Most business owners do not have a review problem. They have a Sunday night problem.
You know the one. It is 8pm, the week is technically over, and instead of closing the laptop you sit there with a vague, itchy feeling that you were busy for five days and are not entirely sure what came of it. You cannot remember what you decided on Tuesday. You know a lead went quiet but not which one. And Monday is going to start whether you have a plan or not.
Big companies solve this with a leadership meeting. Somebody pulls the numbers, somebody builds the deck, and by Monday the whole company knows what matters. You have none of that.
Here is the good news. You can run the same review, at the same quality, in about twenty minutes, and Claude can do most of the heavy lifting. I have trained more than 7,500 people at this point, and the weekly CEO review is the single habit that changes the most for the least effort. Not because it is clever. Because it is the first time all week that somebody senior looks at the business, and that somebody is you.
Quick definitions
CEO review: a short, repeatable weekly look at your numbers, wins, misses, and next week's priorities. A decision meeting, not a status update.
Project: a saved workspace inside Claude that holds your business context, so every chat inside it already knows who you are and what you sell.
Skill: a saved set of instructions Claude follows the same way every time, called by name. Think recipe card, not conversation.
Connector: a safe link between Claude and a tool you already use, like Gmail or Google Calendar, so it can see the real week instead of the one you remember.
Step 1: Decide the five questions your review answers (5 minutes, once)
Before you open Claude, get clear on what a CEO review actually is. It is not a summary of everything that happened. Nobody needs that, including you. It answers five questions and then stops.
One: What did the numbers do this week?
Two: What worked, and is it repeatable?
Three: What did not work, and why?
Four: What is stuck, waiting on me, or quietly rotting?
Five: What are the three things that matter most next week?
That is the whole thing. Five questions, every week, forever. The power is in the sameness. When the questions never change, you start seeing patterns across weeks instead of just reacting to the current one, and patterns are where the actual money is.
Notice what is not on the list: everything you did. If you catch yourself writing a list of tasks to prove you were busy, you have slipped back into being an employee of your own company. Stay in the chair.
Step 2: Set up the workspace so you stop re-explaining yourself (20 minutes, once)
Here is the thing that makes this a twenty-minute habit instead of an hour-long chore: Claude needs to already know your business before you sit down. Otherwise you spend the first ten minutes every week typing "I run a service business, my offer is, my clients are" and by then you have lost the will to live.
Make a project called Weekly CEO Review. In the project instructions, put:
What you sell, in plain terms, with prices. Not marketing language. The real thing.
Who you sell it to, and what they are usually trying to fix.
The numbers you actually care about. Three to five, maximum. Revenue, new clients, leads, calls booked, members. Pick the ones you would check if you could only check one screen.
Your current goal, with a number and a date. "Get to twenty retainer clients by December" beats "grow the business" every time, because one of them can be measured against and one cannot.
How you want to be talked to. Mine says something close to: be direct, tell me what I am avoiding, do not soften the misses, and do not congratulate me for things that were not hard.
That last line is the difference between a review that changes something and a review that makes you feel nice. You are not paying for encouragement. You have friends for that.
Step 3: Gather your inputs (5 minutes a week, and yes it is this boring)
The review is only as good as what you feed it. This is the part everybody wants to skip and it is the part that makes the whole thing work.
You have two ways to do it, and both are fine.
The manual way. Open a note. Paste in this week's numbers (whatever you can get in two minutes from Stripe, your CRM, your booking tool, your bank). Then brain-dump for three minutes: what happened, what you are worried about, what you keep putting off. Do not organize it. Bullet fragments are perfect. Then paste the whole mess into Claude.
The connected way. Turn on the connectors for the tools you already live in. Google Calendar so it can see what your week actually contained instead of what you think it contained. Gmail so it can spot the threads that went quiet. Google Drive so it can pull the notes you wrote during the week. Now your review is built on evidence rather than memory, and memory is a liar, especially on Sunday night.
Start manual if connecting things feels like a project. A messy paste beats a perfect system you never set up.
Step 4: Write the review prompt once, then save it as a skill (10 minutes, once)
Now the actual instruction. Here is the one I use, cleaned up so you can paste it straight in:
The prompt
You are my chief of staff running my weekly CEO review. Here are this week's inputs. Give me, in this order: (1) a numbers table with this week, last week, and the change, plus one sentence on what the trend actually means; (2) three wins, and for each one tell me whether it was repeatable or luck; (3) the misses, said plainly, with your best guess at the cause; (4) anything stuck, waiting on me, or going cold that I have not mentioned; (5) exactly three priorities for next week, ranked, each with the first concrete action and roughly how long it will take. Then end with one question I am avoiding. Be direct. Do not pad. If the inputs do not support a conclusion, say so instead of guessing.
Two lines in there do most of the work.
"Repeatable or luck." A win you cannot reproduce is a nice week. A win you can reproduce is a system waiting to be built. Once you sort wins this way every Friday, you stop chasing the ones that were never going to happen twice.
