The Operating Cadence

There are two ways to think about the set of recurring meetings that governs an organisation. The first, which is nearly universal, is that they are overhead — a necessary tax on real work, to be minimised at the margin, occasionally purged in a burst of enthusiasm and then quietly regrown. The second, which is correct, is that they are the organisation's coordination infrastructure: the mechanism by which information moves between parts of the system and by which decisions get made. Andy Grove put it plainly decades ago — the meeting is the medium of managerial work, in the way that code is the medium of engineering work. You would not describe your deployment pipeline as overhead.

The difference between these two framings shows up in how an organisation responds when the calendar gets too full. Treat meetings as overhead and you cut indiscriminately, lose coordination, watch the problems reappear, and reinstate more meetings than you removed. Treat them as infrastructure and you ask a different question: what is each forum for, what does it cost in latency, and is there a cheaper mechanism that achieves the same coordination.

That second question has a surprisingly precise answer, because a recurring meeting has a calculable cost that almost nobody calculates. It is not the cost of the attendees' hour. It is the wait time it imposes on everything that must pass through it.

The arithmetic of a weekly forum

Suppose a class of decision can only be made at a meeting that happens every Thursday. Questions in that class arrive more or less uniformly through the week. What is the average wait?

Half the cycle, before anything else happens: three and a half days. A question raised on Friday morning waits six days. One raised Wednesday evening waits one. Average, three and a half.

Then add the realities. Agendas usually close a day or two before, so a question raised on Tuesday for a Thursday meeting may miss the cut and wait until the following week. Add the probability of deferral — a forum that gets through four of its six items pushes two into the next cycle, which is another seven days for those. Add the time between the meeting and the outcome being written down and communicated to the people who were waiting, which in most organisations is a further two to four days and sometimes never happens at all.

A single weekly gate is therefore rarely under four working days of average latency and is frequently closer to eight. That is the number to hold in your head when someone proposes taking a decision to the weekly forum. They are not proposing to spend twenty minutes on it. They are proposing to spend a week and a half of elapsed calendar, during which whatever depends on the answer does not move.

Now note what happens when gates chain. A design that must pass the architecture forum and then the investment committee, both weekly with their own agenda deadlines, does not wait four days plus four days. It waits considerably more, because missing one cycle pushes the whole downstream sequence by a week. Two weekly gates in sequence commonly produce three to four weeks of pure wait for a decision that took forty minutes of deliberation. Nobody designed that; it emerged from two recurring calendar entries.

What genuinely needs to be synchronous

The test is not importance. Important things are frequently better handled in writing, because writing forces precision and leaves a record. The test is whether the interaction requires the participants to build on each other's responses in real time.

Genuinely synchronous. Decisions where the right answer emerges from disagreement between people who need to hear each other's reasoning. Conversations where reading the room changes what you say — difficult performance discussions, negotiating a commitment between peers, delivering news that will land badly. Anything requiring rapid iteration across several people, such as working through a design under uncertainty. Relationship-building, which is real work even though it is unmeasurable.

Not synchronous, however traditional. Status transmission. If the state of delivery is visible in a system, a meeting to describe it is a meeting to read a dashboard aloud. Information broadcast, which a written update does better and cheaper and which people can read at the speed they read rather than the speed you talk. Approval of things that were never going to be refused. Review of a document that nobody has read, which is the single most expensive habit in most executive calendars, because it converts twelve people's hour into a reading room.

The last one has a remedy. If a document must be reviewed, either require it to be read beforehand and hold people to it, or open the meeting with twenty minutes of silent reading. The silent-reading pattern is startlingly effective, because it removes the presenter's incentive to spend forty minutes walking through slides that exist mainly to protect against the audience not having read anything.

A useful reframe: the purpose of a forum is to produce decisions, and one that reliably produces none is not a governance mechanism but a recurring appointment. That includes forums where the decisions are made beforehand in side conversations and the meeting ratifies them — not necessarily wasteful, but it should be given twenty minutes rather than ninety.

The cadences that actually earn their place

A functioning operating cadence usually has four layers, each with a different clock speed, and the discipline is keeping the layers from bleeding into one another.

Layer	Rhythm	Purpose	The failure mode
Daily	Short, team-local	Unblock today's work	Expands into status reporting upward
Weekly	Leadership team	Decide what needs deciding, surface risk	Becomes a status round-robin
Monthly or six-weekly	Cross-function	Adjust allocation, remove impediments	Becomes a second, longer weekly
Quarterly	Whole organisation	Re-set direction, review outcomes	Becomes a performance

The most common structural fault is that all four layers do the same thing at different intervals — each one a review of progress against a plan, with slightly different slides. When that happens, the organisation has one mechanism repeated four times rather than four mechanisms, and the quarterly is simply the annual version of the weekly with better catering.

Each layer should do something the others cannot.

