Posted on
by
Jared Spool
The Lives of Links
#linklives
  • trigger words:
    • the things that cause users to click
    • we need words to describe our tools
    • trigger words is an example of that.
  • we’ve been studying how people use the web since 1995
  • we had this theory in 1995: that people who know how to use the web would be better at doing things on the web.
    • we got people with different levels of experience
    • we set them down in front of web sites and had them do things
    • it turn out that people’s experience didn’t matter, but what did matter was design of the web site.
    • (will: this is different than something like using a tool like a circular saw, in which experience is important.)
  • Predicting Failures of Scent
    • Use of the Back Button
    • Pogo-sticking
    • Using search
  • We have thousands of clickstreams
    • we look for patterns.
    • we have two piles: those for people who succeeded, and those who failed.
  • Backbuttons predict failure:
    • For the clickstreams where people use the backbutton once: only 18% are successful.
    • For the clickstreams where people use the backbutton twice: less than 2% are successful.
  • Pogo-sticking also predicts failure
    • Jumping up and down through the site hierarchy
    • People who pogo-stick are only 11% successful
  • Search predicts failure
    • (except on Amazon)
    • When people do search, they type in trigger words
  • Search Pro tip:
    • Your search logs are filled with trigger words
    • Ideally you want the logs to include the page the user was searching from
    • So put the trigger words as links on the page that users were searching from
  • 7-Eleven Milk Experiment
  • Compelled Shopping: Buying Apparel
    • Give the prospective customer $1,000 to buy the clothes they want.
    • “Ideal site”: the customer should spend $1,000
    • Gap: $660
    • Lands’ End: $465
    • Macy’s: $156
    • Newport News: $63
  • Number of clicks to get to final purchase
    • The Gap: 11.9
    • Lands’ End: 15.7
    • Macy’s: 51
    • Newport News: 51
  • Examples of shopping sites: some sites force you to click through to the individual products to get details. No way to compare, no way to see data on individual products. It forces pogo-sticking.
    • Good example: Crutchfield. Shows more data. It’s exactly the differentiated data you want.
    • Bad examples: No data.
    • More bad examples: Meaningless data “technology you trust”
    • More bad examples: Showing the same bullet points for everything: “No annual fee.” “0% interest”. Also useless in comparison.
  • Most useless words in web design:
    • click here
    • learn more
    • click here to learn more
  • Good Design is Invisible
    • It’s like air-conditioning in a room: you don’t notice it, unless it’s bad or it’s not working well.
  • Links secretly live to look good.
    • But they still have to look like links.
  • We used to think that links are supposed to be blue and underlined. Thankfully, we’ve moved back this.
  • In some cases, we can find it out. The page has a clear visual language.
  • In other cases, we can’t tell without waving our mouse all over.
  • Look Good: You have to establish a consistent visual language.
  • Links have to do what you expect.
  • Example of dictionary.com: it’s hard to find the content.
    • The page is full of links and advertisements. You want the user to stay.
  • Other examples: in the middle of articles, there are links to go elsewhere.
    • to related articles
    • to unrelated articles
    • but why?
    • let the person finish the article
  • Taco Bell Advertising Lawsuit example:
    • good article, but…
    • link to everything the tribune has ever written about alabama in the middle of the text
    • link to crossword puzzle and sudoku in the middle of the page
    • advertisements to sue people or to advertise on the page
  • It’s all junk: it’s not really relevant or related.
  • Flyouts:
    • Other sites hide the links
    • you have to put the cursor over stuff to find the links.
    • People want to see the links. That’s why they are using the web.
    • Flyouts are fun to code, but they aren’t fun for the user.
  • Do What The User Expects
    • Deliver users to their desired objective
    • Emit the right scent
    • Look good, while still looking like a link
    • Do what the user expects