Not the week I planned

I had last week mapped out. A PHPOffice spike on one project, a tricky permissions refactor on another, kickoff momentum on two new builds. By Monday morning the plan was irrelevant — a family member was admitted to hospital, and the laptop came with me.

I'm writing this partly as a note to myself, partly because I think the freelance world doesn't talk honestly enough about what these weeks actually look like. The deliverables shipped. The week was hard. Both of those things are true.

What the days actually looked like

Hospital weeks have their own rhythm. Doctors do rounds in the morning. There are long flat stretches of waiting between updates, then short bursts of activity when something needs deciding. You sleep when you can, which is rarely when you'd choose to.

Inside that, the work fit where it could:

  • Early mornings, before rounds. A clear hour or two for the deep stuff — the kind of thinking that needs uninterrupted runtime. The permissions refactor lived here. Modelling roles, policies, and the abstractions between them isn't something you can do in fifteen-minute slices.
  • Mid-morning to early afternoon. Patchier. Good for the PHPOffice work, which was mostly mechanical — reading spreadsheets, mapping cells, writing the conversion layer. You can pick that up and put it down without losing your place.
  • Evenings, after visiting hours. Admin, code review, the small follow-ups that pile up if you don't clear them daily. Email triage. The stuff that doesn't require thinking, just doing.

The discipline that actually mattered wasn't time management. It was honesty about what kind of work could fit in what kind of slot. Trying to do the permissions refactor at 3pm with a beeping monitor next door would have wasted the afternoon and produced code I'd throw away on Friday.

What I told clients

Less than I expected to need to.

I sent two short messages on Monday — to the two engagements with deliverables due that week — saying I had a family member in hospital, that I'd be working but slower, and that I'd flag anything I couldn't hit by Friday. Both replies were warm and unworried. I kept them updated in the existing channels, on the existing cadence. I didn't dramatise it, and I didn't disappear.

The thing I noticed: when you've built a track record of shipping, one rough week doesn't dent the trust. The relationships I was worried about turned out to be the ones that handled it best. The reflex to over-explain or apologise is mostly about your own anxiety, not the client's actual need.

What surprised me

A few things stood out by Friday afternoon.

The deep work got done first. Counter-intuitively, the most demanding piece — the permissions refactor — was the one I finished cleanest. Because it required real focus, I protected the early-morning window for it ruthlessly. The work I thought would be "easier to slot in" was the work that fragmented and dragged.

The tooling held up. Strict types, DTOs, named branches, decent commit messages — all the small disciplines I usually justify as "good practice" are actually fatigue insurance. When you sit down at 6am running on three hours of sleep, you're not reasoning, you're pattern-matching. The patterns need to be good ones, because the reasoning isn't going to save you.

The week wasn't a write-off. It would have been easy to declare it lost on Tuesday and pick up Monday. But the work that did get done was real work — not heroic, not impressive, just steady. Most of what we lose to hard weeks isn't the time. It's the self-defeating decision that the week is already gone.

What I'd tell past-me

Three things, if I could send them back to Monday morning:

Send the client messages immediately. Don't wait until you "know how bad it is" — you don't, and they don't need that detail anyway. Short, factual, no promises you might not keep. Then go.

Pick one thing to ship, not three. I tried to keep everything alive. I should have explicitly demoted two things and protected one. Everything got slower; one thing could have been on time.

Trust the systems you built when you were fresh. This is the real argument for the boring stuff — the linters, the type checks, the templates, the deploy scripts. Past-you, well-rested and thinking clearly, built the rails. Tired-you just has to stay on them.

Wrapping up

I'm back at my desk this week. Things are settling. The work that didn't get done last week is getting done now, and the work that did get done held up under review. That's about all you can ask from a week like that.

If you're in one of those weeks yourself — the practical stuff matters less than people pretend, and the relational stuff matters more. Tell the people who need to know. Pick one thing. Trust your past self's systems. Go.