Operating cadence
Battle testedWhat should run weekly, monthly, quarterly and yearly at your stage, who runs it, and when the rhythm changes.
The layers
Each step has an owner, a time cost and a document. Costs marked estimate are worked out, not published data.
Refresh the numbers
Metric owners update their numbers. At Amazon, owners review on Monday, departments on Tuesday and the company on Wednesday.
Write the Friday update
Shipped, next, blocked, asks and metrics, in writing.
Hold the leadership meeting
A fixed agenda: scorecard, priorities, headlines, to-dos, issues. EOS runs it for 90 minutes; 60 is enough under 15 people.
Run the weekly business review
From about 15 people. Owners explain only the exceptions. Amazon’s company-level review is 60 minutes.
Update the decision log
One line per decision, at the end of each leadership meeting.
Close the month
Working days 1 to 10. APQC’s median is 6 calendar days (2017, large firms). Ventana: 59% close in 6 business days (2022). No reliable startup benchmark found.
Review the month
Plan against actual, the reasons for variance, the forecast and the top risks. Run it only once the close is stable.
Send the investor update
Highlights, lowlights, key metrics, cash and runway, asks. GitLab sends it around the 10th, give or take 1 to 3 days.
Plan the quarter
Review the last quarter and set 3 to 7 company priorities, closer to 3 is better. EOS gives each a single owner and a 90-day container.
Meet the board
Agree the rhythm with the lead investor. Published advice conflicts: every 6 to 8 weeks, or quarterly with written updates between. Send a written pre-read.
Build the annual plan
Start in October for a calendar year, with board sign-off before Q1. One practitioner reports 3 to 6 weeks of drafting at Series A.
Reforecast when plan and actuals drift
Agree the trigger up front, so March is not a fight about the budget. No reliable benchmark found for the right threshold.
Which layers do you run now?
Five questions at most. The rust line is your path. Each faint branch is where the other answer leads. Turn the dial to your stage; slide or type your headcount.
How many people work in the company?
- 1 to 5: Monday plan, Friday note
- 6 to 15: More questions
- 16 to 50: More questions
- 51 to 150: More questions
- 150 or more: Two-pass planning, tiered reviews
Can someone produce 5 to 8 weekly numbers in under one hour?
- No: Build the numbers sheet first
- Yes: Seed cadence
Do the books close within 10 working days of month end?
- No: Fix the close first
- Yes: More questions
Is there a board or an institutional lead investor?
- Yes: Series A cadence
- No: Advisers instead of a board
Does more than one function lead need to commit headcount or budget from a shared plan?
- Yes: Series B cadence
- No: Stay on the Series A cadence
Monday plan, Friday note
Run a 30-minute Monday plan and a written Friday note. Add a short monthly investor email if you have investors. No weekly business review, monthly review or board-style pack.
Two-pass planning, tiered reviews
Plan in two passes, a bottom-up plan and then a revision. Run weekly reviews at team, department and company level, with a capped metric list for the exec review. Amazon’s 400 to 500 metric deck is not a model for a smaller firm.
Build the numbers sheet first
Five metrics, one named owner and one source each. Delay the weekly review until the sheet exists: Commoncog reviews 47 metrics in under 15 minutes only because the data is ready.
Seed cadence
A weekly leadership meeting of 60 to 90 minutes on a fixed agenda: scorecard, priorities, headlines, to-dos, issues. Written Friday updates replace status talk, with an investor update around day 10 and a half-day quarterly planning session.
Fix the close first
Set a close calendar and one owner, outsource bookkeeping if needed, and hold a short monthly cash and burn review only. APQC’s median close is 6 calendar days (2017, not startup-specific). No reliable startup benchmark found.
Series A cadence
A weekly leadership meeting, a 30 to 45 minute weekly business review on 5 to 8 input metrics, a monthly review of plan against actual, and a 1 to 2 day quarterly planning session. Agree the board rhythm with the lead investor, because published advice conflicts.
Advisers instead of a board
Keep the same internal cadence. Replace the board with a quarterly check-in with two or three outside advisers, and keep the monthly written update to any investors.
Series B cadence
A formal annual operating plan with one named owner of the planning calendar. Department reviews feed one exec review, plus a quarterly business review. One Series A practitioner reports 3 to 6 weeks to build the plan, so start in October.
Stay on the Series A cadence
Re-check this question when a second function lead asks for headcount or budget. More planning layers without a second budget holder add ceremony and no decision.
Answer the last question to see your outcome.
Stage
Who runs the rhythm
Set by the dial on the instrument. It follows your headcount until you turn it.
- Runs it
- Founder or CEO, with a co-founder as backup.
- You keep
- Everything: the plan and the investor email.
- Hand off
- Bookkeeping, to an outsourced bookkeeper.
My stance
No calendar of rituals at four people. Talk all the time, keep everything open, write down who owns what.
