Introduction — The Year I Had to Build Twice
Two hundred and fourteen files, a hundred and sixty hours, and a carry-forward rate of four per cent.
Jem built two hundred and fourteen listings between August and November.
Twelve-month wall calendars, monthly planners, weekly spreads, an academic-year set, a family organiser with columns for five people. Four page sizes. Two week-start options. Three colourways. It came to two hundred and fourteen files, and every one of them was laid out by hand, one at a time, because that is how you make things when you are making the first one.
About a hundred and sixty hours across four months. Evenings and most of two weekends in October.
It sold. Not spectacularly, but properly — October was better than September, November was better than October, December was the best month of all. There were reviews. Somebody said it was exactly what they had been looking for. Jem spent Christmas feeling like the thing had worked.
In January, sales fell by about a third. That was expected; people buy calendars before the year, not during it.
In February they fell to almost nothing. That was also, in retrospect, expected.
⚠ What was not expected was March, when Jem sat down to work out what to do next and realised that every single one of the two hundred and fourteen files was now worthless, and would have to be made again from scratch.
What was actually wrong
Nothing was wrong with the files. They were fine. They were correct, they printed well, buyers had liked them.
They were simply about a year that had begun.
⚠ A calendar is the only digital product that comes with an expiry date printed on the front of it. Not a slow decay, not a gradual staleness — a specific day on which the entire catalogue becomes unsellable, all at once, in public, where every buyer can see the year in the title.
That is the first thing, and most sellers know it going in.
⚠ The second thing is the one that costs money, and almost nobody sees it coming: Jem had built finished files rather than sources. There was nothing to regenerate. The dates were part of the layout — typed, positioned, aligned by eye, one month at a time, two hundred and fourteen times.
To sell in the next season, all of it had to be built again. Not adjusted. Built.
Jem's carry-forward rate — the share of the catalogue that could be produced for the following year by regenerating from something that already existed — was about four per cent. Three undated planning sheets and a habit tracker.
⚠ Everything else was a hundred and fifty hours of work that had bought exactly one selling season.
The second year
Jem did it again. August to November, about a hundred and fifty hours, largely the same designs with different dates, laid out by hand for the second time.
It sold slightly better, because there were reviews and a few people came back. And in the following March, Jem was in exactly the same position — with a shop full of dead files and a summer of work ahead.
⚠ Two years, three hundred and ten hours, and a business that started from nothing every August.
The third year was different, and the difference was not design skill or marketing. It was one structural decision.
What the third year looked like
Jem stopped building calendars and started building sources.
A source is a master file in which the dates are not drawn but generated: a grid that takes a year and a week-start convention and produces the right days in the right boxes. From one source you can produce every month, every size, every colourway and every year — because the only thing that changes between them is an input.
Nine sources. Wall calendar, monthly planner, weekly planner, academic year, family organiser, desk pad, undated set, habit tracker, and a year-at-a-glance.
From those nine, two hundred and forty listings.
⚠ The rebuild for the following year took about twenty-two hours.
Not a hundred and fifty. Twenty-two — and most of that was checking rather than making, because the making was regeneration.
Carry-forward rate: about ninety-four per cent.
What generation did to the errors
There is a second half to this, and it is the honest part.
In year one, hand-built, eleven of the two hundred and fourteen files had errors. All different: a month with the wrong number of days in one size but not the others, a weekday column misaligned on a single page, a year label left over from a duplicated file. They surfaced one at a time over five months, each reported by one buyer, each fixed individually.
⚠ In year three, generated, Jem made one mistake — a week-start setting wrong on one source — and it appeared identically in thirty-four listings on the day they were published.
That sounds worse and it is not.
⚠ One buyer reported it. Jem changed one value in one file, regenerated, and re-uploaded thirty-four corrected listings in about twenty minutes.
⚠ Eleven different errors found over five months and fixed eleven times is a worse year than one error found in a week and fixed once. Generation concentrates your risk and collapses your correction cost, and that trade is overwhelmingly worth taking.
But it is a trade, and it is why Rule 3 exists: once you generate, you are never making one mistake. You are making the same mistake everywhere, at once, and the only defence is checking the source rather than the output.
Revenue per maintained source
The arithmetic that follows is the second thing this book is organised around.
Listings are not the asset. Two hundred and forty listings that must be hand-rebuilt every August is not a business; it is a summer job you have given yourself in perpetuity.
⚠ The asset is the source — because a source is the thing that survives the first of January.
So the unit is revenue per maintained source: net revenue divided by the number of sources you actually keep current and can regenerate from. Jem's first year had a denominator of essentially zero. The third year had nine, and earned more.
⚠ And the leading indicator is the carry-forward rate: the share of next year's catalogue you can produce by regenerating rather than rebuilding. You can compute it today, before the season, and it tells you exactly how much of next August you have already lost.
What this book is
Thirty-seven chapters, sixty-eight resources and a hundred and forty-eight prompts, built on four rules:
Rule 1 — It has a death date printed on it. The whole catalogue dies on a known day, together. Chapter 5.
Rule 2 — The season closes. Roughly a third of the year carries almost all the revenue, and outside it there is no demand rather than less demand. Chapter 6.
Rule 3 — Every error is a catalogue error. Variants come from sources, so a mistake is never in one file. Chapter 7.
Rule 4 — It lives on a wall for a year. It becomes physical on a printer you have never seen, then hangs somewhere visible while people write on it. Chapter 8.
What this book will not do
It contains no dates. No public holidays, no term dates, no week-numbering rules, no statement about which day any date falls on in any year. Not because those are hard to find, but because a calendar seller who takes a date from a book — or from a language model — instead of from a source they have verified is exactly the seller this book is trying to prevent. Chapter 21 is how you check; Chapter 37 is why you must never delegate it.
It will not tell you which sizes to sell, which marketplace to use, or what to charge. And it will not tell you that two hundred and fourteen listings is a better business than nine sources.
⚠ What it will do is show you how to build a catalogue that survives the first of January — and how to know, in March, exactly how much of next year you already have.
Jem's summary, given some years later:
⚠ "I spent two summers making the same calendars twice. The third summer took three weeks, and the only thing that changed was that I stopped drawing the dates and started generating them. Nobody told me that was the whole business."
©2026 James Henderson / https://localhandyman.work