April 10, 2007
Jakob Nielsen's new book, Prioritizing Web Usability, is a worthy companion to the previous two. Now it's a trilogy:
- Designing Web Usability: The Practice of Simplicity (2000)
- Homepage Usability: 50 Websites Deconstructed (2002)
- Prioritizing Web Usability (2006)
You can tell Jakob and his co-authors are growing ever more skilled at the practice of simplicity; this book is the first in the series to drop the colon and subtitle.
The very existence of the new, updated book hints that usability guidelines evolve over time. In one of the first chapters, Nielsen makes this explicit by revisiting earlier web usability issues to see how much they've improved in seven years. Each usability issue is rated from zero to three skulls to indicate how severe the problem is today:
1. Usability issues that are still major problems today
|Links that don't change color when visited|
|Breaking the back button|
|Opening new browser windows|
|Design elements that look like advertisements|
|Violating web-wide conventions|
|Vaporous content and empty hype|
|Dense content and unscannable text|
2. Usability issues that are less important due to improvements in technology
|Slow download times|
|Adobe Flash content|
|Low-relevancy search listings|
|Multimedia and long videos|
3. Usability issues that are less important because users have adapted to the web
|Pull-down and cascading menus|
4. Usability issues that are less important because designers have learned restraint
|Plug-ins and bleeding edge technology|
|3D user interfaces|
|Moving graphics and scrolling text|
|Custom GUI widgets|
|Not disclosing who's behind information|
|Inconsistency within a web site|
|Premature requests for personal information|
When comparing the severity of these 34 usability issues with their historical severity in 2000, Nielsen notes that most of the progress can be attributed to designers learning restraint:
|Resolved by user adaptation||11%|
|Resolved by advances in technology||10%|
|Resolved by designer restraint||21%|
|Still an issue||58%|
Relying on user education or technology fixes to address usability issues means you'll be waiting a long time. Most of the immediate benefits are realized by designers who learn to follow usability guidelines. But designers are fallible, too, so there's no guarantee these problems won't crop up again later, or in slightly different forms.
The data presented in Prioritizing Web Usability shows that usability guidelines do evolve over time, but slowly. It also illustrates how the core principles of usability are timeless:
From 1984 to 1986, the U.S. Air Force compiled existing usability knowledge into a single, well-organized set of guidelines for its user interface designers called Guidelines for Desigining User Interface Software, ESD-TR-86-278 (also available as a pdf). Jakob Nielsen was one of several people who advised the undertaking. The project identified 944 guidelines, most of them related to military command and control systems built in the 1970s and early 1980s, which used mainframe technology.
You might think that these old findings would be irrelevant to today's designers. If so, you'd be wrong. As an experiment, we retested 60 of the 1986 guidelines in 2005. Of these, 54 continue to be valid today. Of the total 944 guidelines, we deduced that 10 percent are no longer valid and 20 percent are irrelevant because they relate to rarely used interface technologies. But nearly 70 percent of the orginal guidelines continue to be both correct and relevant 20 years later.
This is one of the reasons I urge software developers to study and understand the principles of usability. It's one of the precious few bodies of knowledge in a developer's toolkit that will still be useful twenty years from now.
Posted by Jeff Atwood
I will never understand why (vertical) scrolling is considered an usability issue. Nobody ever complained about scrolling in Word, Notepad or Windows Explorer. Seems like forcing physical limitations (pagesize) into the virtual space.
The scrolling issue is basically resolved; users have learned to scroll.
Plus computer mice somehow "grew" a scrollwheel between 1996 and 2003. That's kinda significant. (When was the mouse scroll wheel introduced? When did it become mainstream? By 2001?)
The two skulls for scrolling, according to the text of the book, is because most users will visit your site and discard it immediately if nothing VISIBLE looks relevant to them. You have about 25-35 seconds and 100 words to convince them there's something there worth scrolling to before they bother.
Also, in terms of writing for the web, it's important to put the most relevant information at the top of the article or page:
Regarding scrolling, few annoyances are more irritating than horizontal scrolling. Vertical scrolling I don't mind, but scrolling from left to right and back is enough to drive me away from the most informative web page.
:-) it seems to me that we have a usability problem with this article...it's not so easy to discern which skulls relate to which item. a table with headings would make this much easier to read!
His statement that most of the progress can be attributed to designers' restraint is very misleading. It does have the highest percentage of the available choices, but does NOT make up the majority. It makes up half.
With those same statistics, we could very well say, "You can ignore fully half of what Nielsen ever says because users will adapt and technology will improve."
Which has always been my opinion of him, anyway. He has some great insights, but I tend to not worry about the minutiae of his writings.