"One question I am avoiding." Uncomfortable, on purpose. Mine has surfaced things like whether a client relationship is actually working, and whether I am building the thing I want or the thing that is easiest to sell. You will want to skip this line. Do not skip this line.
Once the prompt gives you output you like, save it as a skill named something obvious, like weekly CEO review. From then on you paste your inputs, call the skill by name, and the format is identical every week. That is what makes week twelve so much more valuable than week one, because by then you are comparing like with like.
Step 5: Run it, and argue with it (10 minutes a week)
The output is a draft, not a verdict. This is where most people go wrong. They read the summary, nod, and close the tab, and nothing changes, because a review you merely read is just content.
Push back on it. Real things I say most weeks: "that is not why that missed, the real reason is I never followed up." "Priority two is not actually a priority, it just feels urgent, replace it." "You called that a win. It was one client and they were already warm. Downgrade it."
Every correction makes the next week's review sharper, because you are teaching it how your business really works. After about a month, it starts catching things before you do, and that is a genuinely strange and wonderful feeling.
Step 6: Turn the three priorities into actual calendar time (5 minutes)
A priority that is not on your calendar is a wish. Ask for it directly: "Turn these three priorities into calendar blocks for next week. Look at what is already on my calendar and put the hardest one in my best thinking time. Give me the block titles and times."
If your calendar is connected, it can see where the gaps genuinely are instead of pretending Tuesday is wide open when you have back-to-back calls. If it is not connected, paste in your week and let it work from that.
This step is short and it is the one that converts a review into a different week. Everything before it was analysis. This is where the analysis touches your life.
Quick tip
Put the review on your calendar as a recurring appointment with yourself, same time every week, and treat it like a client call. Friday at 4pm or Monday at 8am both work. The specific slot matters far less than the fact that it repeats. Habits survive on rails, not on willpower.
What this looks like after a month
Week one feels awkward. Your inputs will be thin and the output will be thinner than you hoped. Do it anyway. By week three you will notice the same miss keeps appearing, and it will bother you enough to fix it, which is exactly the point.
By week six you will have something most small businesses never build: a written record of your own decisions. You will catch a lead going cold on day four instead of day thirty. And Sunday night will get quieter, because the itchy feeling was never about working harder. It was about not knowing.
Frequently asked questions
How long does the weekly CEO review actually take once it is set up?
About twenty minutes, and it gets faster. Roughly five minutes gathering inputs, five minutes reading the output, five minutes arguing with it, and five minutes putting the priorities on your calendar. The setup (the project, the prompt, the skill) takes about half an hour the first time and then you never do it again. If it is regularly running past thirty minutes, you are including too much detail in your inputs. Trim to the numbers plus a short brain-dump.
Do I need to connect Claude to my tools, or can I just paste things in?
Pasting works completely fine and is how I recommend everyone starts. Connectors make it better rather than possible, because they let the review be built from what actually happened instead of what you remember happening, which is a meaningfully different thing by Friday. Start manual for a few weeks. Once the habit is real, connect your calendar first, because it delivers the biggest jump for the smallest setup.
What if my numbers are messy or I do not really track anything yet?
Start with the three you can get in under two minutes, even if they are rough. Money in, new clients, and leads or inquiries will carry you a long way. A review built on three imperfect numbers you check every week beats a perfect dashboard you build once and never open. Rhythm has to come first, because the rhythm is what makes the numbers worth having.
What is the difference between this and just journaling about my week?
Journaling processes how you felt. A CEO review produces decisions. The structure is what makes the difference: the same five questions in the same order every week, ending in three ranked priorities that go on a calendar. Reflection with no output is a diary, and diaries are lovely, but they will not tell you that your best client has not replied in eleven days.
How do I stop it from being relentlessly positive about everything?
Tell it not to be, in the project instructions, and reinforce it when it slips. Something like "do not soften the misses, do not congratulate me for things that were not hard, and if the numbers are bad say the numbers are bad" works well. When it does hand you something too generous, correct it in the moment. A review that only ever makes you feel good is not doing its job.
What should I do if the review keeps telling me the same thing every week?
Listen to it. A repeated finding is not a bug in the process, it is the process working. If "follow-up is slipping" shows up four weeks running, that is not a note anymore, that is the most important project in your business. Make it priority one and watch it disappear from the review.
Where to go from here
You do not need a leadership team to run your business like a real one. You need twenty minutes, five questions that never change, and somebody honest in the room. Claude can be the honest one. You still have to be the CEO.
Start this week, even messy. Make the project, paste in whatever numbers you can find in two minutes, run the prompt, argue with the answer, and put three things on your calendar. That is a complete first review, and it beats the one you have been meaning to start since spring.
If you want the templates, the connector setup walked through step by step, and a room full of people comparing what their reviews surfaced this week, that is exactly what Club Jam on Skool is for. It is $47 a month with a 7-day free trial, and inside you will find 220+ modules covering projects, skills, and connectors, plus everyday business owners building the same systems and telling the truth about what worked.
Twenty minutes on Friday. Then close the laptop and actually have your weekend. 🧡