Fifteen-hour days. Sometimes until 2am, because there was a deadline, because someone was waiting, because falling behind wasn’t an option.
And still behind.
That’s the part that never made sense to me. I wasn’t coasting. I was putting in more hours than anyone I knew, and it played out the same way every week — two steps forward, one step back. The effort was real. The deficit never closed.
I want to take that apart properly, because for years I read it as a personal failing, and it isn’t one. It’s structural. There is a whole category of work in a solo business that has nowhere to live except the hours you’re supposed to be asleep in. Once you can see why, you stop trying to fix it with a better morning routine.
Every hour of the day already belongs to someone
Look at how a day actually gets allocated. Not how you plan it — how it gets decided.
The client call happens because a client is on the other end of it. The deliverable gets done because someone is expecting it on a date. The contractor’s question gets answered because they’re standing still until you answer. School pickup happens because a child is waiting outside a building.
Every one of those has a claimant. A person who notices if you don’t show up. That’s the scheduling mechanism in a solo business — not priority, not planning. Who’s waiting.
Now think about the other layer. The reconciliation. The file that needs a name someone can find later. The invoice you’re chasing for the third time. The process you’ve explained out loud eleven times and never written down. The tracker that’s three weeks stale.
Nobody is waiting for any of it. No client emails to ask whether you’ve organised your folders. Nothing external enforces it. It has no claimant at all.
So it loses every scheduling contest it enters. Not occasionally — every time, all day, for years.
Deferrable is not the same as optional
Here’s the confusion that keeps this running.
Every task in your week has two separate properties, and most people collapse them into one. How important it is. And how loudly it complains when you skip it.
A client deliverable scores high on both. It gets an hour.
The admin layer scores high on the first and silent on the second. It gets nothing.
And because it’s silent, skipping it today costs you nothing visible. So you skip it. Then tomorrow, at the same zero visible cost. The real cost is still accruing — it just arrives later and it arrives all at once. The catch-up. The audit. The file you can’t find on the one day it matters. The client who never complained and quietly stopped feeling looked after.
That’s the whole mechanism. Deferrable work migrates toward the end of the day and then straight past it.
Nobody chooses 11pm. You arrive at it. 11pm is simply where the ownerless work ended up after losing every scheduling contest since eight in the morning.
And it lands there because those are the only unclaimed hours in the week. Nobody else has a bid on 9pm to 2am. It isn’t focus time and it isn’t a night-owl personality — it’s the last inventory left, and this is the only work that will accept a slot nobody else wants.
The arithmetic nobody does
Now the part that explains why the hours didn’t fix it.
This work arrives at a rate. Your business produces admin in proportion to its activity — every client generates paperwork, every invoice generates a reconciliation, every project generates files, every campaign generates follow-up. Grow the business and the pile grows with it. That isn’t a flaw in how you run things. That is what growth is.
You clear it at a rate too. And your rate has a hard ceiling, because it’s one person, and a day has a fixed number of hours in it, and some of them have to be sleep.
If the arrival rate is higher than the clearance rate, the pile grows. How hard you work doesn’t change the direction. It only changes the slope.
That’s the trap. Adding hours moves your clearance rate up a little. Adding clients moves your arrival rate up a lot. So the better the business does, the further behind you get — which is the most disorienting thing about running eight operations, and almost nobody says it out loud.
Fifteen hours instead of ten is maybe fifty percent more throughput, thrown at a pile that’s growing with the business. Two steps forward. One step back. Indefinitely, in principle, because the arithmetic doesn’t care how tired you are or how much you want it.
That’s an arithmetic problem. Not a willpower problem. If you take one thing from this, take that.
Why time management can’t reach it
Every productivity system I tried assumed my day had slack in it.
Prioritise, and the low-value work falls off the bottom. Batch it. Time-block it. Do the worst thing first.
Those all work when the problem is order. They cannot work when the problem is volume, because none of this work is actually able to fall off the bottom. It isn’t optional. Prioritising it doesn’t delete it — it relocates it. You do the important thing first and at 6pm the rest is still sitting there, still needing doing, still with nobody else on the planet assigned to it.
