A Real Person Having a Bad Day
A design education teaches you to look at the page. A master's in industrial-organizational psychology teaches you to look at the person reading it. I only ever had the second one, and what it produces is a design tested against a real person having a bad day — not a persona, one person, in one room, at one hour, in the condition they are actually in when they reach the thing I made.
Built from Ten sites, platforms and portals built in eleven months with no formal design training, the four ship tests carried across them, and a written decision record kept for each. This is an argument about method rather than a study of one. No outcome was tracked and nothing was measured against a control. The systems themselves are not named here. Some belong to organisations I cannot name on this site, so you cannot open one and check that the rule I describe is the rule that shipped. That is the weakest thing about this piece and it does not have a fix.
Does not establish That design training is unnecessary, or that this route is better than the one through a school. It is not a recommendation — I took it under conditions nobody would choose, and I do not know what I would have made with four years of instruction behind me. Nor that designing this way produces better outcomes: no reader was tested, and I cannot tell you the tinted ground reduced anyone's strain. What the method produces is not a proven result. It is a decision with a reason attached, which can be argued with.

In September 2025 I had never designed anything.
That month took my father and it took my job, and I am not going to tell you about either here. What matters for this piece is what the calendar looked like afterwards: I had time, no employer, and no plan that survived contact with either fact.
Four things were available. I want to put all four down, because three of them were reasonable and I turned them down anyway, and a decision you can only see the winner of is not a decision you can check.
What I turned down
Another disability-management role. The work was right. The structure around it was not, and I had just watched what that structure does to the people inside it. Taking the same job at a different address would have meant betting that the structure was local. It is not.
Waiting until I felt ready. Ready was not coming. I had the time, and time was the only thing I had a surplus of — which is a strange sentence to write about the worst months of your life, and it was still true.
Learning design properly first. This is the one that would have looked sensible from the outside, and it is the one I am most certain about. A course would have taught me to look at the page. I already knew how to look at the person reading it.
What I took
Build, knowing nothing.
I had just finished a master’s in industrial-organizational psychology. That degree is not about layout. It is about what happens in a person when a system acts on them — what they notice, what they can hold, what they do when they are tired and frightened and the thing in front of them is the only thing standing between them and an answer.
So I designed the only way I knew how, which was to ask what the person on the other side would feel.
That is not a philosophy I arrived at. It was the only question I had.
Why that question is different from the usual one
Most design decisions are defended with a reason that cannot lose. It feels cleaner. It is more modern. It breathes. None of those can be wrong, which means none of them can be checked, which means they are not reasons.
Starting from a person changes what kind of claim you are making. The person is specific, they are not having a good day, and the question stops being what looks right and becomes what happens to them — which has answers you can get wrong.
Here are four of them.
Eleven at night
An exhausted director of human resources, reading on a laptop in a dim room, because that is when the reading actually gets done.
The page ground is tinted rather than white. White throws glare at that hour into eyes that are already done for the day. No grey text on grey ground. No fourteen-pixel body copy. No tight leading.
Every one of those is a normal, defensible, contemporary choice that I am refusing, and the thing doing the refusing is not taste. It is a nervous system at eleven at night.
A reader in distress
Somebody who is not reading so much as scanning for the one line that applies to them.
Headlines wrap on text-wrap: balance, so the break still lands somewhere that
makes sense to a person taking in half of it. And the crisis affordance loads
before the hero — the page is ordered by what someone might need first, not by
what looks best first.
That second one costs something visible. The most composed moment on the page no longer arrives first. I have never been able to argue myself out of it.
The one place red is allowed
One red, reserved across an entire system, for emergency affordances only. Never a brand colour. Never decorative. Never a sale.
Which means the worst moment a reader could have gets a slot in the palette before any of the good ones do, and every cheerful use of that red — the urgency badge, the accent rule, the hover state — is spent scarcity. A signal that appears everywhere has already stopped being a signal.
The greyscale printer
Every surface has to survive being photocopied.
This is not a nostalgia test. It is how documents actually travel inside an organisation: printed, scanned, forwarded, printed again by someone three departments away who was never sent the link. By the time a document reaches the person it is about, it is very often grey.
If colour is carrying meaning, the system has already failed. The photocopier is just where you find out.
A test has to be able to fail
None of those four is about the page. Each is a named person in a named condition, and each produces a rule that a mark can fail.
That is the difference between a system and a look. Every visual treatment in my work declares the information status of whatever it touches — solid means settled, dashed means open, struck means rejected and kept. A treatment that cannot name the status it is communicating does not ship.
Which makes the whole thing checkable. I can be told I got one wrong in a way that has an answer.
And a method that can fail has to show its failures, or the claim is worthless.
So I publish the rejections
Every design you have ever been shown is the survivor. The composition that shipped, the palette that stuck, the price that got published. What you are never shown is the four that lost, which means you have no way to tell the difference between a decision and a default. Both look identical once they are the only thing on the page.
On my own studio site there is a section, in the same type and at the same scale as everything else, listing six things the practice turned down — the price anchor it used to publish and no longer stands behind, the separate shop section, running its own checkout, selling the site itself as a template. Each one struck through, each one with the reason beside it.
None of that is flattering. A visitor arrives and the first structural thing they learn is what I will not do. It is also the only part of the page that proves a decision was made at all.
The harder discipline is the unfinished one. Two commercial terms on that page — what the files are built with, and the refund window — read not decided yet, in a state mark that wears a dashed edge instead of a solid one. They have looked like that for as long as the page has existed.
The alternative was not silence. The alternative was a sentence. It would have taken thirty seconds to write something plausible and true-ish, every reader would have accepted it, and I would then have been holding a commitment I never actually made. An undecided term wears the dashed edge. It never wears an invented sentence.
Then the same move, at product scale
This is where it stops being an argument about pages.
Eunosa is an AI tool for disability and return-to-work questions. It refuses to answer without a citation. Not prefers — refuses. The people relying on it are making decisions about someone’s livelihood, and against that, a confident guess is worse than no answer at all. The refusal is the product.
Lechio is built for neurodivergent workers, and its daily briefing has to be finishable. A list you can never get to the end of is not a tool. It is one more thing to feel behind on. So the design’s job there is to stop generating, which is the opposite of what every engagement instinct in software tells you to build.
Both are the same discipline as the dashed edge. Declare the limit. Build the limit in. Make the thing say what it cannot do, in the same voice it uses to say what it can.
What the counts do not tell you
Ten sites, platforms and portals. Eleven months. Zero days of formal design training. Five design systems, each governed by one rule a mark can fail.
Those tell you something got built. They do not tell you why it looks the way it does, and the why is one substitution: where a trained designer has a trained eye, I have a trained question. Both are instincts. Mine was installed by a different discipline, and it points at a different thing.
Making someone feel seen and heard is why I went into disability management.
It turns out that is also what design is.
https://www.meaganangelucci.com/writing/a-real-person-having-a-bad-dayPublished Aug 22, 2026