I have had the same conversation for ten years.
I say I'm a product designer, and the person across from me pictures someone drawing. Screens, colours, layouts. The word carries that meaning almost everywhere, and in Arabic it's heavier still — تصميم sits so close to graphic design and illustration that the two are nearly fused in ordinary speech.
So I've learned to answer with other fields, where nobody makes the mistake:
Designing a building is not drawing a building.
Designing a curriculum is not drawing a curriculum.
Nobody thinks designing a drug, or a perfume, is a matter of illustration.
In all of those, design is the reasoning that produces the thing. You place one decision on top of another, one piece of information on top of another, and the accumulation is what arrives at the result. The drawing, where there is one, is the last step and the least of it.
That's what I mean. Not that visual craft doesn't matter — it does, and I came to this from fine art. But that the drawing is the output of design, and mistaking output for practice is how you end up with beautiful products that don't work and organisations that hire designers too late to matter.
If design is building, then the decision is the unit of work.
Not the screen. So when I document a project, what gets recorded is the decision, its reasoning, its alternatives, and its consequences — because a screen without its reasoning can't be maintained by anyone who wasn't in the room.
If design is building, then structure is the real deliverable.
The hardest thing I've designed isn't a screen; it's a permission architecture, or a system where a bank's query to a customer became a structured object with a reason, a stage, an owner, and a specific file it points at. That object is why a replacement lands correctly and automatically. You can't draw your way to that.
And if design is building, then the constraint is the material.
A regulation, a country without digital identity, a bank that won't shorten its own declarations — these aren't obstacles in front of the work. They're what the work is made of. The Egypt journey exists in the shape it does because the country has no verifiable digital identity; the UAE journey looks nothing like it for exactly the same reason, inverted.
Security over usability. When they conflict, security is the target and good design is the route to it — not the other way round. A smoother screen is never worth a weaker one on a product that touches money or infrastructure.
Context over consistency. It is better to be right than to be consistent. Consistency is a means; treating it as an end produces interfaces that are uniformly wrong. The same data can deserve two opposite presentations when two different people are reading it for two different reasons.
Errors should be attributed, not forbidden. Past a point, stacking reviewers stops being safety and becomes paralysis. In systems whose entire purpose is speed, the honest protection is that every action is owned and timestamped, and that the person who decides is not the person who assessed.
Accessibility is comprehension, not contrast ratios. A person cannot consent to what they cannot understand. Reading level, unfamiliar terminology, and error prevention on legal commitments are inside the standard — and they are the parts that actually decide whether a regulated journey is usable.
When a digital journey removes a human, it inherits their job. The branch officer who read the declarations aloud and explained them was performing an accessibility service nobody had named. Removing him doesn't remove the obligation; it moves it into the interface.
I trust behaviour over statements.
In interviews, people can rarely tell you what's wrong. They've adapted; the friction has become the job. The most useful finding I've had in a bank came from sitting beside officers and watching them work through a screening report one row at a time — something none of them had thought to mention, because to them it was simply how the work was done.
So I ask, and then I watch. And where those two disagree, I believe the watching.
Every case file I publish contains something that broke.
That isn't modesty as a style. It's that a portfolio of only successes is a portfolio you can't learn anything from, and neither can I. The confirmation button that only appeared in one direction, the commercial register uploaded as a single page, the sequence I enforced that served one user and obstructed another — those taught me more than the parts that worked, and they're the parts I'd want to be asked about.
And where I have no evidence, I say so. Some of my work has no numbers I can stand behind. Writing no metrics are claimed costs me an impressive line and buys something worth more.
How this thinking shows up in practice → Systems · Work