Or else put the skulls on the left, so you could easily make the association between the skulls and the text.
The link to ESD-TR-86-278 is broken - s/hef/href/
Sheesh, everyone's a critic. :) Corrected issues from Mike, Barry, and Morgan. Now can we discuss the topic of the article instead of the article itself?
I still disagree with a:visited colors. It was a valid point when the same colors were applied to every single link on every single page but, now that each website defines its own colors, either you'll have a difference so subtle it won't be evident, or you'll make a confusing technicolor mess of your content.
(Re: vertical scrolling, and reading long text in general on the web, I think the popularity of blogs has done as much in this regard as scrollwheels.)
From the interview:
"I'll give you an example from my own company: we sell a report with usability guidelines for email newsletters. The previous edition was 293 pages. The new edition is 544 pages because we've added findings of how people read email from a recent eyetracking study."
I refuse to take any advice from a guy who touts usability but can't manage to cut down a 544-page report to a, um, *usable* size.
About time somebody drops the colon and subtitle. The whole pattern of...
Catchy Title: But Here in the Subtitle Is What the Book Is Really About, Because You Can't Tell From the Catchy Title
...is getting a bit old.
Here is irony - to buy the book, there is a link to Amazon, a web site that I have vowed never to use again because of its bloat, low signal-to -noise ratio, and LACK OF USABILITY.
But for your sake, Jeff, I'll hope for a high click-though rate in defiance of my bad attitude.
Scrolling: Get rid of the ads, giant banners and mastheads, and sparse "intro" text.
Technicolor mess: Get rid of the gratuitous color.
Cut the bloat: I don't care how much bandwidth I have, my time is still valuable.
Sneaky popups, Flash, PDFs: We need a blacklist proxy we can all use and list these suckers with. These things are like deer, potholes, and speedbumps on the Information Superhighway.
Vertical scrolling is still my biggest complaint! Nobody does a decent job of it. I want one screen to start with, and when I hit the page down key, I want the next screen. I do not want the last line of the first screen and then the next screen. I want it to work like a book. Is that so hard? Yes, I know, everybody has a different screen size, and they all use different text sizes, and gee, it's really hard to figure out how many lines will be on each screen. Well, geez, gumby, you've got a computer, let if count the friggin lines.
And I don't want half a line a text (where the upper or lower half of the letters is cut off).
I don't want to have to figure out where to pick up where I left off when I have scrolled down with a scroll wheel.
What is absolutely the worst is when you get to the end of an article, and there is only half a page left, and then you try to find your place, and you end up rereading the entire last page to find out that there was only one new line on the last page.
I don't like scrolling text at all. It hurts my eyes to look at it. Computer screens have lousy resolution (compared to paper).
Wikipedia is the worst. Page down works a few times, and then it does not work anymore.
The problem is: Our client don't like usability, they want the new thing they saw in the new-cool-whatever-website.com
I'm just tired to try to convince them.
"now that each website defines its own colors, either you'll have a difference so subtle it won't be evident, or you'll make a confusing technicolor mess of your content."
One extra color isn't going to break the bank, especially when it's actually _useful_, unlike most of the rest. (Technicolor is certainly preferable to taking monotone to the extreme, blending links into text, though a competent designer can differenciate visited links without breaking the theme at all.)
It's Nielsen, N-I-E-lsen, like in the picture of the book cover.
I agree with Garoo. Having a different colour for visited links just is not usable for everything. Why would I want to mark a link "My Preferences" (for example) as a visited link? The link will be visited at different times for different reasons.
I can maybe understand marking an outgoing link, or a link to specific content, but even then it is not always required.
I would agree though that links should probably be underlined, and underlining should not be used for anything else, purely because this is immediately recognisable.
I guess I should read the book for details, but I tend to have problems with laundry lists of what not to do, and no advice for what you should do.
Not being a bad site is not the same as being a good site.
Plus, how do you measure things like "Violating web-wide conventions" and "Vaporous content and empty hype".
I could list quite a few of my own ideas on what I consider web-wide conventions or empty hype.
Learn usability, but don't do it at a University, huh? :-)
Having a different colour for visited links
just is not usable for everything
You're wrong. The typical case is a user who tries to find something on your site and clicks from page to page. Knowing when a link leads to a page he has already visited is extremely valuable.
Oddly enough, even though I read this article, every time I see the headline I think it says, "Timeablity is Useless" instead of "Usability is Timeless". Weird brain.
Actually LKM I would agree with Mladen. For the majority of links, yes you should colourise visited links. However, for times such as his cited example of frequently visited links with a specific and obvious purpose, such as Preferences, it doesn't necessarily need colourising. Though it doesn't hurt too much to do it, and it's probably more work to remove the colourising from just that link.
