A meeting budget works when it caps the team’s recurring load, counts in person-hours, and has one named person who says no when it is spent. It fails when it is written as a per-person quota, because then the person with the least power in the room is the one who runs out of budget first and has to decline a director to stay inside it.
That is the whole design, and it is nearly the whole post. The rest is how to pick the number, what the document says, and the four things that go wrong in the first month.
What a meeting budget actually caps
The unit is person-hours, not meetings and not calendar hours. A 60-minute meeting with eight people costs eight person-hours. The same meeting with three people costs three. This matters because almost all of the available savings on most teams are in attendance rather than in the number of meetings, and a budget counted in calendar slots cannot see that.
So the budget is one number: the person-hours per week this team may spend on recurring internal meetings. Four words in that sentence are doing work.
- Per week, because that is the cycle a recurring meeting repeats on.
- This team, because a team is the unit that can actually renegotiate its own meetings. An individual cannot.
- Recurring, for the reason in the next section.
- Internal, because a cap that counts customer calls turns into a cap on revenue and gets repealed in a fortnight.
Setting the number
Start from what you currently spend, not from a figure you read somewhere. There is no benchmark worth importing here; the only number with any authority is your own.
- Run the audit across the team for two weeks. You need every recurring internal meeting, its duration, and its actual attendance — not its invite list.
- Compute current spend: for each recurring meeting, duration in hours × attendees, summed, divided by two if you audited a fortnight.
- Compute the team’s capacity: headcount × contracted hours.
- Current spend ÷ capacity is the share of the team’s week that is already committed before anyone does any work.
- Set the budget as a cut from that, in person-hours, and write both figures down.
A worked example with invented inputs, so the arithmetic is visible. Eight people, 40-hour weeks, so 320 person-hours of capacity. Suppose the audit returns a recurring internal load of 88 person-hours a week — 27.5% of the team’s time. A 25% cut gives a budget of 66 person-hours, about 20.6% of capacity.
Two rules about that cut. Make it a cut from the measured figure rather than a round target, because a round target invites an argument about whether the target is right and a cut only invites an argument about the size. And make it a single-digit-percentage cut the first time if the team is sceptical: a budget that is met in month one survives, and one that is missed in month one is treated as aspirational from then on.
Budget the recurring load, not the ad-hoc
This is the distinction that makes the instrument work, and most published versions miss it.
Recurring meetings are decided once and paid for every week, by people who were often not in the room when the decision was made. Nobody re-approves the Tuesday sync. That makes them a standing commitment against the team’s capacity, which is exactly what a budget is for.
Ad-hoc meetings are decided by somebody with a reason, that week, in response to something. Capping them means the team cannot respond to things. You get meetings that are declined for budget reasons while the problem they were called about goes unhandled, and the team learns that the budget is an obstacle rather than a tool.
So: the budget covers recurring internal meetings. Ad-hoc meetings are governed by the ordinary decline, by agendas, and by whoever called the meeting having to justify it — not by a number.
The exception is a team where ad-hoc load is genuinely the problem. There, cap the ad-hoc separately and loosely — a stated upper bound for the week, reviewed rather than enforced — and keep the two numbers apart.
Who enforces it
There are exactly three answers, and one of them is what happens by default.
The manager. They see every recurring invite for the team and refuse the ones that take the budget over. This works and it costs the manager something real: they are now the person who said no, every time, to peers. Most managers will not do this for long without being told to by someone above them.
The team, in a standing review. The budget is reconciled in public, monthly or quarterly, and the team decides together what to cut to make room for something new. Slower, more durable, and it distributes the cost of saying no — which is the actual reason it outlasts the manager version. This is the one to pick.
Nobody. The budget is announced, nobody is named, and it becomes a number in a document that gets quoted when convenient. This is the default and it is what happens if you do not choose one of the other two.
Whichever you pick, the budget needs a name against it in writing. The properties that keep a policy alive past month three apply to a budget without modification: written where a new person finds it, an exception process that existed before the first exception, and someone senior visibly inconvenienced by it at least once.
The written budget
Short. Longer than this and it does not get read.
Recurring internal meeting budget: 66 person-hours per week.
How it is counted. Duration × expected attendance, for every recurring internal meeting on this team’s calendar. Customer and external meetings are not counted. Ad-hoc meetings are not counted.
Current spend. 58 of 66 as of . The ledger is .
Adding a recurring meeting. Anything new has to fit inside the remaining budget. If it does not, the proposer names what comes off the calendar to make room. “We will find the time” is not an answer.
Who reconciles it. , in the monthly team review. Disputes go to .
What is exempt, decided in advance. Incidents and on-call handover. The weekly all-hands. One-to-ones — these are not budgeted and not reviewable here.
Review date. <date, one quarter out>.
Exempting one-to-ones is deliberate and worth being explicit about. They are the meeting most likely to be sacrificed to a cap and the one that costs the most to lose, and a budget that quietly pressures people into cancelling them has made the team worse while reporting success.
The ledger is the part that does the work: a single table of every recurring meeting, its duration, its attendance and its person-hour cost, sorted by cost. Teams who build it usually find that two or three meetings account for most of the spend, which changes the conversation from a general complaint about meeting load into a specific decision about which standing meetings survive the quarter.
The quarterly reconciliation
Thirty minutes, and it is the same four questions each time.
- What is current spend against the budget? If spend has drifted up without anything being approved, something was reclassified — find it.
- Which meeting has the highest person-hour cost, and did anybody defend it this quarter?
- What was added since last time, and what came off to make room?
- Is the budget still the right number? Raise it if the team has grown or the work has genuinely changed. Lower it if the last cut held comfortably.
Write the answers down in the same place as the budget. A reconciliation that happens and is not recorded reads, three months later, as a reconciliation that did not happen.
What breaks first
- Reclassification. A recurring meeting becomes “ad-hoc, but weekly”, or internal becomes “customer-adjacent”. This is the first failure and it arrives within a month. The countermeasure is that the ledger lists meetings by calendar entry, not by category — if it is on the calendar every week, it is recurring, whatever it is called.
- Attendance inflation. Duration stays fixed, attendance creeps, and the cost goes up without any meeting being added. Count attendance from the audit each quarter rather than from the invite list.
- The 25-minute meeting. Teams shorten meetings to fit the cap, which is a real gain, and then schedule more of them, which is not. Watch the person-hour total rather than the average duration.
- It moves to DMs. The meeting disappears and the same conversation happens in a thread with six people in it, synchronously, for forty minutes. The budget did not reduce the work; it moved it somewhere nobody counts. This is the failure mode with no clean countermeasure, and the honest thing is to look for it at the reconciliation and say so if it has happened.
If you cannot get a budget for the team
Set one for yourself and do not announce it as a policy.
Compute your own recurring load in hours a week. Pick a number you will hold — say, no more than a quarter of your contracted hours in recurring internal meetings. Then, when a new standing invite arrives and you are at the cap, you have a reason that is not about preference: something has to come off, and you can say which.
“I’m at about ten hours a week of standing meetings and that’s as much as I can carry and still ship. Happy to join this one if I can drop the Thursday review — or I can read the notes on this and come to the ones where I’m actually deciding something.”
That is a budget doing its only real job, which is to turn a vague reluctance into a specific trade. The cost of declining does not disappear because you have a number — but the number is what makes the decline legible to someone who is not inside your week.