Quarterly Planning Sessions That Do Not Drift

Quarterly planning drifts for three reasons, and none of them is a lack of enthusiasm in the room. The session starts without a decision that genuinely has to be made, the candidate list never gets properly cut, and people leave holding a document rather than a commitment. Fix those three and two days of argument turns into one day that produces an actual plan.

Drift is not the same as overrunning. A planning session can finish bang on time, cover every agenda item and still have drifted: you have a deck, a spreadsheet and forty initiatives, and by week five nobody can tell you which six things the quarter is about. That is a design failure rather than a discipline failure, which is good news, because agendas are easier to change than people.

What follows is the pre-work that has to land before anyone sits down, a block structure that survives contact with a real argument, the cutting mechanics that take a long candidate list down to a small set of owned commitments, how to keep capacity honest when the room is optimistic, and what to do in week six when the plan has slipped anyway.

The three ways a planning session drifts

The first is context drift. The session opens with a business review - last quarter's numbers, a market update, three functional reports - and by the time that finishes it is half eleven, everybody has been passive for two hours, and the actual planning is squeezed into a rushed afternoon. Context is necessary. Presenting it live, to people who could read it faster, is not.

The second is list drift. The room generates forty candidate initiatives, discusses thirty of them and commits to twenty-two, because cutting felt political and nobody wanted to tell a colleague their proposal was out. Twenty-two commitments across a team of thirty is not a plan, it is a lottery, and the quarter will decide for you which of them actually happen.

The third is ownership drift. The plan is agreed in principle, the document says 'Marketing and Product to align on launch messaging', and no human being has that in their calendar. Work with two owners has none. This is the cheapest of the three failures to fix and the most commonly skipped, because by the time you reach ownership the session is already running late.

Decide what the session must produce before you book it

Write the output down as an artefact, not a verb. 'Align on Q1 priorities' is not an output; 'a ranked list of six committed outcomes, each with one named owner, a done-by date and an agreed measure' is. If you cannot describe the artefact in one sentence, the room cannot tell whether the session is working, and drift becomes invisible until the quarter is half gone.

Then decide what is already fixed, and say so in writing. Most planning sessions pretend everything is open when in fact the budget is set, two initiatives are contractually committed and headcount is frozen until March. Pretending otherwise wastes half a day and, worse, teaches people that the session is theatre. Put the constraints in the pre-read: these three things are not up for discussion, here is why, everything else is genuinely open.

Finally, decide who decides. A room of thirty people cannot make a prioritisation call by consensus, and pretending it can produces a plan nobody privately supports and everybody publicly agreed to. Name the decider before the session: the room generates, ranks and argues, and one person or a small group commits at the end. That is not less democratic, it is more honest, and it makes dissent cheap to voice because voicing it is no longer the same as blocking.

The two weeks before the room

Everything on this runway exists to make the session about argument rather than information transfer. If people are reading in the room, you are paying thirty salaries to do something each of them could have done at a desk, and you are doing it with the most expensive attention of the quarter.

The written positions step is the one worth protecting when the calendar gets tight. Asking each function to submit, in writing, the two things they most want the quarter to contain and the one thing they are most worried about gives you a map of the real disagreements before anyone sits down. You can then build the agenda around those disagreements rather than discovering them at three o'clock on the second afternoon.

A block structure that survives contact

These blocks fit one long day or two half days. Two half days on consecutive mornings beats a single ten-hour day for anything involving numbers: people are sharper, the overnight gap lets uncomfortable proposals settle, and the second morning routinely produces a cut the room refused to make on the first afternoon.

The block most often sacrificed for time is capacity reconciliation, and it is the one that determines whether the plan is fiction. If you are running short, cut the context block, cut the risks block, shorten the pitch round - but never arrive at the close without having matched commitments to actual weeks of actual people's time.

Cutting forty candidates down to six

Steps three and four are where a tool genuinely earns its place. Hidden-then-revealed multi-vote ranking is fiddly with sticky notes, and impossible to run fairly when a third of the room is dialling in. Session Flo runs the private vote and the simultaneous reveal from a single join code, and shows the spread rather than just the average - which matters, because the spread is the part you are going to spend your discussion time on.

Keeping capacity honest

Every quarterly plan rests on an implicit capacity claim, and the claim is almost always wrong in the same direction. Rooms plan as though each person has thirteen usable weeks, when the honest figure after support rotas, interviews, holidays, onboarding and the two incidents nobody can predict is closer to nine.

Make the claim explicit and put it on the wall. Take each team's headcount, subtract known leave, subtract a fixed allowance for run-the-business work, and show the remaining weeks next to the weeks the shortlist actually requires. When the second number is larger than the first, you do not have a stretch plan; you have a plan that will be cut by circumstance rather than by you.

Then make the cut in the room instead of in week seven. The uncomfortable ninety seconds where somebody says 'so which of these six is not happening' is the highest-value ninety seconds in the whole session. If it does not happen there, it happens later, quietly, without the people whose work depends on the answer.

Running the session across time zones

Distributed planning fails differently from co-located planning. The problem is not that people cannot hear each other; it is that the two-hour overlap window gets spent on the presentation blocks, and the argument gets pushed into asynchronous comment threads where it dies politely.

Invert it. Everything that is information transfer - context, numbers, candidate pitches - goes asynchronous: written, or recorded as short clips, consumed before the window opens. The live overlap is spent exclusively on the two things that genuinely need synchronous humans in the same call, which are the contested items and the capacity reconciliation.

Run the voting asynchronously too, with a deadline, so the ranking already exists when the call starts and the live time goes straight to the splits. Ninety minutes of overlap across three regions is enough for a real decision, but only if none of it is spent on things a document could have done better.

The close: owners, dates and the first check-in

Week six: what to do when it has drifted anyway

Some drift is information rather than failure. If a commitment has stalled because the world changed - a customer left, a dependency slipped, a regulation landed - the right move is to kill it formally and say so, not to leave it sitting at amber until March while its owner quietly works on something else.

Run a fifteen-minute mid-quarter review with the same shape as the close. Each owner says green, amber or dead in one sentence, and anything amber comes with a specific ask. Do it in week six, when there is still time to move people onto something that will land, rather than week ten when there is not.

Keep a running record of what was committed against what actually happened, and open the next planning session with it. A team that sees its own hit rate before it starts generating candidates plans differently - usually smaller, usually with more slack, and usually better.