One Small Door at a Time: My December of Tiny Fixes
There's a specific sort of madness that arrives around the second week of November. The nights draw in, the heating clicks on, the supermarket starts playing the same three songs on a loop, and I get a very sincere urge to relaunch my entire website before Christmas.
I have done this four years running. I have never once succeeded. What actually happens is I spend a fortnight sketching a grand new information architecture in a notebook, then December arrives and I've done nothing except rename a folder and feel guilty about it.
So this year the plan is different. Twenty-four doors. One small job behind each one. Nothing more.
The rules, such as they are
I've had to write this down, because otherwise I'll cheat within about three days and try to sneak a full redesign into door eleven.
- One job per day, and it has to be finishable in a single sitting. If it needs research, it doesn't qualify.
- No job may depend on another job. The whole point is that I can skip a day without the rest collapsing.
- It has to be a visible improvement to someone other than me. Tidying a CSS variable nobody can see doesn't count.
- If I miss a day, I miss a day. No doubling up. December is not a punishment.
- The list is written in advance and I'm not allowed to swap things in. This is the rule that stops me spending the 14th rebuilding the contact form.
That's it. Five rules and a paper list stuck to the side of the monitor with the corner curling up.
What 'tiny' actually means
This is the bit people get wrong, including me. Tiny doesn't mean pointless. Tiny means bounded. There's a difference between a small job and a job that just hasn't grown up yet.
A tiny job has a clear finish line. Rewriting a single image caption has one. 'Improving accessibility' does not, and if I put that on the list I'll be ankle-deep in ARIA attributes on the 22nd, weeping into a mince pie.
So the entries on my list look more like this:
- Rewrite the alt text on the six photos in the most recent post.
- Fix the dead link to the thing I mentioned in March.
- Reorder the projects page so the finished ones sit above the abandoned ones.
- Cut one paragraph from the About page that I wrote in 2023 and no longer believe.
- Add a last-updated line to the colophon.
- Change the footer copyright to say 2026 properly.
- Check the site on a phone, on purpose, for ten minutes.
None of that is exciting. All of it is the difference between a site that feels looked-after and one that feels vaguely abandoned. Visitors can't articulate the difference, but they feel it. A site with three stale dates on it reads like a shop with the shutters half down.
Why not just do the big relaunch?
Because the big relaunch is a lie I tell myself to avoid doing the small things.
There's a satisfaction to the fantasy: one weekend, a fresh coat of paint, everything fixed at once. But the fantasy has a price. It means the site sits untouched for six weeks while I plan it, and then it sits untouched for another six weeks because the relaunch turned out to need a content audit, and a content audit needs decisions, and decisions need a version of me who isn't currently wrapping presents.
Daily maintenance is unglamorous in a way that's genuinely better. Each job takes ten to forty minutes. The list is finite. By Boxing Day the site looks noticeably tidier and I haven't once had to draw a wireframe.
It also suits how I work, which is in scraps. Twenty minutes between meetings. A train journey. The bit of the evening after the dinner things are cleared but before the good telly starts. A big project needs a big block of time I don't reliably have. A small job fits in the gaps that already exist.
The bit where it goes wrong
Let me be honest before this reads too much like a self-help post.
Some of the doors will be rubbish. Day nine will be 'fix a typo', which is a genuine improvement but not a satisfying way to spend an evening. Day nineteen will be something I've forgotten how to do and will have to look up, which turns a fifteen-minute job into a fifty-minute one and eats the telly.
And there's a real risk that a tiny job quietly becomes a medium job. You go to fix one link and discover the whole page it points at has been redirecting for eight months. You go to rewrite a caption and end up deleting the post. The rules say I'm allowed to stop and put it back on next year's list. I suspect I'll need reminding of that.
The other trap is treating the calendar as a content commitment. It isn't. I'm not writing twenty-four posts, I'm not filming anything, and nobody has subscribed to this. It's a maintenance schedule with a paper chain attached. If I fall behind by a week, the internet will cope.
Why the calendar shape helps
There's something about numbered doors that does the motivational work for you. A to-do list is a nag. A calendar is a ritual. Twenty-four boxes on a bit of card by the kettle, and the day's box gets ticked off — that's a small, repeatable pleasure in a month that otherwise consists of fighting for a parking space at the retail park.
It also makes progress visible in a way that a git history doesn't. I can look at twenty-two ticked boxes on Christmas Eve and know exactly what happened. That's worth more than a vague sense that I've been 'working on the site'.
What I'd suggest if you're tempted
Write your list before December starts, and be brutal about the wording. Every item should be something you could describe to a friend in one sentence, and they'd know when it was done.
Then put the list somewhere physical, not in a task app you open twice a week. Mine's on the back of an old envelope. It's not elegant, but I can't ignore an envelope taped to a monitor.
And leave Christmas Day blank on purpose. Twenty-three doors is fine. Nobody's marking it.
If you want to follow how mine goes, I'll be nudging the odd update onto the site as I tick things off — mostly boring ones, occasionally something worth a whole post. Which, I suppose, is the point. Small doors, opened in order, adding up to something that looks suspiciously like a better website.
