Measure your week before you fix it.
When you sell your time, your calendar is your capacity. Most service providers know their revenue to the dollar and their hours only as a feeling: busy, slammed, underwater.
This move turns two weeks of your calendar into numbers: hours by type of work and by client. Every move after this one uses what it shows you.
- Data Two weeks of your calendar, plus the work that never makes it onto it.
- Operations Sorting the work by type: delivery, sales, admin, and what only you can do.
- AI The sorting, the totals and the questions you'd skip.
I want to see where my working hours actually go. Below are two weeks of my calendar, plus the work that never makes it onto the calendar (email, messages, admin, the late-night catch-up). [upload screenshots of your calendar, or paste it here] [add the off-calendar work with rough hours, e.g. "email, about 1 hour a day"] 1. Sort every block into one of these: client delivery, client communication (calls, email, messages), sales and marketing, admin and money, internal operations, interruptions. 2. Total the hours for each type, and for each client. 3. Tell me what share of my week is paid client work and what share is everything else. 4. Mark each type of work as "needs my judgment" or "just needs someone with the context." 5. Give it all back as one table I can save. If a block is unclear, ask me what it was. Don't guess.
You can't make room in a week you've never measured.
How to run it
- Take screenshots of your calendar in week view for the last two weeks. If your Claude or ChatGPT is already connected to your calendar, ask it to read those two weeks instead.
- Jot down the work that never makes it onto the calendar, with rough hours.
- Paste the prompt, add both, and answer its questions.
- Save the table in your Project. Moves 2, 3, 7 and 10 use it.
Stop something before you add anything.
The fastest way to add capacity isn't a new tool or a new hire. It's stopping something. Most weeks are full of work that made sense two years ago: the report nobody reads, the weekly call that could be an email, the favor that turned into a standing task.
You won't find these by thinking about them. You find them when someone asks, task by task, what would actually happen if you stopped.
- Operations Every recurring task and what it's actually for.
- Data The hours each one takes, from Move 1.
- AI Asks the questions you'd skip and adds up what you get back.
Here's where my hours went over the last two weeks: [paste your table from Move 1] Go through every recurring task, one at a time, and ask me: - Who would notice if I stopped doing this for a month? - What would stopping cost: money, a client, or just my habit? Then sort each task into: Stop: nobody would notice, or the cost is small Shrink: do it less often, shorter, or a lighter version Keep: it earns its hours Total the hours a month I'd get back from Stop and Shrink. Be direct. I tend to keep things out of habit.
The fastest way to add capacity is to stop doing something.
How to run it
- Open your table from Move 1.
- Paste the prompt and answer each question in a line.
- Pick two things from the Stop list and stop them this week. Tell anyone affected in one plain sentence.
- Put a reminder on your calendar 30 days out: did anyone notice?
Find out which clients actually pay you.
Every client pays you one rate on paper and a different rate in real life. The real one is what they pay divided by every hour they take: the calls, the revisions, the messages after 6pm.
Work it out and the picture usually changes. This move shows who to keep as is, who to keep on new terms, and who to let go.
- Data What each client paid in the last six months, and the hours they take.
- Operations Scope creep, revisions, and how each client communicates.
- AI The rate math, the ranking and a draft of the hard conversation.
Here are my current clients. For each one: what they paid me in the last six months, roughly how many hours a month they take (meetings, delivery, messages), and anything that makes them easy or hard to work with (scope creep, late payments, constant changes, referrals they send). [upload your invoice or payment export, and paste the hours and notes] 1. Work out each client's real hourly rate: what they paid divided by the hours they took. 2. Rank them on that rate, and separately on how much time and energy they take outside the work itself. 3. Sort them into: keep as is, keep but change the terms (price, scope or how we communicate), or plan to let go. 4. For anyone in "change the terms," draft a short, kind message I could send. 5. Tell me how many hours a month I'd get back if I let go of the bottom group, and how much revenue I'd need to replace. Use only the numbers I give you. If a number is missing, ask me instead of estimating.
Not every client is worth the hours they take.
How to run it
- Export the last six months of invoices or payments as a CSV from wherever you bill (Stripe, QuickBooks, your bank).
- Add rough hours per month for each client. Your Move 1 table helps.
- Upload the export and paste the prompt. Initials instead of names are fine.
- Pick one client to act on this month: a new price, a tighter scope, or a kind goodbye.
Charge for results, not hours.
If your revenue only grows when your hours grow, you've hit the ceiling of trading time for money. The way through is pricing: charging more where demand is strong, and charging for a result instead of the hours it takes.
Most service providers set their prices once and then just get busier. This move shows you, with your own numbers, where a price increase costs you the least.
- Data Price, hours to deliver and sales for each offer.
- AI Your real hourly rate on each offer, and what different prices would do to your hours.
