each picture is internally consistent, which is exactly why nobody thinks to check it against anyone else's.

They usually surface about three weeks in, at the first review. That's what happened when Pagedeck built permission management for enterprise customers. Theo was building a screen where the customer's own admin edits permissions directly. Sofia was selling a flow where the customer emails support and support makes the change. Both had read the same document and interpreted it reasonably. Neither was wrong, and three weeks were gone.

A second problem shows up in the same place: who decides. When nobody owns a decision, every question that comes up starts with a hunt for the right person to ask, and the work stops during the hunt. This is how a document collects twenty comments and sits untouched for two weeks.

A kickoff exists to close both of those before the work starts. Agreements that get postponed don't disappear; they come back with interest. A conversation that costs thirty minutes at the start costs several days three weeks later, because by then it also includes undoing what was built.

The agenda walks through six questions in order, each one narrowing the answers available to the next.

1. Why this, why now — the problem being solved and the reason it's worth solving this quarter. Everything after this is judged against it, so a vague answer here leaves every later decision without a test, and arguments later in the project get settled by whoever is most insistent.
2. What's in and what's out — scope. The out list matters more than the in list: "permission screens this round, audit logs next" gets said out loud and written down. That line is the defense against scope creep, the slow expansion of work without anyone agreeing to it. If nothing is recorded as out of scope, nobody notices the expansion.
3. What counts as done — the finish condition, written as a number or a state. Something like "permission-change requests drop from 30 a week to under 5."
4. Who does what, and who decides — these are two different lists. Many teams reach for RACI here, a grid that splits involvement into responsible, accountable, consulted, and informed, mostly to force exactly one decision-maker per item.
5. What appears when — not one final date but checkpoints along the way. A milestone is such a checkpoint, and the kickoff records what has to be visible at each one, since a date with nothing attached to it can be declared met by anyone.
6. What's still unknown — open questions and risks, each with a name and a date attached for who brings back an answer. A kickoff that produces an empty list here almost always means the conversation stopped early.

Afterward the six answers go into a document that also reaches the people who weren't there. The meeting evaporates; the document is the thing that lasts. It's written well if someone joining a month later can catch up from it alone.

Two groups get invited: everyone who will actually do the work, and whichever stakeholders — the people affected by the outcome or able to change it — genuinely need to be in the decisions. Sending one engineer as a representative and briefing the rest later guarantees that whatever context drops in the handoff becomes someone's different interpretation.

Eight people is about the ceiling. At fifteen, more than half say nothing, and someone who said nothing didn't agree to anything — they listened. People who are interested but not needed for a decision get the document instead of the invitation.

The timing sits between two conditions: what to build has been written down once, and nobody has started building it yet. Too early and there's nothing concrete to settle, so the hour turns into speculation. Too late and the meeting happens on top of work that already exists, which makes changing direction expensive. The moment a draft product requirements document (PRD) starts circulating is usually about right.

That document goes out before the meeting and people read it beforehand. When they don't, the first half hour is spent reading it aloud, and the disagreements that actually needed the room get pushed to next time. Some teams put a line at the bottom of the document where readers add their name, purely so the organizer knows before the meeting whether it's worth holding.

"I'll book the kickoff for Tuesday. Please read the PRD before you come."

The document is settled; the meeting is for the disagreements only.

"That was explicitly out of scope at the kickoff."

Pulling a new request back to what the group already agreed on.

"We started without a kickoff and now nobody knows who decides."

Roles were never assigned, so decisions have stalled.

A Pagedeck kickoff runs under an hour. Mia circulates the six agenda lines in advance, and in the room the group stops only on the items where opinions differ. On the permissions project, item four took twenty minutes: whether support should be able to make permission changes on a customer's behalf. Grace settled it in the room — phase one is admin-only, direct edits.

What can't be settled gets recorded as unsettled. The open item that day was how long audit logs are retained, and Sofia took two days to check what the enterprise contracts promise. The output of a kickoff isn't that everything is decided; it's that everything undecided has a name and a date on it.

The most common mistake is treating a kickoff as a ceremonial start. Plenty of them really do consist of one presentation and some applause, so the impression is earned. But the deliverable here is decisions, not enthusiasm. If the out-of-scope list, the decision owners, and the open questions with deadlines aren't in a document afterward, the meeting produced nothing. Those projects hold the same meeting three weeks later, except by then it costs whatever has to be unbuilt.

The opposite mistake is expecting a kickoff to settle everything. Trying to close every question in one sitting produces three hours of meeting and a stack of decisions made without evidence. Questions without answers can simply be written down as open, with an owner and a date. This is the same instinct behind meeting hygiene: in a meeting whose purpose is deciding, decide the things that are actually decidable today.