SEO copywriting: Searchwritten

According to a recent publication by Jakob Nielsen, it is better to not use many words in a web text: less is more. However, another study by the Poynter group reveals that web content (in online newspapers) is read way, way better!

How can we explain this difference? And: what should SEOs learn from it?

Ok, so now i’m confused.

Last week Jakob Nielsen wrote an article entitled: ‘How little do users read?’ In the article, he repeats an idea we all know: average web users don’t read a text as a whole, they scan it. He also states that about 20% of the words in texts can actually be read in the time people spend on a web page.

Another study however, (Poynter group Eyetracking study 2007) says something different. They say that 77% of a text that is being read, actually gets read, even if the text is very long.

What’s up with that? I got frustrated and began reading to find out. Is one of them talking BS, or is there more to it?

Why less is more: Jakob Nielsen about web reading

Nielsen provides some data to back up his claim:

‘On the average Web page, users have time to read at most 28% of the words during an average visit; 20% is more likely‘.

The data set he used contained 59,573 pageviews, of which 45,237 were usable after cleaning out error pages, bounces, and pageviews that lasted more than 10 minutes. The data was collected in 2005 among mostly high-end users (university employees).

Time to read: 25 seconds + 4,4 seconds for every 100 words extra

The calculations are simple, i’ll summarize and explain them here:

• an average webpage contains between 30 and 1,250 words. Reading behavior beyond 1,250 words becomes erratic;
• an average pageview lasts at least 25 seconds (30 words), plus 4,4 seconds more for every 100 words extra;
• assuming an average reading speed of 250 words per minute, users can read 18 words in those 4,4 seconds. So when an extra 100 words are to be read, people actually read only 18 of them.

Amount of read words: only 20% to 28%

The next step is to calculate how much words people are able to read when an article contains a given number of words. In the data set Nielsen used there was an average of 593 words on a webpage. I have no way of validating that number, but it seems reasonable.

Source: http://www.useit.com/alertbox/percent-text-read.html

Look at the chart above. The smaller the amount of words, the higher the percentage of words people have time for. People read 50% of a text when it’s 111 words or less. At Onetomarket we generally use 250 words as an ideal quantity for both search engines and humans. It seems that when a text is 250 words long, about 35% it will be read. The average pageview of 593 words would only be read for 28%.

Nielsen lowers the average number to 20% because people do not spend all their time on a webpage reading. They also need time to figure out the navigation and lay-out, etc.

Poynter group says: less it NOT more

Ok. So 20% of web texts are read, on average.

But that is not what the Eyetracking Study of the Poynter Group says! For those of you who don’t know: the Poynter group is a non-profit organization that does all kinds of teaching stuff for journalists.

In 2007, they conducted a large eye tracking study: 582 subjects, 50% men/women, average age of about 35 to 40. About 200 newspaper websites were examined (also 200 tabloids and 200 newspapers).

The research revealed the following, among other things:

• 77% of the words are read after a subject decides to read an article;
• half of the people read from top to bottom (methodical); the other half scans;
• alternative story forms (Q&A’s, timelines, lists, facts boxes) draw in a higher amount of visual attention;
• when online, people search for navigational elements first. When reading print, people search for big headings and photos first.

The most striking difference between this study and Nielsen’s statements, is that ‘less is more’ doesn’t seem applicable for news articles. If someone starts to read, they go for the full monty:

Source: http://eyetrack.poynter.org/keys_01.html

Conclusion: users are picky, and quality matters

So one study concludes that less is more; the other says exactly the opposite. How can we explain this?

My two cents on it are that they differ because of the way the research is set up. There are some striking differences:

• the Poynter institute explicitly told test subjects to read a newspaper, tabloid or website. It makes sense that a much higher percentage would actually read stuff. It’s not that this is a bad way to do research! But because of this, it can never, ever account for the way people interact in general with websites;
• also, newspaper websites are high authority websites, and are packed with high quality content. They are the traditional news publishers. It’s logical that people read more on these websites;

It seems to me that there are two very logical observations here:

1. people are very picky when choosing which articles to read;
2. but if they decide to read, they probably read more than 20%. This however, only applies to newspaper websites, but we can assume that websites that are known to deliver ‘quality’ information, can reach this high a percentage as well. I do not think the same percentage is applicable for web shops and your average business website.

What about SEO?

The reason I wanted to dive into these researches is because of my daily routine as an SEO. I give a lot of workshops about the relationship between copywriting and search engine optimization. Usually, I teach copywriters about keyword research, writing title tags and meta-descriptions, how many keywords to use in your text and where to put them, etc.

These are the things I will use in my presentations:

1. Time to read: 25 seconds + 4,4 seconds for 100 words extra; (Nielsen)
   20% to 28% of web texts can ON AVERAGE be read given the time people spend on webpages; (Nielsen)
2. however, IF people read a news story, they read 77% of the text (Poynter);
3. this high percentage is presumably lower on web shops and business sites (IMHO);
4. but if people find a text interesting, they WILL read it (IMHO);
5. people read in different ways: they scan (50%) AND read from top to bottom (50%) (Poynter);
6. navigational elements AND alternative story forms attract readers. Secondly, headings and images are important (Poynter).

Besides the conclusions above that are more about web reading in general, there are also several hidden messages for SEOs.

1. If a visitor is supposed to ‘do’ something on a webpage (like reading or converting into a customer) it is paramount that the user experience comes first, and keywords second! That might be hard for many SEOs to swallow. But because REAL user attention seems so hard to obtain, everything should be done to reach that goal first. Never, ever sacrifice the quality of a text for search engines!
2. the thing a reader should ‘do’ on a webpage cannot be reached if they must read the WHOLE text first. That is something visitors simply do not do. We have seen that they read something between 20% and 77%, but never 100%. So, when doing textual conversion optimization you must make sure the conversion goal is present on several locations of the page. Otherwise it might not even get noticed;
3. Also, they read texts in different ways: they scan and read methodically. Also, visitors tend to read more when you use alternative formats, like Q&A’s, lists, etc. So, diversify your conversion points.

Well, this article has become quite big too, but i hope you find it interesting enough to read the full 77% ;)

Read more:

• Jakob Nielsen: How Little Do Users Read? www.useit.com/alertbox/percent-text-read.html
• H.Weinreich et al., Not quite the average: An empirical study of Web use doi.acm.org/10.1145/1326561.1326566
• Poynter Group Eyetracking 2007: eyetrack.poynter.org/.