~/wiki / figma-i-makety / ierarkhiya-i-struktura-ekrana

Nobody reads your screen. Here's why - and how to change it

Main chat

A chat for vibe coders: news, guides, live cases, marketplace, and finding executors.

$ cd section/ $ join vibe dev
Nobody reads your screen. Here's why - and how to change it - обложка

How the eye actually moves

There are two main scanning patterns.

F-pattern on the text pages. The view goes horizontally on the top, then horizontally again a little lower, but shorter, then vertically down the left edge. The right-hand side of the screen and the bottom paragraphs hardly read.

Z-pattern - on pages with visual content. Looking diagonally: left top → right top → left bottom → right bottom → right bottom. It works well for landing pages and marketing pages.

The F-pattern is not a law of nature. This is a symptom of a poorly designed hierarchy. If the design does not direct the view, the user switches to fast scan mode and passes on F.

A good hierarchy breaks the F pattern and makes the eye stop where it should.


Why most screens don’t work for scanning

There are a few typical reasons.

All the same weight. Headings are not much larger than subheadings, subheadings almost do not differ from the main text. There's nothing to catch on to. This is one of the most common problems in AI generation: models build a formally correct H1 → H2 → body hierarchy, but without enough visual contrast between levels.

The important thing is not at the beginning of the line. The F pattern reads the first words of each line better than the next. If the paragraph begins with the introductory word – “Also”, “In addition”, “Should be noted” – the key content in the middle will be lost.

Long blocks without interruptions. Continuous text without subheadings and visual breaks is a signal for the brain to read carefully. Most users at the sight of such a block do not include a careful mode. They're missing the whole thing.

  • Too many accents. * Bold everywhere, a few accent colors, accents. When everything is highlighted, nothing stands out. The emphasis works only in contrast to the neutral background.

How to fix it

Three levels of typography - and they should be markedly different

Working system for mobile screen:

Уровень Размер Начертание
H1 28–32px bold / semibold
H2 20–22px medium / semibold
Body 16px regular

On the H1 desktop from 36px. If you remove the titles and the screen looks similar, the hierarchy doesn’t work.


Keywords for the beginning

The F-pattern reads the beginning of the line best. This means that the keyword should come first – not after the introductory design.

Bad:

plaintext
For the convenience of teamwork, you can use...

Good:

plaintext
Teamwork - without unnecessary tools

Bad:

plaintext
It is also worth noting that the download speed...

Good:

plaintext
Download speed – less than 2 seconds

Break the text with visual anchors

A long block without interruptions, the user will run his eyes and leave. We need stop points.

What works as an anchor:

  • subtitle every 3-5 paragraphs
  • keyword bold – no more than one or two per paragraph
  • short list instead of comma listing
  • callout block with an important thesis

What does not work: emphasize without reference, italic for the entire paragraph, highlighting random words in color.


Scan test

Close the entire text, leaving only the headings and first words of each paragraph. Is the page clear?

If so, the scanning is good. If not, the headlines are decorative, not semantic.


Separately about mobile screens

On mobile, the F-pattern persists, but attention spans shift.

The lines are shorter, so the first words are even more important. On a narrow screen, the user sees fewer words in a line, and the beginning of the line gets even more attention.

Skroll is faster - the content flickers. Each block has less time to catch the eye.

The Thumb zone, the comfortable thumb-touch zone, is the lower central part of the screen. The main CTA is better placed there, not in the upper corner.


How to check screens after generation in AI tools

Figma Make and Google Stitch generate neat screens, but the hierarchy is often weaker than it seems. The model knows that the title should be bigger – but doesn’t know how big it should be.

After generation, check three things.

**Blurr test.**Wash the screenshot or step back at arm's length. What stands out? If the hierarchy is not read in blurry, it is not contrasting enough.

** Heading test. **Read only headlines and subheadings. Is the structure and essence of the page clear?

**Test the first words. **Take the first words of each paragraph. Do they make sense or are they introductory designs?


Outcome

Проблема Симптом Решение
Всё одинакового веса Взгляд не знает, куда идти Контрастные уровни типографики
Важное не в начале Ключевые слова теряются Ключевое — в первые два слова
Длинные блоки без разрывов Пользователь пропускает текст Подзаголовки, списки, callout
Слишком много акцентов Ничто не выделяется Один акцент на блок

Users are scanning - and that can't be changed. You can only design for it.

$ cd ../ ← back to Figma and layouts