How to Run a Weekly Owner Review
Why this matters
You already know your week got away from you. What you do not know is where, or whether it was the same place as last week. Without a fixed weekly look backward, every stolen hour reads as a one-off, so you never see that the same four questions ate a full working day across five days. This review is 45 minutes, it happens on the week that just ended, and it produces exactly one written change. It is not a plan for next week and it is not a team meeting. It is you, reading the evidence your own calendar and phone already collected, and closing one leak.
Step 1: Fix the slot and cap it at 45 minutes
Same day, same time, every week. Friday late afternoon works because the week is fresh and nothing new is landing. Monday first hour works if Friday is your worst day. What does not work is "when I get a chance", because a review that floats never happens and a review that runs 90 minutes gets cancelled by week three.
Cap it hard at 45 minutes. The cap is what makes it survive a bad week. If you cannot finish all five passes, you finish the ones you got to and you still write the one change. A partial review beats a skipped one, and the tally in Step 3 is worthless with gaps in it.
Step 2: Reconcile the calendar, planned against actual
Open last week's calendar next to what actually happened. For each block you had set aside for work on the business (planning, quoting strategy, hiring, pricing, the thing that is not a call), mark it kept, partially kept, or eaten. Write the two numbers: hours planned, hours actually protected.
This is the pass owners skip because they think they already know the answer. They do not. The felt sense is "I got nothing done"; the calendar usually shows a specific pattern, like every block after 2 p.m. survived and every block before it got eaten by the morning board.
Do not reconcile field hours here. If you still run calls, those hours are already accounted for on the job record. This pass is only about the blocks you set aside and then lost, because those are the ones nobody else is tracking.
Step 3: Tally the interruptions by class, not by count
Through the week, every time someone reaches you with a question that pulled you off what you were doing, you make a tally mark and write two or three words for the class. Pricing question. Schedule conflict. Parts. Customer complaint. Approval. That is the whole method, and it takes about four seconds per interruption.
At the review, sort the marks into classes and count each. You are not looking at the total. The total is discouraging and not actionable. You are looking for the class that dominates, because a single class covering a large share of your interruptions is a single missing written answer, and one written answer is a thing you can actually produce this week.
The rule: any interruption class accounting for 25% or more of the week's tally (unit: per week, on a tally of at least 20 interruptions) gets one standing written answer before the next review, AND the class must have appeared at that level in two consecutive weekly tallies. Both gates, not either. One week at 25% is often an artifact of one bad job. Two weeks running is a structure.
The step size is deliberate: one standing answer per review. Not the whole list. An owner who writes six standing answers in one sitting produces six documents nobody reads, because none of them was written under the pressure of a real recurring question.
Step 4: Sweep the open loops and the check-back queue
Two short lists.
Open loops are things you are personally holding that have no next action and no owner: a customer waiting on a call back, a quote half-written, a vendor dispute, a decision you have deferred twice. Write each one, then either name the next action and when, or hand it off, or kill it. An open loop you have carried three weeks running is not a task, it is a decision you are avoiding, and naming it as such is usually enough.
The check-back queue is the handoffs currently in flight. For each one you handed off recently, note which check-back cycle it is in and whether it came back to you this week. A handoff that returned is not a failure, it is data. Write the cause in one word: trigger, authority, information, or skill.
Step 5: Write the one change and the one commitment
The review ends with two lines on paper, not in your head.
The change is the standing answer from Step 3, or the gap closed from Step 4, whichever the evidence pointed at harder. One thing.
The commitment is the single block next week you will defend at the cost of something else. Name what it is for and name what you will drop if the week comes at you. An unnamed sacrifice means the block dies quietly.
A worked review
An owner of a five-tech shop runs the review for the first time on a 52-hour week.
Calendar pass. He had blocked 9 hours of on-the-business time across the week: three 2-hour blocks and one 3-hour Friday block. Actual protected time was 2.5 hours. So 6.5 of the 9 planned hours were eaten, about 72% of the 9. The Friday block survived whole; every weekday block died before it reached the halfway point.
Interruption pass. 31 tally marks across 5 working days, roughly 6 a day. Sorted: 19 pricing questions on jobs already quoted, 5 schedule conflicts, 4 parts availability, 3 miscellaneous. The pricing class is 19 of 31, about 61% of the week's 31 marks. The previous week's tally, which he had also kept, showed 16 of 27, about 59%. Two consecutive weeks above 25% on a tally above 20, so both gates are met.