Here's what I sell. For each offer: the price, how many hours it takes me to deliver (including calls, emails and revisions), how many I sold in the last 12 months, and roughly how many people I offered it to. [paste your list] 1. Work out my real hourly rate on each offer. 2. Flag where my rate is lowest and demand is highest. Those are the likely candidates for a price increase. 3. For the top candidate, model a 15%, 25% and 40% increase: how many fewer clients could I take and still earn the same? How many hours would that free up? 4. Point out anything I charge by the hour that could be a fixed price for a defined result, and what the scope of that result would be. 5. Draft a short, plain note telling existing clients about the new price, with a start date. Don't recommend a price just because it sounds premium. Show the math.
Charge for the result, not the hours it takes you.
How to run it
- List your offers with price, hours to deliver (be honest about calls and revisions) and how many you sold in the last 12 months.
- Paste the prompt and the list.
- Look at the math for your top candidate and pick the increase you could say out loud without flinching.
- Use the new price on the next proposal you send. Give existing clients a date.
Cut the offer that eats your week.
Every offer you sell comes with its own way of delivering it: its own templates, steps, tools and follow-ups. Five offers means five ways of working, and all five live in your head.
Cutting one isn't mainly about the revenue you lose. It's about how much simpler everything else gets.
- Operations How different each offer is to deliver.
- Data Revenue, clients and hours for each offer over 12 months.
- AI Compares them side by side and drafts a simpler menu.
Here's everything I sell, including the custom one-offs I keep saying yes to. For each: revenue in the last 12 months, number of clients, hours to deliver, and how different the delivery is from my other offers (its own templates, tools or steps). [paste your list] 1. Which offer earns the least per hour, and which takes the most one-of-a-kind work to deliver? 2. Which offers share most of the same delivery process and could become one? 3. Recommend one offer to cut or pause. Tell me the revenue I'd lose and the hours I'd get back. 4. Show me a simpler menu: three offers or fewer, each with a clear scope. If cutting something would hurt more than it helps, say so.
Fewer offers, fewer processes, more room.
How to run it
- List everything you sell, including the custom work you never meant to offer.
- For each, note revenue, number of clients, hours, and what's unique about delivering it.
- Paste the prompt and the list.
- Pick one offer to pause for the next 90 days and leave it out of new proposals.
Write down the question you answer every week.
Every "quick question" that lands on you is a piece of your business that isn't written down yet. Each one costs a few minutes and a lot of focus, and the same few come back every week.
Don't try to document everything. Document the one that comes back most, and it stops routing through you.
- Operations The question or decision, written as a page someone else can follow.
- AI Finds the repeats in your list, interviews you, and writes the page.
For the last week I kept a list of every time someone (a client, my team, a contractor) needed me for something. [paste the list] 1. Group the list. Which questions or decisions came up most, even when they were worded differently? 2. Pick the one that comes up most and costs me the most focus. 3. Interview me about it, one question at a time, until someone else could handle it without me. When I say "it depends," ask what it depends on. 4. Write it up as one page: When this comes up / What to do / What still comes to me / Why it matters. Keep it short enough that someone would actually read it.
Every repeat question is a missing page.
How to run it
- Start a note on your phone today. Every time someone asks you something, add one line.
- Solo? Log the client questions and the things you keep looking up again.
- After a week, paste the list into the prompt and answer the interview.
- Put the page where people will look, and point them there next time.
Give away the work that doesn't need you.
Most of what routes through you doesn't need your judgment. It needs context that only you happen to have. Those are two different problems. The first is real. The second is a missing page.
Once you can tell them apart, every task has somewhere to go: you, a person with a short how-to, or AI with the right inputs.
- Operations Each task, what it needs, and who could do it.
- AI Sorts the tasks and drafts what each handoff needs.
Here are the recurring tasks from my week, with roughly how many hours each takes a month: [paste from your Move 1 table] Sort each one into: A. Needs my judgment: my experience is truly the input B. Needs context only I have, but not my judgment: write it down once and someone else can do it C. AI can do a first draft if it has the right inputs (notes, templates, past examples) D. A person can do it today with a short how-to For B, C and D, tell me what's missing for the handoff: the page, the template or the inputs. Then pick the one handoff that frees the most hours for the least setup, and write the first draft of what it needs. Push back if I'm calling something "needs me" just because I like it done my way.
If it needs context, not judgment, it can leave your plate.
How to run it
- Copy the recurring tasks and hours from your Move 1 table.
- Paste the prompt and the list. Argue with it where it's wrong.
- Take the one handoff it picks and finish the draft it wrote.
- Hand it off this week. Check the first two results yourself, then let go.
Spend your hours where clients come from.
Marketing comes out of the same hours as delivery. Posting, networking, podcasts, a newsletter: you pay for each one in time, and most service providers can't say which one actually brings the clients who pay.
This move puts one number on each channel: revenue per hour you put in.
- Data Where each client came from, what they paid, and your hours on each channel.
- AI The totals, the per-hour math and a simple way to track it from now on.
