All-hands

The number of people who make a plan and the number who work to it are never close. At a company of forty-five, six or seven people sit in the quarterly planning meeting. The other thirty-eight have to hear what got decided from somewhere.

When that somewhere isn't defined, the information arrives in fragments. One person hears it over lunch, another catches a single line in a chat thread, a third hears only what their manager chose to pass along. Everyone ends up holding a different version of the same decision, and the gap surfaces in hallway questions like "I heard we're killing that feature?" The person asking got half a rumor, and the person answering has no idea where the rumor came from.

Worse than information leaking sideways is information that never arrives. Someone who spends three months without knowing which company goal their work attaches to has no basis for deciding priorities. Two urgent requests land and they either escalate every time or pick one at random. Neither habit shows up as a complaint; it shows up as work that keeps landing slightly off what the company needed.

An all-hands makes that path a single defined one. What has been decided gets delivered at a set time, to everyone, in the same words. The point is less the volume of information than the fact that everybody received the same content at the same moment.

Most companies land on a similar order. Last period's numbers first, then what happens next, then questions. An hour is the usual ceiling.

1. Last period's numbers — revenue, users, and performance against target, shown as they are. If the target was $1,000,000 and the result was $850,000, both numbers go on the same slide. Numbers come before interpretation, because a room that hears the framing first spends the rest of the hour checking it against the figures.
2. What worked and what didn't — go through what the company said it would do and mark what happened. Anything that missed gets a reason attached. Skip this part and the rest of the agenda turns into marketing.
3. Next period's work — what was chosen and why. The output of company-level prioritization gets published here. If goals are managed through a framework like OKR, that list is what gets shown.
4. Team news and new joiners — recent changes in each org, plus everyone who started this month. Past forty people there are faces nobody recognizes, and this is often the only introduction they get.
5. Questions — answers to what was submitted in advance and what comes up live. Many companies pair this with an anonymous channel, because the questions most worth asking are exactly the ones people won't raise a hand for.

Materials go out the day before, and a recording and a written summary stay behind afterward. People who missed it and people who join later need the same content for the meeting to work as a record.

Everyone is invited, with no split between full-time staff, contractors, and interns. The moment you split the list, the meeting stops being about company news and becomes a signal about which group you belong to. Typically the chief executive opens and closes while each org lead speaks to their own section — if one person reads all of it, nobody in the room can answer a detailed question.

Timing attaches to the planning cycle. Right after the quarterly plan is locked is the default slot, since holding one before anything is decided leaves nothing to say but general updates. Many companies add a shorter monthly session that covers only the numbers, so decisions land quarterly and progress lands monthly.

Sessions also get called outside the schedule. Layoffs, a funding round, or a change in product direction are the kind of news people will otherwise hear as a rumor first, and the later it comes through the official channel, the more it reads as something the company tried to hold back.

Frequency follows headcount. At fifteen people there's little reason to hold one, because everyone already hears it in the same room. Past forty, things start happening that other teams don't know about. Past three hundred, gathering everyone at once becomes the hard part, and recordings and regional replays get attached. With people in several time zones, some companies drop the live gathering entirely and treat the recording and the written summary as the real artifact, keeping the live slot only for the question session.

"You're covering that at the all-hands, right? If it doesn't come from there, people will hear it as a rumor first."

A request to use the official channel before the informal one gets there.

"We'll take questions anonymously and answer the three with the most votes."

A commitment to surface the questions people find hard to ask out loud.

"Two of last quarter's three goals didn't land. Let me start with why."

A signal that this is a status update rather than a performance showcase.

Pagedeck books an hour on the Thursday after the quarterly plan is locked. Grace puts last quarter's numbers up first and speaks to the reasons behind anything that fell short. Mia then covers what product is doing first this quarter and what got pushed back. Naming what got pushed back matters because sales or customer success may have been waiting on exactly that.

The last fifteen minutes are questions. Submissions open three days ahead through an anonymous form and get answered in vote order. When something uncomfortable comes up — compensation, headcount plans — the answer is the current boundary of what can be said, not a deflection.

The most common mistake is running it as a showcase. With everyone in the room, the pull toward good news is natural. But the people listening already know what isn't working on their own teams. A list of wins tells them the company either doesn't know or isn't saying, and that conclusion sets after two or three sessions. From then on, genuinely important news from the same stage gets filtered. Transparency is exactly what is under test here.

Treating more sessions as better is the other one. Forty-five people for an hour costs the company forty-five hours. Run it weekly and there isn't enough new material, so last week's content comes back lightly reworded and people leave it playing while they work on something else. Match the interval to the rate at which decisions actually accumulate.