Diagnosis. He had read this as a discipline problem, that he was letting himself get pulled off. The tally says otherwise. Nearly two thirds of his interruptions were one question with one shape: a tech on site finds the job is bigger than quoted and calls to ask what to charge. That is not a discipline problem, it is a missing authority boundary.
The one change. A single page: on any job already quoted, the lead may approve added work up to a stated multiple of the original quote without calling, and anything above that multiple, or any change to scope the customer disputes, comes to the owner. He picked the multiple by looking at the last twenty of these calls and finding the point below which he had approved every single one without thinking. Approving what you already always approve is the cheapest authority you will ever give away.
The commitment. The Tuesday 2-hour block, defended, for the hiring pipeline. What he drops if the week comes at him: the Tuesday afternoon supplier visit, which can be a phone call.
Reading the next review against the last one
Four weeks later the same passes read: 34 interruption marks, of which 7 are pricing. The class fell from 19 to 7, about a 63% drop in that class against its own starting count of 19, while the total tally went up slightly because he was tallying more consistently by then. Protected time went from 2.5 hours to 6 of 9 planned.
Note what did not happen. The pricing class did not go to zero, and expecting zero is how owners conclude the standing answer failed. Seven calls a week is the residue you should expect: the genuinely above-the-line jobs plus a couple where the lead wanted a second read. That residue is the boundary working, not leaking.
The read on a standing answer that is not working: if the class has not fallen by at least half against its own starting count within three weekly reviews, the answer is written wrong, not being ignored. Almost always it described a process instead of granting an authority. "Check the job scope carefully before adding work" is a process and changes nothing. "You may approve up to X without calling" is an authority and changes everything.
Spending the 45 minutes
The cap only works if the passes are budgeted, because the calendar pass will eat the whole slot if you let it. Rough allocation:
- Calendar reconciliation, 5 minutes. It is two numbers. Owners spend twenty minutes here re-litigating why each block died, which is interesting and produces nothing, because the reason is almost always in the interruption tally anyway.
- Interruption tally, 10 minutes. Sorting and counting. Most of this is the sorting, and the sorting gets faster after about three weeks as your class names stabilise.
- Open loops, 10 minutes. The longest pass in a bad week and the one worth protecting, because an open loop carried three weeks is the thing most likely to cost you a customer.
- Check-back queue, 5 minutes. Quick, and often zero if nothing came back.
- The change and the commitment, 15 minutes. This is the largest allocation and it surprises people. Writing a standing answer that actually grants an authority takes real thought: you have to look at what you have historically approved and decide where the line goes. Fifteen minutes is roughly what one good boundary costs.
If a week runs short, cut the calendar pass and keep the last one. A review that produces no written change is a review you will stop running within a month, because it will correctly feel like it did nothing.
What changes the shape of this review
If you have no team yet. The interruption tally still runs, but the classes will be customers and suppliers rather than techs, and the standing answer is usually a published policy rather than a delegated authority: your callback window, your after-hours rule, your deposit terms. Same instrument, different lever.
If your week is genuinely seasonal. In peak season the calendar pass will read badly every week and you will be tempted to stop. Keep the tally, drop the calendar pass. The interruption classes in peak season are the most useful data you will collect all year, because they are the ones that recur under load.
If you are the only one who reads it. That is the design. This is not a report. If you find yourself writing it for an audience, you will start softening the numbers, and a softened calendar pass is worse than no calendar pass.
How to verify it is working
Three readings, in order of how quickly they move:
- The tally's top class changes. Within four to six reviews, the class that dominated week one should no longer be the largest. If the same class leads for six consecutive weeks after you wrote its standing answer, the answer is the problem.
- Protected hours rise as a share of planned hours, not in absolute terms. Going from 2.5 of 9 to 6 of 9 is the win. Going from 9 planned to 4 planned so you can hit 4 of 4 is not.
- A handoff completes its check-back cycle and stops appearing. If your check-back queue only ever grows, you are handing off faster than the handoffs are landing, and the review has become a list of things you are worried about rather than an instrument.
References
- See related: The Sunday Setup: Starting the Week Ahead
- See related: Batching vs Switching: The Cost of Interruptions
- See related: How to Hand Off a Recurring Task Permanently
- See related: The Two-Hour Rule: Time on the Business
- SBA guidance on owner time management and delegation in small businesses