Jakob Nielsen posted a great article, "First 2 Words: A Signal for the Scanning Eye," where he goes into the details of his research in this area.
Jakob starts with reminding us how our customers have many lists to read, err scan -- it's not just search results.
- Search result pages
- List of current and/or archived articles and press releases
- Product listing
- Table of contents
- Question lists on an FAQ page
- Bulleted and numbered lists, checklists, etc
11 characters is used as the baseline measure the number of characters a user actually reads when looking through a list (link text). The test is to see just the first 11 characters, and see if you can predict what's behind the link -- what will you get when you click.
The best link text has these characteristics.
- Plain language
- Specific terminology
- Follow naming conventions
- Front-loaded, action-oriented terms
The first 3 have something in common -- no slang or internal jargon.
What's the characteristics of bad link text?
- Generic words
- Made-up terms
- Having the information-carrying text at the end
Jakob concludes with a reminder that our customers do have the option to read past the first 11 characters. What we need to remember is to front-load our titles and links, so the customer, when necessary will read the rest of it, i.e. it needs to grab their attention to read enough to validate it's the correct link (which would be followed by a click). "Nanocontent (first bit of a link) just needs to be good enough that users will sniff the most promising links in full." How?
- Provide enough additional context, so the user knows what to expect (and make sure the clicked-to page actually delivers).
- Clearly differentiate links -- don't make me choose between 2 very similar links.
- Don't mislead or over-promise.
It's all so simple. Right?