ingridleiria Operating Cadence Design
When the rhythm is absent, coordination does not stop happening, it just happens by interruption. Decisions get made in corridors and messaging threads by whoever is available, information reaches the leadership team late and by accident, and the person holding the middle of the company spends the week as a human router, carrying context between people who should have been in the same room on a known date.
The cost shows up in three places. Decisions take weeks that should take days, because there is no scheduled moment at which they can be made and nobody wants to call a meeting. The same topic gets discussed in three separate meetings and owned by none. And the calendar fills with recurring sessions that nobody can cancel, because cancelling one would remove the only place a particular conversation happens, even though nothing is ever written down afterwards. A cadence is the set of moments where a company looks at itself, decides, and adjusts, and designing it is cheaper than living without it.
When to use this, and when not to
Use it when a leadership team has no rhythm or a broken one, when the calendar has grown by accretion, when a company has changed size or shape and the old rhythm no longer fits, when decisions are visibly slow, when a new leader wants to establish how they will run, or when someone asks for a management operating system.
Do not use it to decide what the company is doing; that is strategic-plan-and-action-plan. Do not use it to set the goals the rhythm reviews, which is okr-planning, or to run the planning cycle itself, which is annual-planning-and-headcount. Do not use it to write the weekly report, which is weekly-status-update, or the record of a single meeting, which is meeting-to-decisions; both of those are artefacts this design specifies and then hands off to. Do not use it for a single badly run meeting, which usually needs an agenda and a pre-read rather than a redesigned rhythm, and do not use it to run a specific programme's coordination, which is program-management.
The boundary: this skill decides which recurring moments exist and what each one must produce. What happens inside them belongs to the skills named above.
What you need before starting
The current recurring calendar, with attendees and durations. Missing: reconstruct it by asking three or four people to forward a normal week, which takes an hour and is more accurate than a list written from memory. Where a calendar tool is connected, export the recurring series directly; where it is not, the forwarded weeks are sufficient.
The last five significant decisions and where they were actually made. This is the diagnostic that matters most, because a cadence problem shows as decisions made in corridors or not made at all. Missing: ask the leader directly to name five and say where each was settled.
The company's size, shape and time zone spread. A fifteen-person company, a sixty-five person company across four countries and a four-hundred-person division need different rhythms, and distributed teams need more written cadence and fewer synchronous meetings. Missing: ask, because designing for the wrong shape produces a rhythm that is quietly ignored.
What the leader wants to stop doing personally. The strongest single input, because it identifies which meetings exist to compensate for a missing artefact. Missing: ask what they read on a Sunday evening, and why.
The planning and reporting dates already fixed. Board meetings, month-end close, payroll, quarterly scoring. Missing: get these before proposing anything; a cadence that collides with the close will be abandoned in the first month.
The method
Inventory before designing. For each recurring meeting: name, frequency, duration, attendees, stated purpose, actual purpose, and the artefact it produces if any. Multiply duration by attendees to get weekly hours consumed. The gap between stated and actual purpose is where most of the redesign comes from.
Trace the decision flow. For the last five significant decisions, record where they were made, who was present, and the elapsed time from question to decision. Decisions made outside any scheduled moment tell you which forum is missing; decisions that took six weeks tell you which forum meets too rarely for the altitude of its content.
Find gaps and duplicates. Information reaching leadership late or by accident, topics discussed in three meetings and owned by none, and meetings that exist only because a tracker or a report is not trusted. That last category is important: it is repaired by fixing the artefact, not by keeping the meeting.
Present the diagnosis as one page before proposing anything. Hours consumed, decisions made outside forums, gaps, duplicates. A proposal that arrives without a diagnosis is read as a preference; the same proposal after a page of evidence is read as a finding.
Assign each topic an altitude and match it to a frequency. Weekly for execution and blockers, monthly for metrics and resourcing, quarterly for strategy and priorities, annually for direction and budget. A weekly meeting that discusses strategy and a quarterly that discusses tickets are both broken, and the fix is to move the topic rather than change the meeting.
Give every meeting one owner and one named artefact. The owner is responsible for the agenda, the pre-read, the record and the follow-up. The rule that does most of the work: if nothing is written afterwards, the meeting did not need to happen synchronously. Where a meeting cannot name its artefact, delete it and see what breaks.
