Morkes & Nielson's study was done in 1997 but it is still a classic

Participants in an authoritative  study by Morkes and Neilsen (1997) said they use the Web for
•    technical support,
•    product information,
•    research for school reports and work,
•    employment opportunities,
•    sales leads,
•    investment information,
•    travel information,
•    weather reports,
•    shopping,
•    coupons,
•    real estate information,
•    games,
•    humor,
•    movie reviews,
•    email,
•    news,
•    sports scores,
•    horoscopes,
•    soap opera updates,
•    medical information,
•    historical information.

Scannable, concise, objective

People read differently on the web, so the writing style used needs to match this.

Studies have shown that readers scan pages rather than reading from start to finish, and if they do not quickly find the information they want, they move to a different web page.

What do users like?
•    users do not read on the Web; instead they scan the pages to get the information they want
•    users do not like long, scrolling pages: they prefer the text to be short and to the point
•    users prefer factual information and detest anything that seems like marketing.
•    users like summaries
•    users like inverted pyramid writing style, with the most important message upfront.
•     hypertext structure can be helpful,
•    graphical elements are liked if they complement the text

What helps users to recall the information they read?

Recall of information read on the web is variable, and is influenced by the style of writing used.
An early study (Morkes and Neilsen 1997) showed  the combined effect of using scannable,  concise and objective writing styles to be much larger than any of the individual improvements taken alone: a 124% improvement in measured usability, whereas the three individual improvements “only” scored from 27% to 58%.

So the message is clear- to be read, understood and recalled, writing on the web needs to be scannable, concise and objective.