The weekly leadership meeting is for decisions, not updates. Structure it as a list of open decisions with owners and ages, taken in order of how long each has waited and what it blocks. If there are no decisions, cancel it. Replacing the round-robin with a decision queue is the most useful change most leadership teams can make, because a round-robin guarantees the loudest function gets the most airtime and the most urgent decision gets whatever is left.

The monthly is for reallocation and structural work. Where you move money, change team boundaries, kill something, and address impediments no single team can remove. It has a longer clock because structural changes need time to show an effect, and changing them weekly produces thrash.

The quarterly is for direction and honesty. More on this below.

The daily belongs to the teams. The most reliable way to destroy a team's daily coordination is for a senior person to attend it, at which point it converts from a planning conversation among peers into a status report delivered upward.

Quarterly reviews that surface truth

The quarterly business review is the most expensive recurring event in most organisations and frequently the least informative, for a structural reason: the incentives all point towards performance rather than disclosure.

Consider the preparation from the presenting team's point of view. Weeks of work producing a pack. A rehearsal. Numbers chosen because they are defensible. A narrative that explains shortfalls by external factors and successes by the team's efforts. This is not dishonesty; it is the rational response to a forum where being questioned sharply in front of peers is the observable downside and where nobody has ever been rewarded for volunteering that their strategy is not working.

You will not fix this with an exhortation to be open. You fix it by changing what the forum rewards.

Ask for the bets that failed and what was learned. Make a slide mandatory. A team that reports no failed bets in a quarter either took no risks or is not telling you, and both are worth knowing.

Ban the pack, or cap it hard. A six-page written narrative, circulated beforehand and read in silence at the start, produces more truth than forty slides, because prose is harder to hide in than bullet points. Slides permit assertions without connective reasoning; sentences do not.

Review outcomes, not activity. The question is what changed for customers or for the business, not what was shipped. This is the point of outcomes over output and it lives or dies in this forum specifically, because whatever the quarterly review asks for is what the organisation will spend the next quarter producing.

Respond well to the first piece of bad news. How you react to the first team that tells you something uncomfortable determines what every other team does for the rest of the year. This is the psychological safety mechanism in its most concentrated form, and it is decided in about four seconds of your visible reaction.

Separate the review from the reallocation. If a review's outcome is a budget decision made in the room, every presentation becomes a funding bid. Hold the review, then decide the allocation a week later with the benefit of comparison across teams.

Auditing the calendar without losing coordination

The naive purge — cancel everything, reinstate what is missed — is a popular intervention that mostly fails, because what breaks is not visible for six weeks and by then the connection is lost. Do it more carefully.

Export every recurring meeting in your leadership group's calendars. For each one, record four things: what decision it produces, what would break if it stopped, how many people attend, and how long a decision waits because of its frequency. Most organisations discover three categories.

Forums that produce decisions. Keep them. Consider running them more often with fewer people, which reduces latency without increasing total cost.

Forums that transmit information. Convert to writing. The obligation is on you to make the written version good; a bad written update genuinely is worse than a meeting, and people who defend the meeting are usually defending against the alternative they have experienced rather than the one you describe.

Forums that exist because they have always existed. The majority by count. Pause them for six weeks rather than cancelling, which turns an irreversible decision into a reversible experiment and removes most of the resistance. Note what actually breaks.

Two refinements matter more than the audit itself.

Attack the frequency, not just the existence. A forum that meets fortnightly imposes roughly twice the average wait of one that meets weekly. If a forum genuinely produces decisions, meeting more often and more briefly is almost always better for flow, and it costs no additional attendee hours in total.

Attack the attendee list ruthlessly. A recurring ninety-minute meeting with fourteen people costs twenty-one hours a week. More importantly, its size determines its behaviour: past about eight people a meeting stops being a working session and becomes an audience, at which point the real decision moves to a smaller conversation elsewhere and the forum becomes ratification. If you want the forum to do work, keep it small enough that silence is conspicuous.

And protect unstructured time deliberately. A leadership team booked at ninety-five percent has no capacity to absorb an urgent decision, which is exactly the asymptotic wait problem from why busy teams are slow teams applied to the most expensive server in the organisation. Slack in a senior calendar is not laziness. It is the buffer that keeps decision latency bounded, and it is the cheapest capacity you can buy.

What to do on Monday

Pull your own recurring meetings for the last month and, against each one, write the decisions it produced. Not the topics discussed — the decisions, with outcomes. Many will have none. That is your candidate list.

Then take your weekly leadership meeting and change one thing: open it with a list of decisions waiting, sorted by how many days each has been open, and work down the list until the time is gone. Do not do a round-robin. If you get through the list in twenty minutes, end the meeting. Run it that way for a month and compare your decision ageing before and after — that comparison is the only evidence anyone needs to keep the change.