Move status to writing and reserve meeting time for exceptions and decisions. Status is written and circulated beforehand and read before the meeting starts, or read in the first five minutes in silence, which works better than expecting people to have read it. The artefact is specified by
weekly-status-update; the meeting is for what the writing could not settle.Consolidate rather than add. Fewer, longer, better: one well-run ninety-minute leadership meeting beats three forty-five minute ones, because context is loaded once. Every new meeting in the proposal names the meeting it replaces or absorbs, and the proposal shows total recurring hours before and after.
Treat time zones as a design constraint. Rotate the pain of a synchronous meeting rather than fixing it on one region permanently, and move anything that can be asynchronous into writing. A distributed company that copies a co-located rhythm ends up with a second, undocumented rhythm in the region that was not designed for.
Roll out once, publicly, with a date. Announce the whole change at a single moment, say what is being removed and why, and start the new rhythm on a named date. Changing a cadence in pieces produces a period where both rhythms are half-running and neither is trusted.
Set a review date eight to twelve weeks out and put it in the calendar during the rollout, so that what does not work gets fixed rather than tolerated.
A reference cadence for a scale-up
Adapt this, do not copy it. The shape holds for a company of roughly fifty to three hundred people; a smaller company collapses several rows into one, and a larger one adds a layer below the leadership meeting.
| Cycle | Moment | Attendees | Duration | Artefact |
|---|---|---|---|---|
| Daily | Written check-in in the team channel | Each team | 5 min async | Three-line updates |
| Weekly | Leadership meeting: exceptions, decisions, blockers | Chief executive and direct reports | 60 to 90 min | Decision log entries, action list |
| Weekly | One-to-ones | Chief executive with each report | 30 min | Private notes, follow-ups |
| Weekly | Rollup to the chief executive | Chief of staff | Written | Leadership status rollup |
| Fortnightly | Pipeline or product review | Function lead and team | 60 min | Updated forecast or roadmap |
| Monthly | Metrics and financial review | Leadership, finance | 90 min | Dashboard, variance notes, reforecast |
| Monthly | All-hands | Whole company | 45 min | Recording, questions log |
| Quarterly | Priority setting | Leadership, then teams | Half day plus a week of cascade | Quarterly priorities document |
| Quarterly | Board meeting | Board, chief executive, finance | Half day | Board pack, minutes, follow-up list |
| Annually | Strategy and budget offsite | Leadership | Two days | Annual plan, budget, priorities |
| Annually | Performance and compensation cycle | Managers, people team | Multi-week | Reviews, decisions |
How to collapse it below about fifty people
Merge on shared attendees, not on shared topic, and work up the table from the bottom. With one leadership layer, the weekly leadership meeting and the fortnightly function review are the same room meeting twice, so keep the weekly one and delete the fortnightly. The weekly rollup goes, because the leadership meeting is the rollup. The all-hands and the leadership meeting become one weekly session of thirty minutes. Quarterly priority setting keeps its half day and loses the week of cascade, because there is nobody below to cascade to. Below about fifteen people, fold the monthly metrics review into the weekly session and run three moments in total, as the second scenario does. Four things survive every collapse: one weekly decision moment, one look at the numbers each month after close, one quarterly priorities session, and a decision log.
How to add a layer above about three hundred people
Do not lengthen the leadership meeting, and do not add attendees to it. Add a tier beneath it and give that tier its own artefact. Each function runs its own weekly leadership meeting at the same frequency as the one above, producing its own exceptions and decisions list, and the company leadership meeting reads those lists rather than hearing them narrated. Two things have to be designed at the same moment or the new tier degrades into a status meeting within a quarter. First, decision rights: write down what each tier decides alone and what it escalates, because anything unwritten floats upward by default. Second, the monthly metrics review splits in two, a functional review inside each function first and then one leadership review that looks only at the variances the functional reviews escalated. Above roughly a thousand people, or wherever the business has genuinely separate units, repeat the whole reference table inside each unit and keep only the quarterly and annual moments at group level.
