---
name: Spotted pattern
slug: spotted-pattern
category: layout
status: published
created: 2026-08-21T00:00:00.000Z
modified: 2026-08-21T00:00:00.000Z
definition: A scanning shape where the eye jumps to isolated words scattered
  down the page, hunting a keyword, a link or a number rather than reading
  anything.
aliases:
  - name: spotted scanning
    source: community
tags:
  - perception
relations:
  contrastWith:
    - marking-pattern
    - lawn-mower-pattern
    - commitment-pattern
    - f-pattern
  variantOf: []
  partOf: []
  seeAlso: []
implementations: []
sources:
  - title: "Text Scanning Patterns: Eyetracking Evidence, NN/g"
    url: https://www.nngroup.com/articles/text-scanning-patterns-eyetracking/
demo: inline
exhibit: false
useWhen: predicting a keyword hunt rather than reading
---

A reader arrives with a specific question: what does it cost, is there parking, which day.
The gaze plot that comes back is not a path at all but a scatter of isolated fixations down
the block, each one landing on a single word and none of them on the sentence it sits in.
Nielsen Norman Group calls this the spotted pattern. The reader is not reading and does not
intend to. They are running a visual search over the text for the shape of an answer.

What makes the search work is that some words look different from their neighbours. A
number, a bolded phrase, a link, a capitalised name, a currency symbol: each is a shape the
eye can find in peripheral vision without reading the words around it. That is the whole
argument for formatting keywords rather than trusting prose to carry them. Bold the two or
three words that answer the question, keep numerals as numerals instead of spelling them
out, and let links say the thing being linked to. In flat, unformatted text the same reader
finds nothing to land on and the scan degrades: a couple of fixations at the start of the
first lines, then the page gets abandoned or scrolled past.

The formatting has to be honest, and that is the constraint people miss. Bolding a whole
sentence gives the eye nothing to spot, because a uniform block is uniform whether it is
bold or not, and bolding words for emphasis rather than for findability sends the search to
the wrong place. Two or three spots per screen is the working budget. It is also worth
remembering that a spotted scan is often the *first* pass rather than the only one: the
reader lands on a number, then reads the sentence around it, which is why the answer has to
survive being found out of context.

The pattern sits alongside the other scanning shapes eye tracking has named, and it is the
one most directly changed by typography rather than by layout. The [F pattern](/f-pattern)
and the [layer cake pattern](/layer-cake-pattern) describe what dense text and subheaded
text get; [bypassing](/bypassing-pattern) describes a list whose items all start the same
way; [commitment](/commitment-pattern) is the rare case where somebody reads every word.
Spotted is the one to design for when the reader has a question, which is most of the time.