Understanding how people read online

In order to design effective web content, we need to first understand how people read.

The fundamental truth is people rarely read word for word. This finding about online reading ibehaviour that hasn’t changed since Nielsen Norman Group conducted their eyetracking studies two decades ago.

Users don’t read, they scan

Users still primarily scan. Scanning all of the text, or even a majority is extremely rare. In fact, users only read 20 to 28% of a webpage. People scan the page to look for keywords that can point them to what they are looking for.

Eye tracking studies conducted by NN/g has identified multiple scanning patterns. The scanning patterns that arise are dependent on the user’s task, familiarity with the site, page layout and type of content (whether it’s text or image heavy).

While light scanning is the norm, the amount of time and level of detail is influenced by the users’ motivation and how important the information is to them. In most scanning patterns below, you can see that the fixation points are on the headings, sub-headings, images and call-to-action buttons. If a particular heading interests the user, they will then scan the description.

The only exception is the commitment pattern when the user is no longer scanning, but reading most of the words on the page. This is seen when users are very motivated to read and learn the content.

Word skipping

Word skipping is closely related to word length: as the length of the word increases, the probability that it will be skipped decreases, as research studies have shown. Short words (4–6 letters) are much more likely to be skipped than longer words (7–8 letters).

Users are more likely to skip shorter words that follow a longer word. So if a text uses longer, more complicated words, users will skip more.

Recognising words

While it is commonly thought that words are recognised by their shapes, the model that most cognitive psychologists accept as the most accurate is the parallel recognition model. Letters within a word are recognised simultaneously and we use that visual information to recognise a word. We also use contextual information to recognise words.

It has been established that our eyes don’t move smoothly across the page when we read, but make quick jumps from word to word. However, people with learning disabilities read letter for letter, they do not bounce already like other users. They may also have difficulties understanding a sentence which is too long.

The diagram shows points where a typical reader’s eyes fixates at and quick movements (saccades) across the text.

What users want

Users do not usually read text unless they want information. A successful scenario of information-finding looks like this:

1. I have a question
2. I can find the page with the answer easily
3. I have understood the information
4. I trust the information
5. I do not need anything else/I know how to proceed to the next step.

Therefore, an effective website should enable users to find what they need quickly, complete their tasks without making them think too hard about it.

Designing content that supports scanning

When we design content, we need to know:

• How users behave, what they are interested in or worried about so that the writing answers their question
• The audience’s vocabulary, so that we know what are the search terms and phrases they use

To support users in quickly finding what they need, these are some established guidelines:

• Use clear headings and subheadings
• Frontloading: Placing important information/keywords up front in structure of content
• Chunking: Breaking up content into small, distinct groups. Methods of chunking include having short paragraphs, short lines of text, grouping related items together.
• Highlighting keywords using colours, typeface weight, variation or hyperlink texts
• Having a short summary paragraph for longer sections of text
• Use bulleted lists
• Use simple vocabulary
• Use short sentences
• Highlight keywords (hyperlink texts, use of colours, typeface weights and variation)
• One idea per paragraph

This list is non-exhaustive and the above guidelines all contribute to designing content that is easy to scan and understand, supporting users in finding what they need on a website.