Worked example
Situation. Alder Health Systems, 145 people across three countries, eight months after a funding round that had doubled headcount. The chief executive's complaint was that she was in meetings from Monday to Thursday and still learned about a slipped launch from a customer. The leadership team had grown from five to nine.
Task. A cadence proposal within two weeks, and a rollout that did not require a month of negotiation with nine leaders.
Action. The inventory took four days and produced the number that carried the whole proposal: 61 recurring meetings, consuming 412 attendee-hours a week, of which 143 hours sat in meetings with no artefact of any kind. Two meetings had the same six attendees and overlapping agendas, one owned by product and one by engineering, and neither owner had known the other existed until the inventory was circulated.
The decision trace was worse and more useful. Of the last five significant decisions, three had been made in one-to-ones with the chief executive and communicated afterwards, one had been made in a messaging thread over eleven days, and one had not been made at all after seven weeks. There was no forum in which a cross-functional decision could be taken, so every decision routed through one person, which was the actual cause of her calendar.
The first proposal was wrong and was abandoned after a day. It added a weekly cross-functional decision forum on top of the existing calendar, which would have taken the total to 62 meetings and 430 hours. The chief executive's reaction was that it solved her problem by making everyone else's worse. The rewrite made the leadership meeting the decision forum: it moved from a 60 minute status round-robin with nine people to a 90 minute session with a written pre-read due 24 hours ahead, an agenda of exceptions and decisions only, and a decision log maintained in the record. Status left the meeting entirely and became a written rollup.
Consolidation followed. The overlapping product and engineering reviews merged into one fortnightly session with a single owner. Eleven recurring meetings were deleted outright, all of them ones that could not name an artefact, and four were converted to written updates. The monthly metrics review moved to the fourth working day after close, because it had previously run before the numbers were final and spent its first twenty minutes on figures that changed later. Time zones were handled by moving the leadership meeting to 15:00 central and rotating the fortnightly product review between two slots so the same region did not always take the late one. Rollout was one announcement at an all-hands with the before-and-after hours on a single slide, and the new rhythm started on the first Monday of the following month.
Result. Recurring meeting load fell from 412 to 289 attendee-hours a week, a reduction of about 30 percent. Nine decisions were logged in the leadership meeting in the first six weeks, against three in the previous six. The chief executive's own recurring load fell from 23 hours to 14.
Two things did not work and were fixed at the eight-week review. The written pre-read was late in four of the first six weeks, which was solved by moving the deadline from 24 hours to 48 and by the chief of staff chasing on a fixed schedule rather than ad hoc. And one deleted meeting turned out to have been load-bearing: a weekly support and engineering triage, whose absence produced a backlog of unrouted escalations within a fortnight. It was reinstated at 30 minutes with a named artefact, which is the correct outcome of a review rather than evidence the design was wrong.
A second scenario, where it goes differently
An eighteen-person company with a founder, no management layer and everyone in one time zone.
Almost all of this method is too heavy. The inventory takes twenty minutes and the decision trace is unnecessary, because everyone already knows where decisions are made, which is wherever the founder is. The risk is the opposite one: imposing a scale-up rhythm on a company this size creates ceremony that consumes the founder's week and slows the thing that is currently working, which is that anybody can decide anything in ten minutes.
The right design is three moments and one artefact. A weekly all-hands of thirty minutes that doubles as the leadership meeting, a monthly metrics review, and a quarterly half day on priorities. The one artefact is a decision log, because that is the thing an eighteen-person company loses first as it grows and the thing that is most expensive to reconstruct later.
What changed: the number of moments, the depth of diagnosis, and the direction of the risk, which is over-designing rather than under-designing. What did not: every moment still has one owner and one named artefact, and the review date is still set.
Output
The diagnosis, one page: current meeting count and weekly attendee-hours, hours in meetings with no artefact, where the last five decisions were made and how long each took, and the named gaps and duplicates.
The proposed cadence:
| Cycle | Moment | Owner | Attendees | Duration | Artefact | Replaces or absorbs |
|---|
Hours before and after, as a single comparison. Where this is charted, keep it monochrome, separate before and after by fill texture rather than colour, put the legend outside the plot area, and reserve one accent for the category that changed most.
Agenda templates for the two or three most important meetings, with time allocations:
LEADERSHIP MEETING, 90 MINUTES
Pre-read circulated 48 hours ahead; assumed read.