- Runs it
- CEO, with the first ops hire or a Chief of Staff as backup.
- You keep
- Investor update, board relationship, quarterly priorities.
- Hand off
- The scorecard, the staff meeting agenda, the close with the bookkeeper.
My stance
No heavy structure yet. Async first, with the leaders aligned on what matters and open about everything else.
- Runs it
- CEO plus a head of ops or Chief of Staff, with a finance lead as backup.
- You keep
- Board, quarterly priorities, the narrative.
- Hand off
- Weekly review facilitation, monthly review prep, the close calendar.
My stance
Add a ritual when a fix needs one to stick. Drop it the moment it stops helping.
- Runs it
- CFO or VP Finance for the plan and monthly review. Chief of Staff or head of BizOps for the calendar and quarterly review.
- You keep
- Priorities, the board, the final cut of the plan.
- Hand off
- Plan build, board pack assembly, department reviews.
My stance
The annual plan is a calendar problem before it is a numbers problem. Fix next October’s dates in January and publish who owes what by when.
- Runs it
- CFO plus a strategy and operations lead, with a Chief of Staff to the CEO.
- You keep
- The narrative and the few decisions only the CEO can make.
- Hand off
- Everything that repeats, with a finance partner in each org.
My stance
Two passes, and writing before slides. Cap the exec review metric list: 400 to 500 metrics suits Amazon, not a 200-person firm.
Efficiency savers
What each step saves. Where nobody publishes the number, it says estimate.
Friday written update instead of a status meeting
No published data. A 10-person, 60-minute status meeting costs 10 person-hours a week. Writing for 15 minutes plus one 30-minute read costs about 3. Roughly 7 hours saved.
Estimate, no published data: estimateFixed-format scorecard
Same metrics, same order each week. One practitioner reviews 47 metrics in under 15 minutes. It needs ready data.
Commoncog, the Amazon weekly business review: secondary source (opens in a new tab)Review exceptions only
Owners explain only exceptions and say “nothing to see here” for routine variation. No timing data published.
Commoncog, the Amazon weekly business review: secondary source (opens in a new tab)Close calendar with owners and bank feeds
No reliable benchmark found. Roughly 2 to 5 days saved, as an estimate.
Estimate, no published data: estimateReuse the investor update as the board pre-read
No published data. Roughly 4 to 8 hours saved per board meeting.
Estimate, no published data: estimateWritten pre-read before the board meeting
The meeting becomes “decisional, not informative”. Qualitative only.
Unusual Ventures: secondary source (opens in a new tab)
How the rhythm goes wrong
Failure modes
- Ritual stack too early
Eight people, five recurring meetings. Add a layer only when the pain has shown for two months, and use the tree.
- Dashboard theatre
The review reads numbers aloud with no action. Review exceptions only, and drop an input metric that stops predicting the output.
- Metric proliferation
Hundreds of metrics and no owners. Give each metric one named owner and cap the list.
- Investor update only when good
Gaps in the months with bad news. Use a fixed template with lowlights and asks, and send it monthly.
The biases behind them
- Planning fallacy
The annual and quarterly plans run optimistic. Plan to the base rate of the last four quarters.
- Parkinson’s law
Meetings and prep fill the slot you give them. Give the review a short slot and a hard stop.
- Goodhart’s law
A tracked metric becomes a target and stops measuring. Pair each output metric with an input metric.
- Sunk cost
You keep a ritual because it was hard to set up. Ask each quarter whether you would start it today.
Where the evidence disagrees
Board cadence
Often
Every 6 to 8 weeks at seed and every 8 weeks at Series A, 2 to 3 hours each. Unusual Ventures, a VC firm.
Unusual Ventures: secondary source (opens in a new tab)Quarterly
Quarterly at seed, with written updates and 1:1s in the off-months. A 2026 seed playbook, also reported there.
Unusual Ventures: secondary source (opens in a new tab)No reliable benchmark found. A third source says 8 to 10 meetings a year after seed. Agree the rhythm with the lead investor.
Live staff meeting or async
Live
EOS says the Level 10 meeting is not a roundtable of updates. Use it to solve issues.
EOS Worldwide, Level 10 meeting: secondary source (opens in a new tab)Async
Async-first teams replace even the meeting with writing. GitLab also warns against using async to avoid a hard conversation.
GitLab handbook, sync and async: secondary source (opens in a new tab)The two can share one week: updates in writing, the meeting for issues.
Rules by country
GoBD record keeping
KPMG, GoBD amendment: secondary source (opens in a new tab)Applies to the books. The BMF issued a second amendment on 14 July 2025. For e-invoices, archiving the structured XML part is enough.
Receiving is required from 1 January 2025. Issuing phases in from 2027 for larger firms and 2028 for the rest. Secondary sources describe the turnover threshold inconsistently.
General information, last checked 29 Oct 2026. Not legal or tax advice.
Disagree with a stance, or have data that should be here?
Email me