Tuesday, September 1st, 2026

Labor Day isn't until next Monday, which means you're sitting in the strangest week of the business year. September has technically started. Nobody's acting like it yet. Half the people you need answers from are still out, the inbox is unusually quiet, and there's this shared agreement that the real work resumes on the 8th.

Which gives you about six days of cover to do something almost nobody does, and it's the reason I'm writing this today instead of next week.

Here's what set it off. Sunday night I pulled up the doc where I keep the running list of what I'm building this year. Fourteen items. I read them top to bottom, and somewhere around item nine I got that specific cold feeling you get when you recognize your own handwriting from a while back.

Six of those fourteen had been on the list in January. Not partially done. Not blocked on something outside my control. Just sitting there, quietly re-approved every single time I looked at the list, because looking at a list and deciding it's still a good list feels enough like planning that you can do it for eight months without noticing.

So I did the arithmetic I'd been avoiding, and the arithmetic is the whole point of today's letter.

From today to December 31st is one hundred and twenty one days. Call it seventeen weeks.

That number by itself is useless. Seventeen weeks sounds like a lot. It's the kind of number that makes you feel like you've got room. So let's take it apart, because the useful version is a lot smaller and a lot more honest.

Out of those seventeen weeks, you're going to lose real time you already know about. Thanksgiving week is functionally gone for most businesses. The last two weeks of December are gone. Depending on your world, the week of Labor Day is soft. That's roughly four weeks off the top before you've made a single choice, and you didn't choose any of it.

Thirteen weeks left.

Now the part people flinch at. Out of a working week, how many hours do you actually spend building things that didn't exist before, as opposed to delivering, servicing, responding, and administering? Be honest. For most people running a small operation, the real number for creation work is somewhere between six and twelve hours a week, and the ones who insist it's twenty five have never measured it.

Take the middle. Nine hours a week, thirteen weeks. That's about one hundred and seventeen hours of genuine building time between now and the end of the year.

One hundred and seventeen hours. Roughly three working weeks, spread thin across four months.

Now go look at your list of fourteen things and tell me with a straight face that all of them fit.

This is the entire problem. We plan in items and we execute in hours, and we never convert between the two. A list of fourteen things looks like ambition. Converted into hours it's a fantasy, and the fantasy has a cost, because every item on a list you can't finish is quietly taxing your attention every time you read it.

Almost every planning method you've been handed starts with what you want and works backward. Set the goal, reverse engineer the milestones, break it into tasks, schedule the tasks. It sounds rigorous. It fails for one reason: nowhere in that sequence does anything check whether the total will fit in the container.

Flip the order. Measure the container first, then decide what goes in it. You'd never pack for a trip by listing everything you want and then looking for a suitcase. You look at the bag, then you choose.

Here's the whole system. Four passes. Give it ninety minutes this week while everyone else is still on the beach.

Don't estimate. Estimating is how you got here.

Go back through the last three weeks in your calendar and your inbox and put every hour into one of four buckets. Delivery, which is work a client is paying for right now. Admin, which is invoicing, email, scheduling, tools, taxes, the whole tax of existing. Sales, which is anything aimed at getting the next dollar. Build, which is the only bucket where new things come from.

Three weeks is enough to get a real read and short enough that you'll actually do it. Add up the Build column and divide by three. That's your weekly building capacity, and I promise it's lower than the number in your head.

If you want this measured rather than reconstructed, this is the one thing I'd automate. I run Rize in the background specifically because my self reported hours and my actual hours were off by a factor I'm not going to print. The tool isn't magic. It just removes the part where you get to be generous with yourself.

Then subtract the weeks you already know are gone. Write the final number down somewhere you'll see it. Mine's on a sticky note on the monitor. It's an unpleasant number and that's the job it's doing.

Take your list. Every item goes into exactly one of three lanes. No item goes in two. If you can't decide, it goes in the third one, and you'll understand why in a second.

Keep. Things already running that produce something and require maintenance. Client work. The newsletter. The thing that pays. Keep items don't get budgeted out of your Build hours because they're not building, they're operating. But they do get named, because unnamed maintenance is how people end the year convinced they did nothing.

Build. New things that don't exist yet. Here's the rule that makes this work and the one you'll want to argue with. One Build item at a time. Not one per month. One, until it's done or dead.

I know. You've got three that all feel urgent. Run the math from Pass One. If your real capacity is nine hours a week, and a meaningful new thing takes forty to eighty hours to get to a state where it's actually live and not just started, then you have room for one, maybe two, before the year ends. Three parallel Builds at nine hours a week means three things at thirty percent completion on December 31st, which is worth exactly nothing. Sequence beats parallel every time when capacity is the binding constraint, and capacity is always the binding constraint for people like us.

Kill. Everything else. And yes, everything else, because if it didn't make Keep and it isn't the one Build, it's Kill by definition. Not "later." Not "Q1." Kill has a specific meaning here, which we'll get to.

This is the one that gets skipped and it's the one that does most of the work.

Before you add anything to the next seventeen weeks, you remove something. Not because minimalism is a virtue. Because you already established you don't have room, and the only honest way to make room is to take something out.

Three places to look, in this order.

Subscriptions and tools. September is renewal season for a lot of annual plans, which makes this the cheapest month of the year to audit. Pull your card statement for the last three months and list every recurring charge. For each one, answer a single question: if this vanished tonight, what specifically breaks tomorrow? If you can't name the thing that breaks, cancel it. Most people find between eighty and three hundred dollars a month of stuff they'd genuinely forgotten they were paying for, and more importantly, each one of those tools was carrying a little bit of your attention.