The problem with "The Horizontal Way" is that your content really does have to be in discrete blocks of self-contained text. Without that tracking your last position when you scroll is a complete nightmare. When I scroll down a page my eyes can keep track of my position, and even keep reading. When I scroll horizontally my position is completely disrupted because all the text is moving the opposite way to the direction in which I read.
"I want it to work like a book. Is that so hard?"
It's not a book. Books don't update themselves automatically when new, relevant information comes out. Making new technology comparable to old technology is moving backwards. Don't get me wrong, I love books, but I can read a computer screen much faster (as fast as it scrolls when you click the down arrow).
"I don't like scrolling text at all. It hurts my eyes to look at it. Computer screens have lousy resolution (compared to paper)."
Maybe you should adjust your text dpi. You can find it in windows under Display Properties-Settings-Advanced. I did this for my Mom(who has a similar problem), and now she can read the screen better. Or just get thicker glasses ;-)
Jakob Nielsen has some good points, but still it can be so draining for creativity and soul to read his stuff. Just looking at his homepage useit.com, which has looked like that for year still hurts my eyes. Jakob at least has proven that usability and design dont have to go hand in hand. :)
I agree with Ali, the web is not books.
About the "back button" breaking things, I think the back button has broken much more then the "not working back button". When people go back and re-submit http post requests and result in dublicate submitted forms. Yes it is a matter of implemenatation, but still...
Peter: The back button does not re-submit post requests. You're thinking of that refresh button, which is a bit harder to break.
I definitely agree about the appearance of Nielsen's site. I think Jeff posted a link some time ago to a redesign of it that looked a lot better.
Seems to me that he basically rules out doing anything. For example, let's say I use an unusual word that 50% of the readers won't know, and I want to provide a definition of it.
Well, I can't do a popup div, because that's a custom widget, violates web conventions and it might look like a popup window or an ad (since there are a lot of ads in popup divs these days).
Can't do a new browser window, that's got three whole skulls!
Can't put the definition inline because that leads to dense content, not to mention missing the point of hypertext. You could argue it should be defined the first time it's used, but this is the web. How do I know if it's the first time the user's seen it when there are multiple paths to the same content?
I guess that leaves me with creating an entire new page for the definition and linking the user off to that. But isn't taking the user out of the context of what he's reading bad as well? Isn't it even worse than a popup window?
Maybe I'll start caring what Nielsen says it stops amounting to "Don't do anything!"
Anyone have a link to any of the articles that discuss link colors?
The traditional colors used (before everyone started changing colors of everything with css) were shown to get better click ratios than non-standard colors.
Seems to me that he basically rules out doing anything. For example, let's say I use an unusual word that 50% of the readers won't know, and I want to provide a definition of it.
You could use the, say, abbr tag. At least in Firefox, it will add a dotted underline to the link and then on hover, pop up a tooltip with the expansion of the abbreviation. Semantically, it's poor form, but if you're *that* desperate to provide a definition...
My problem with Nielsen, although I agree with him on many points, is that his philosophy caters to the lowest I.Q. It's a Harrison Bergeron situation, which the Government (through various lobby groups) actively endorses. To his credit, he's just trying to make a living. It's his dogmatic disciples I fear most.
While the Government is meant to serve the people, and therefore needs to include ALL people when considering its services, I believe it is up to the company or individual to decide how to serve its audience.
Design Creativity needs a sponsor. Art for Art's sake is still a new concept (in the history of humanity), and Captialism continues to demonstrate that Art in its purest form is not for general consumption. Industrial design remains a critical pillar of Art's influence in the world.
If we strip design from the web, we will never have Internet design periods equivalent to the Rennaisance, the Baroque, Gothic, Classical, etc. We will move directly into the shapeless, uncharacterized designs that plague our modern cities' architecture (and alas, our very neighborhoods). Embellishment may not be required for functionality, but who among us doesn't appreciate the fluted columns of Roman architecture, a detailed moulding around the door of one's abode, or that line of rivets around the arm of an old leather recliner.
Aesthetic is as important as usability and can actually drive desire, no matter how unworkable it is. Witness: high-heeled shoes. You'd figure we could get shoes right after millenia of testing.
About splash pages.
They can be required at times. In situations where there are several languages involved (multi-lingual countries or Europe) where languages issues can get people pretty finnicky, a splash page that says "Welcome" in the several languages involved is a good way to avoid bias perception, which can sometimes be detrimental.
Of course, the splash page should be quick to load, and the site be left unframed so one can bookmark the page in his favourite language right away.