Time-blocking is the fix everyone recommends. Put admin in the calendar, two to four. Protect it.
It’s the first block that yields, every time, because it’s the only appointment in your week where nobody notices you didn’t show up. A client calls at two. You take the call, because the call is the reason you have a business. The block moves to tomorrow, and tomorrow it moves again.
There are only two real levers. Reduce the arrival rate, or add clearance capacity that isn’t you. Reducing the arrival rate means running a smaller business, and that’s not what any of us are trying to do.
The category needed an owner
The mistake I made for fifteen months was treating this as a personal efficiency problem. Better tools, better system, better me. Learn faster, work smarter, get up earlier.
What it actually was: an entire function of the business with nobody assigned to it.
In a company with staff, this work has a name and a desk. Ops. Admin. The person who keeps the machinery running. Nobody expects the CEO to name the files. Solo, there is no such person, so it defaults to the only one available, in the only hours available.
I didn’t need to be faster. Faster was never going to close a gap that arithmetic was widening. I needed the function to have an owner who wasn’t me.
That’s what an operator is, and why I’m careful about the word. Not a tool I open when I have a spare minute — a tool makes me faster, and faster wasn’t the problem. An owner. Something that holds a whole function, keeps its own memory of how that function works in my business, and does the work whether or not I’m awake for it.
Ellestra runs that team as my AI COO. Atlas owns the admin layer — the files, the naming, the finding, the pile that used to be my 11pm. Quill holds content. Sage handles research. Charter turns processes I’d only ever explained out loud into maps. Pixel builds the visuals off an outline instead of me making shapes at midnight.
None of them are faster versions of me. They’re different clearance capacity. That’s the only thing that actually changes the arithmetic.
Mine was the books
Not at a desk, either. In bed.
Laptop open, propped up on pillows, doing the books to get caught up — and it was never the routine part that kept me there. Entering things is fine. Entering things is almost restful. What kept me up was finding the problems: a receipt that didn’t match anything, a reconciliation that wouldn’t close and wouldn’t tell me why. Investigative work, at midnight, which is the single worst hour to go looking for a discrepancy.
And I remember thinking, in complete sentences: I do not want to be sitting in bed doing admin. This is not what I want to be doing. I should be here reading a book, or listening to a podcast, or doing absolutely nothing before I sleep. Not working on receipts.
That’s the part that stayed with me. Not the hours — the location. Bed was the last place in the house that wasn’t work, and eventually the work took that too.
The night shift still exists
Here’s the part I find genuinely funny now.
That work still happens at night. The morning note sitting there when I sit down with coffee — what came in, what’s flagged, what needs a decision from me, what’s already handled — was written while I was asleep.
The night shift didn’t disappear. It just isn’t mine anymore.
And the “still behind” feeling went with it. Not because I got faster, and not because I found more hours. Because the arrival rate finally has something meeting it that isn’t one tired person at 11pm with the last of her judgment.
If you want to see what that did to the shape of an ordinary day, hour by hour, that’s the Wednesdays post. This one is about why the night shift existed in the first place.
What to do next
If you’re reading this at 11pm — and statistically, some of you are — the thing I’d want you to take away is that the hour is not a character trait. It’s a destination. It’s where ownerless work lands after being outbid all day by work with a person attached.
So pick the one function that reliably ends up in your night hours. Files and admin, for most people. Content for some. Research for others.
Give that one away first. Properly — to something that holds the whole function and remembers how it works, not to a tool you open when you have a spare minute, because that’s just you again with a faster keyboard.
The AI COO Setup Kit is the setup I’d start from: the dedicated machine, the memory that persists overnight, the identity file that teaches it how your business actually runs, and the first operator to hand a function to. It’s the fifteen months I spent going in circles, compressed into something you can do in a day.
And if you want the number before you commit to anything, there’s a calculator on the automate page. Put in what your week actually looks like — the email, the invoices, the content — and it’ll show you the hours you’d get back. I’m not going to quote you mine, because yours won’t be mine. That’s exactly why it’s worth working out rather than being told.
Then find out what your evenings were supposed to be for.