Recurring meetings and commitments. Look at every repeating block on your calendar for the last month. For each, ask what decision it produces. Not what gets discussed. What gets decided. A recurring meeting that doesn't produce decisions is a status update wearing a meeting's clothes, and status updates belong in writing. Kill it or convert it.

Half-finished things. This is the painful one. Every project sitting at forty percent is occupying real estate in your head and returning nothing. Go through them one at a time and give each a verdict: finish it in the next seventeen weeks, or formally kill it. Not shelve. Kill.

Here's what killing actually means, because "shelving" is a lie we tell ourselves. Write down what you learned, delete or archive the working files, remove it from every list, and tell anyone who's expecting it. If it turns out to matter next year, you'll rebuild it faster with what you learned. What you're really buying is the mental line item, and you can't buy that back by putting something in a folder called Someday. Someday folders are just lists with better branding.

Now you protect the thing you just built, because the plan isn't what kills you. The next four months of small yeses are what kills you.

Make a running ledger. Two columns. Left side, hours available, which is the number from Pass One. Right side, hours committed. Every time you say yes to anything that isn't Keep or the one Build, you write it in the right column with an honest hour estimate.

The rule is simple and it's the only rule: you don't commit past the line.

What this does isn't discipline. It's translation. Right now when someone asks for a favor, a call, a quick review, a podcast appearance, the question in your head is "do I want to do this," and the answer is usually some version of yes because you're a decent person and it sounds interesting. The ledger changes the question to "what am I removing to make room for this," and that's a completely different question with a much better hit rate.

Two practical notes. Estimate in hours, not minutes, and round up. A "quick call" is never a quick call, it's a call plus the prep plus the follow up plus the twenty minutes on either side where you're not usefully doing anything else. And review the ledger at a fixed time weekly, not when you feel like it, because a ledger you look at when you're already worried is a ledger that only measures your anxiety.

I keep the capture end of this automated because the capture end is where every system I've built has died. Commitments get made out loud, in calls and in messages, and if logging them depends on me remembering to go type something afterward, they don't get logged. I run Fathom on calls so the record exists whether or not I'm sharp that day, and a Make.com scenario that watches for anything I've tagged as a commitment and drops it into the ledger with an estimate attached. That took one afternoon to build and it's the reason this one survived past week three, unlike the four versions before it.

For the thinking part, I'll write the plan out in my own words first and then hand it over and ask it to argue that the plan doesn't fit, to name what I'm assuming about my own speed, and to tell me which item I'm keeping for emotional reasons rather than business ones. I run that through Galaxy.ai because I like being able to bounce the same question off a couple of different models without paying for four subscriptions. I create, it critiques, I refine. Never the reverse. The second it's doing the deciding you've handed over the exact thing you were trying to strengthen.

I built the worksheets for all four passes into one document because I needed them for myself and it seemed stupid to keep it.

It's called The Seventeen Week Close Kit. It's a working doc, not a lead magnet with a video series behind it. Inside you'll find the Honest Hours worksheet with the four buckets and the math laid out, the Three Lanes sorting sheet, the Subtraction Pass checklist with the tool audit and the meeting audit and the kill list, a blank Commitment Ledger, and a one page seventeen week map you can print and put on the wall.

No cost, no upsell attached to it, no sequence waiting for you. Make a copy and use it.

Get The Seventeen Week Close Kit here

If you go through it and land somewhere ugly, which happens, hit reply and tell me what the number came out to. I read all of those. And if it turns out the problem is bigger than a worksheet, that's what Pinnacle Masters exists for, and we can talk about what a working session looks like.

You don't have to run all four passes today. Run Pass One.

Go back through three weeks of calendar, sort the hours into the four buckets, add up the Build column, divide by three, and subtract the weeks you already know are gone. Write the number on something you'll look at.

That's maybe forty minutes and it will do more for the rest of your year than any amount of goal setting, because everything downstream of it becomes a real decision instead of a wish. You can't choose between fourteen things. You can absolutely choose one thing when you're staring at a number that tells you one is what you've got.

Back to my list. I killed nine of the fourteen. Two of them I'd been carrying since January and had genuinely convinced myself were nearly done, which is a thing your brain will do for you if you never open the file. Killing them took eleven minutes and I felt lighter walking out of the office than I have in a month.

The other thing I noticed, and this is the part I didn't expect: the one Build I kept got noticeably better within a day. Not because I had more hours, though I did. Because it stopped competing. When something is the only new thing you're building, you think about it in the shower, on the walk with the dog, in the gap between calls. Attention consolidates when you stop splitting it. That compounding is invisible on a spreadsheet and it's most of the actual return.

Seventeen weeks. Everybody else starts theirs on the 8th.

You've got a week's head start and a calendar nobody's fighting you for. Go count your hours.

One step, one day. Grace over guilt. — Dan Kaufman

———————————————————————————

Tools I use and trust. If you sign up through these links I may earn a small commission at no extra cost to you. I only recommend things I actually use.

Make.com — Automate your workflows without writing code

Fathom.video — AI meeting notes that actually capture what matters

Galaxy.ai — All your AI tools in one place

Rize.io — Time tracking that makes you honest about your hours

Beehiiv — Where serious newsletter operators publish