Here's where my clients from the last 12 months came from, and what each has paid me so far: [paste: client initials / source / revenue] And here's roughly how many hours a month I put into each channel: [e.g. LinkedIn posts 8h, networking events 6h, podcast guesting 4h, referrals 0h] 1. Total revenue by source. 2. Revenue per hour of my time on each source. 3. Which source brings the clients who stay longest or pay the most? 4. Recommend where to put more time, where to cut back, and what to stop. 5. Suggest one simple way to record the source of every new lead from now on. If a source is marked "unknown," leave it out of the recommendations and tell me how many clients fell there.
Put your hours where your clients actually come from.
How to run it
- List your clients from the last 12 months with where they came from and what they've paid. If you're not sure of a source, write "unknown."
- Estimate your hours a month on each channel.
- Paste the prompt and both lists.
- Add one question to your intake form or call notes: "How did you hear about me?"
Read your numbers in 15 minutes a month.
A lot of service providers find out a month was bad two months later. The numbers exist: in the bank, in Stripe, in a spreadsheet. They're just in five places, and pulling them together happens at 10pm, if it happens at all.
Same prompt, same 15 minutes, the first Monday of every month. Over time, the month-to-month comparison tells you the most.
- Data Last month's money in and out, open invoices, pipeline and hours worked.
- AI Reads it all and gives you the same one-page summary every month.
Here are last month's numbers: [upload or paste: revenue by client or offer, expenses, money in and out of the bank, open invoices, and the deals in my pipeline] I worked about [X] hours last month. Give me a one-page read: 1. Revenue, expenses and what I actually kept, compared with the month before if you have it. 2. Revenue per hour I worked. 3. Money owed to me, and anything overdue. 4. Pipeline: what's likely to close next month, and how that compares with what I need. 5. The three numbers that changed the most, with one plain question about each for me to answer. Facts from my numbers only. If something's missing, list it instead of guessing. Use the same format every month so I can compare.
Same 15 minutes, same questions, every month.
How to run it
- Put a 15-minute block on your calendar for the first Monday of every month.
- Export last month as a CSV from where your money lives (bank, Stripe, QuickBooks), and add your pipeline and hours.
- Run the prompt in your Project, so next month it has something to compare against.
- Answer the three questions it asks. That's the review.
Hire when the math says so, not when you're tired.
Hiring is the biggest capacity move there is, and the one most often made on a bad week. If the work isn't written down, a new person adds a manager's job to your plate instead of taking work off it.
Hire when the math works and the work is ready to hand over. This move checks both.
- Data Your hours, your rates and what the person would cost.
- Operations Whether the work is written down well enough to hand over.
- AI The break-even math and a clear read on whether you're ready.
I'm trying to decide whether it's time to hire someone (employee or contractor). Here's what I know: - Hours I worked last month, and how many I want to work: [X / Y] - The work I'd hand to this person, with hours per month: [paste from your Move 7 list] - What I charge, and how much more work I could sell if I had the hours: [describe] - What this person would likely cost: [hourly rate or salary] 1. Is the work I'd hand over defined well enough for someone to take it on? If not, list what needs writing down first. 2. Compare their cost with the hours I'd free up and what those hours are worth, whether I sell them or use them elsewhere. 3. Break-even: how many extra hours of paid work do I need to sell for this to pay for itself? 4. Recommend one: hire now, start with a small contractor test, or fix the process first. Explain why in three lines. Don't assume I can sell the freed hours. Ask me how confident I am about demand.
Hire for work that's already written down.
How to run it
- Pull your handoff list from Move 7 and your hours from Move 1.
- Get a realistic cost for the role: a contractor's hourly rate or a salary.
- Paste the prompt with your numbers and answer its questions.
- If the answer is "fix the process first," go back to Move 6 for the work you'd hand over.
These moves get much better in Claude Code, with a second brain
In a chat, every move starts from zero. You paste your calendar, your invoices and your offers. Next month you paste them again, and the chat has forgotten what you decided.
In Claude Code, connected to a second brain, it doesn't:
- Your business is already there. What you sell, who your clients are, how you work and what you decided last month live in files Claude reads every time.
- Your numbers live in one place. Move 9 runs on this month's real numbers in seconds, not a copy-paste.
- Every move saves its result. Move 3 builds on Move 1, and next month builds on this month.
- The prompts become routines. The same moves run on their own, on a schedule, and send you the answer.
That takes two things most service providers haven't built yet: a second brain that holds how your business runs, and data you can actually collect, read and decide with.
That's exactly what the Second Brain Workshop Series builds, live:
- Mon Oct 19 · Deep Dive: Making Decisions with Data. How to collect, read and decide with your numbers.
- Thu Oct 22 · Technical Workshop: Data Visualization. Build a live dashboard of your own numbers.
- Mon Oct 26 · The Second Brain Workshop. Get your business out of your head so Claude can work from it.
For today, start with Move 1.
Puja
All 10 prompts in one place
The same prompts as above, in order. Copy one, or copy them all into a note.