Home / Blog / Weekly Operating Cadence
Day 29The Weekly Operating Cadence: One Page, Five Numbers, Same Time Every Week
Nobody notices the quarter going wrong in week two. That is precisely when it is cheap to fix.
Ask a struggling founder how the business is doing and you get a story. Ask a good operator and you get five numbers, this week against last week, and a sentence about which one they are working on. The gap between those two answers is not intelligence or effort. It is whether the company has a rhythm.
Without one, information arrives by accident. Someone mentions in a corridor that two customers cancelled. The bank balance surprises you in month three. A project everyone assumed was progressing turns out to have been blocked since the second week. Every one of those is cheap to fix on the day it happens and expensive to fix a quarter later.
A company without a cadence does not run slower. It runs blind, at the same speed.
Andy Grove made the argument in High Output Management decades ago: meetings are the medium through which managerial work is done, and the question is never whether to have them but whether they produce decisions. A weekly review that produces three decisions and two owners is not overhead. A weekly review where everyone reports status they could have written down is.
The five numbers
Five is not a style preference. It is the number a small team can actually hold in their heads and act on. A dashboard with forty metrics gets scrolled past, and a company that watches everything watches nothing.
Cash and runway. Balance today and months remaining at current burn. Never delegate this one out of the room, and never let it be reported monthly.
Revenue won. New and expansion, this week, against target. Actual signatures and payments, not pipeline.
Revenue lost. Churn and downgrades, named. Losing customers by name changes the conversation from a percentage into a set of decisions.
One quality measure. Activation rate, delivery on time, error rate, first response time. Whatever tells you the thing you sell is still good while you push for growth.
One leading indicator for the current bottleneck. This is the only slot that changes. If sales is the constraint, it is conversations booked. If delivery is the constraint, it is units shipped per person. If retention is the constraint, it is week-one activation. Change it when the bottleneck moves, and only then.
Each number gets a target, last week's value, and one person's name. A number without an owner is a fact. A number with an owner is a job.
Monday: the page
Numbers published before the meeting. Fifteen minutes on variances, blockers and decisions.
Midweek: unblock
Founder clears the specific obstacles named on Monday. No new initiatives introduced.
Friday: close
Update the numbers, update the 13-week cash forecast, write three lines on what changed.
Monthly: financials
Actuals against plan, margin by segment, and one structural question nobody has time for weekly.
Quarterly: reset
Three priorities for the next 13 weeks. Everything else is explicitly not being done.
The rhythm matters more than the format. Weekly catches execution problems, monthly catches financial ones, quarterly catches strategic ones. A company missing the weekly layer discovers execution problems on the monthly cycle, by which point four weeks of effort has already gone into the wrong thing.
Make the meeting produce decisions
Four rules turn a status meeting into an operating meeting, and they are all unpopular for the first fortnight.
Publish before you meet. The one page goes out an hour ahead, or the meeting starts with everyone reading it in silence, which is the practice Amazon made famous with its written memos. Either way, nobody spends meeting time reading numbers aloud to people who can read.
Discuss only variances. A number that hit its target gets no airtime. This feels wrong and it is correct: attention is the scarcest resource in the room and it belongs on what is not working.
Every action gets one owner and a date. Not a team, not "we should." One name, one date, in the document, visible to everyone.
Start with last week's actions. The first two minutes review what was committed to last week and whether it happened. This single habit is what makes the cadence real, because it is the only mechanism that connects a commitment to a consequence.
Under 5,000 a year. Cheap. Worth keeping if it produces even one decision a month. 5,000 to 25,000 a year. A real line item. It should produce decisions, not status. 25,000 to 75,000. You are spending the cost of a junior hire on this recurring meeting. Halve the attendees or the frequency. Over 75,000 a year on one recurring meeting. Cancel it for a month and see what actually breaks.
Attendees times duration times loaded hourly cost times annual frequency. Eight people, one hour, weekly, at 60 an hour loaded is about 25,000 a year and 416 person-hours. The point is not to abolish meetings. It is to know the price before you add another one, because recurring meetings are almost never cancelled.
Run that calculation on every recurring meeting in your calendar this week. Most founders find one that costs more than a part-time hire and produces nothing that could not have been an update in a document.
