Nobody asked for the footer

Yesterday I spent an hour on the footer of a website almost nobody scrolls to the bottom of. It's a paper landscape cut into six layers, and each one moves at its own speed as you scroll.

Nobody asked for it. No user is ever going to write in about a missing parallax footer.

TL;DRThe last twenty percent, the polish nobody requests, used to be rationed by what I could build myself, so anything needing a designer never got done. Now that part costs an hour instead of weeks, which means the list of details worth doing just got a lot longer.

I already made this argument

Back in February I wrote that the last 20% is the product. Empty states, error states, micro-feedback, the small interactions nobody requests and everybody feels.

I spent three days that week on microinteractions in the app. A copy button that slides up only after an answer finishes, a short vibration, the icon flipping to a checkmark and back. There are a hundred or two hundred of those in Edicek now.

I still believe every word of it. But there's something I didn't say, mostly because I hadn't noticed it yet.

The polish I could afford was the polish I could code

Every one of those hundred interactions had one thing in common. I could build it myself, in code, at midnight, for free.

That quietly decided which details got done. Anything that lived in code was fair game. Anything that needed artwork was not.

So the list of things I wanted and never did kept growing. Illustrated empty states. A proper onboarding sequence with visuals. Animated infographics on the landing page. A footer that isn't just a row of links. None of those were rejected on merit. They were rejected because each one meant a brief, a designer, a budget, and weeks of waiting for something I could look at but not change.

The last twenty percent wasn't rationed by discipline. It was rationed by which half of it I could reach.

An hour

The footer needed artwork and animation, so under the old rules it was never happening.

It took an hour. I generated the scene, cut it into six transparent layers, and tied each one to the scroll position. The reason that's possible at all is the brand system underneath it, which took months to settle and which I wrote about when the tooling finally caught up.

An hour is not free. But an hour is a decision you make on a Tuesday. Weeks and a budget is a decision you defend.

An hour is still an hour

It wasn't an hour of admiring my own work.

The scene is generated as one image, and each layer is then pulled back out of that same picture. Six separate generations invent six different shapes and nothing lines up.

Then the part I didn't see coming. Every layer has to be the whole sheet of paper, not the piece you can see. In a flat scene the back wave is mostly hidden behind the ones in front of it, so what you get handed is that thin visible strip. Standing still it looks perfect. Then it starts moving and the strip is floating in mid-air with a gap underneath it.

Colors drift too. Isolate a muted green and it comes back as pale mint, so you end up naming the hex in the prompt and checking it against the scene it came from.

The list is the point

Toast style. Notification style. Empty states. Onboarding. The way an infographic moves on a landing page.

None of those are features. You can ship without every one of them and the product still works, which is exactly why they get cut first. In isolation each one looks like a waste of an afternoon.

Documentation sits on the same list and is the clearest case of it. Everyone wants to build the next thing. Almost nobody wants to go back and write down how the last one works, and fewer still want to keep it true six months later. It ships nothing. It just quietly decides whether anyone can use what you already built.

But nobody meets them in isolation. They arrive stacked, across the ten minutes somebody spends with your product for the first time. My guess is that thirty or forty percent of whether a product lands sits in how those ten minutes feel. I can't prove that number and I won't pretend it's research. I've just never seen anything win on correctness alone.

What changed isn't that the list matters. It always mattered. It's that most of it used to be out of reach, and now it's an afternoon each.