00:00 Decisions to be taken today (from the pre-read) 40 min
00:40 Exceptions: workstreams at risk or blocked 25 min
01:05 One deep dive, named in advance 20 min
01:25 Confirm decisions, owners, dates 5 min
Artefact: decision log entries and action list, circulated same day.
The rollout plan: what changes, what is deleted, who is told and how, the start date, and the review date eight to twelve weeks out.
Failure modes
Designing before diagnosing. Recognise it when the proposal has no hours figure in it. Do the inventory; the number is what makes the change acceptable to people losing a meeting.
Adding a meeting to fix a meeting problem. Recognise it when the proposed total is higher than the current one. Make the existing forum work harder before creating a new one.
A meeting with no artefact. Recognise it by asking the owner what is written afterwards and getting a pause. Either name the artefact or delete the meeting.
Status consuming the decision forum. Recognise it from the last four agendas: if more than half the time went to round-robin updates, the written artefact is missing or nobody trusts it.
Rolling out in pieces. Recognise it when two rhythms are half-running. Announce the whole change once, on a date.
The rhythm that ignores the close and the board calendar. Recognise it when the monthly review keeps being rescheduled. Anchor recurring moments to fixed external dates rather than to a weekday.
No review date. Recognise it when a rhythm that is not working has survived six months. Set the date during the rollout, not after.
Edge cases
A leader who will not give up their one-to-ones as the decision channel. Do not fight it directly. Introduce the decision log first, so that decisions taken in one-to-ones become visible, and let the visible duplication make the case for the forum.
A company in a crisis or an incident. Suspend the normal rhythm deliberately and say for how long, replacing it with a daily standing session and a single written update. An undeclared suspension is how a cadence dies permanently.
Fully asynchronous or heavily distributed teams. Replace most synchronous moments with written ones on a fixed publishing schedule, keep one synchronous decision forum, and be stricter about artefacts, since writing is the only channel that reaches everyone.
A newly appointed leader inheriting a calendar. Do not redesign in the first month. Run the inventory and the decision trace, attend everything once, and propose in month two, when the diagnosis carries evidence rather than assumption.
Quality bar
- Every recurring meeting in the proposal has one named owner and one named artefact.
- The diagnosis states current and proposed weekly attendee-hours, and any increase is justified in writing.
- Topic altitude matches frequency: no strategy in the weekly, no tickets in the quarterly.
- Every new meeting names the meeting it replaces or absorbs.
- Decisions have a named home and a log.
- Status is written and circulated before the meeting it feeds.
- Synchronous burden across time zones is rotated rather than fixed on one region.
- The rollout has a single start date and a review date eight to twelve weeks out, both in calendars.
Adapting this to your context
The reference cadence, the meeting lengths and the eight to twelve week review come from venture-funded scale-ups of fifty to three hundred people with one leadership layer and two or three overlapping time zones.
- The frequency ladder. Weekly, monthly, quarterly, annual assumes a quarterly business rhythm. Where the natural cycle is an academic year, a public budget round or a growing season, move priority setting onto that cycle.
- Sixty to ninety minutes for the leadership meeting. Sized for five to nine direct reports with a written pre-read. More than nine is a signal to add a tier rather than to extend the meeting.
- Attendee-hours as the headline diagnostic. In billable or shift environments, count lost billable hours or the cost of covering the shift, because that is the number the reader will actually feel.
- The monthly review on the fourth working day after close. Anchor it to your own close timetable. A finance function that closes on day ten cannot hold a day-four review.
- What not to change. Every recurring moment has one named owner and one named artefact, and status is written and circulated before the meeting rather than narrated inside it.
Related skills
strategic-plan-and-action-plan sets the direction the quarterly moments review, and okr-planning sets the goals the monthly and quarterly moments score. annual-planning-and-headcount supplies the planning calendar this rhythm has to accommodate. weekly-status-update is the artefact the weekly rhythm produces, and meeting-to-decisions is the artefact each individual meeting produces; this skill specifies that both must exist and where they land. decision-memo is what a leadership forum should receive when a decision needs preparing rather than discussing. program-management handles coordination inside a single large programme rather than the company rhythm.