Audit your current rhythm
0 to 2. You are running on memory and adrenaline. Start with one weekly meeting and five numbers on Monday. 3 to 5. A rhythm exists but leaks. The missing items are usually ownership and reviewing last week's actions. 6 to 7. Solid operating system. The last item is the real test and it only gets scored during a bad month. 8 of 8. Rare. Protect it: the cadence is the first thing sacrificed when things get busy, and busy is when it earns the most.
Eight checks. Item eight matters most: any company can hold a meeting during a good quarter. The cadence is only real if it survives the week when everything is on fire, because that is the week when decisions get made badly and nobody is watching the numbers.
The quarterly reset
Weekly reviews keep you honest about execution. They do nothing about doing the wrong thing efficiently, which is what the quarterly reset exists to catch.
Three priorities, no more, written as outcomes rather than activities. "Cut time to first value from 3 days to under 4 hours" is a priority. "Improve onboarding" is a wish. Each one gets an owner and a measure, and each one should be something you would notice failing.
Then write the list of what you are explicitly not doing this quarter, and share it. That second list is the one that changes behaviour, because in most companies nothing is ever cancelled, it just gets quietly starved while everyone pretends it is still alive. Naming the dead projects returns real capacity and it removes the guilt attached to work nobody was going to do anyway.
Open a document. Write the five numbers with three columns each: last week, this week, target. Fill in what you know and write "not measured" where you cannot, which is itself the most useful output of the exercise. Put a name against every row, including your own. Then book 30 minutes at the same time every Monday, permanently, and send the page an hour beforehand. Do it for four weeks before judging it. The first week feels bureaucratic, the second reveals a number nobody was tracking, and by the fourth you will catch something in week two that would previously have surfaced in month three. That single catch usually pays for the entire practice.
Why cadences collapse
Too many numbers. The dashboard grows because adding a metric feels productive and removing one feels like giving up. At fifteen numbers, nobody reads any of them. Cap the list and force a swap when someone wants an addition.
The meeting becomes reporting. If people are describing what they did rather than deciding what to change, the format has failed. Move the reporting into the written page and reclaim the time for variances.
It gets suspended during a crisis. This is the fatal one. The crisis week is exactly when decisions are made under pressure with the least information, and a founder who cancels the review because things are busy has removed the instrument at the moment the readings matter most.
Nobody owns the numbers. Metrics that belong to everyone belong to nobody, and the review becomes a discussion where each person can explain why the figure is somebody else's fault.
The founder does not attend. If the cadence is optional for you, it is theatre for everyone else, and they will correctly allocate their effort accordingly.
Average detection delay is roughly half the review interval, so a monthly rhythm lets a problem run for about two weeks before anyone sees it and a quarterly one lets it run for six. Multiply that delay by your weekly burn to price your current cadence.
The takeaway
- Five numbers, one owner each, same time every week. Cash and runway never leave the list.
- Publish before the meeting, discuss only variances, assign one owner and a date to every action.
- Start each review with last week's commitments. That is what makes the cadence real.
- Calculate what each recurring meeting costs per year before you add another one.
- Quarterly, pick three outcomes and publish the list of what you are not doing.
- Never suspend the cadence during a crisis. That is the week it earns its keep.
Frequently asked questions
What is an operating cadence?
The fixed rhythm at which a company reviews numbers, decides what to change, and checks whether the last change worked: weekly for leading indicators, monthly for financials, quarterly for priorities. The value is in the regularity, because a problem found in week two costs a fraction of the same problem found in month four.
What metrics should a startup review weekly?
Five at most: cash and runway, revenue won, revenue lost, one quality measure, and one leading indicator tied to your current bottleneck. Each needs an owner, a target and last week's number beside it.
How do you run an effective weekly meeting?
Publish numbers beforehand or read them in silence at the start, spend the time only on variances, blockers and decisions, and assign every action to one person with a date. Begin the next meeting by reviewing those actions.
How much do meetings cost a startup?
Attendees times loaded hourly cost times duration times annual frequency. A weekly hour with eight people at 60 an hour is roughly 25,000 a year. Calculate it before adding a recurring meeting, because recurring meetings are rarely cancelled.
Nobody saw it coming, because nobody was scheduled to look.
Kill My Startup is about the failures that were visible for months to anyone with a routine for noticing.
Buy on Amazon →Sources
- Andrew Grove, High Output Management, on meetings as the medium of managerial work and on leading indicators.
- John Doerr, Measure What Matters, on objectives, key results and quarterly cycles.
- Amazon's written narrative and silent reading meeting practice, as described in its shareholder letters.
- Standard management accounting practice on loaded labour cost.