It is also a good idea to put a link on each page to the other language, such as done on this website I am currently maintaining: http://www.st-armand.com
I just scanned through the government standards document and was impressed with the simple language. Easy enough for a computer programmer to follow. It even lays out the unspoken rules such as "reduce transitions between keyboard and light pen" and "provide fast feedback to user actions". OK, I don't use a light pen, but the mouse is the analog and many forms designers could use some lessons in that regard. And we don't worry about the lag on a keypress any more, but I still encounter drop down menus and combo boxes that take 2 or 3 seconds to do their thing when you click on them.
I was actually talking about the back button, but yes you could say it is the refresh button. But I was thinking of the scenario below:
User see a typical form, fills out all the fields and click submit.
The server then generates alot of things, like ordering a product and sends the result generated to the user.
The user go to something else but remember that he/she forgot something and click back and then the browser asks if the user want to send the post request again, and suddenly they have submitted the old form twice.
Back/forward is very usefull, but seems like an old navigation hack.
I swear I had no idea Vonnegut had died when I mentioned Harrison Bergeron. If you don't know the reference, I encourage you to read the short story in Vonnegut's honor today.
ielsen's work is a helpful guideline. It can be summed up by Jakob's Law of the Web User Experience :
"Users spend most of their time on other websites"
That's it. People are used to certain behavior. If your behavior is different, you'd better have a very good reason. People feel stupid when they don't understand how your site works, and making people feel stupid is a great way to lose customers, friends, etc.
Sure, people will learn the new behaviors if they spend enough time on your site. Unless your are Yahoo or Amazon, they probably will not spend enough time on your site to learn the new behaviors. The users will just come by every once in a while and feel stupid.
Make stupid people feel smart! That is the key to success.
This isn't art, it's design. Think crescent wrench not Michelangelo's David
A simple, understandable interface is not "dumbed down." It is easy to use. It is good design.
Nobody cares about some super-duper color scheme that can't handle one more color (for visited links). People only notice horrible color schemes, so focus on function.
Users don't come to your site for the experience, they come for the content. That is why browsers use vertical scrolling: our eyes scan up-and-down. (think grocery list vs. line of prose)
It is a little mis-leading to say that people don't scroll, but it is a good design rule of thumb. People only scroll your page if they are already reading. Only the content and the footer should ever exist more than 1500px down.
Great information Jeff, thank you!
I bought the book, but it's mostly focused on how usability issues have changed over time, which is interesting, but I'm more concerned about what's a problem now. I find the O'Reilly Designing Interfaces book to be a much better usability problem solver.
"When people go back and re-submit http post requests and result in dublicate submitted forms"
Don't blame the user for shitty web applications.
Using GET/links for views allows users to bookmark pages and navigate freely.
Use POST/submit-buttons for actions, the user will be notified when resubmitting the page. The page won't be resubmitted when using the back button, only on F5/refresh.
I also use transaction IDs in FORMs, usually these are the UUIDs used as primary keys for appending to the database. If the user hits F5, this will simply yield a "PRIMARY KEY CONSTRAINT" violation which can be formatted to a nice, comprehensible text message.
I think we have two kinds of people here: 1) people who are totally in tune with the web and their browser and can read scrolling pages and write 90 words a minute, and 2) me, the neanderthal, who likes books, doesn't like things moving around, and types 90 characters a minute.
And then there are all the people who are not here (reading this blog), who use the web, and they come in about 16 different flavors. So your web site design really needs to take into account your audience, intended or not. People read different ways. Some people skim and just absorb the overall sense, others read every word and absorb (intake?) every sentence.
"Violating web-wide conventions"
- "Links that don't change color when visited"
- "Breaking the back button"
- "Opening new browser windows"
- "Pop-up windows"
I'm okay with Nielsen. He's got a paycheck, he seems to enjoy what he's doing, and he tends to practice what he preaches. What gives me pause is his antagonism. He treats the designers of today as some variety of Luddite who are incapable of accepting new trends, let alone new technologies. And most importantly, he forgets that most websites (and the usability attached to it) should be designed for specific audiences. Slashdot is not designed for general users. MSDN is not designed for general users. Design portfolios are not designed for general users.
Nielsen artificially inflates the prospected ROI for companies who want to be all things to all people with his repeated mantras. I, personally, do not want to see the web be so sanitized. I like my corners that break all the rules, knowingly, because of specific demands by their limited audience. His lambastings must, therefore, be taken with a grain of salt--unless you're in the process of redesigning msnbc.com or yahoo.com.
Remember, the most used site, Google, violates several rules (scrolling, complex URLs, dense content, et al). And Google appears to appeal to everyone--how many people are confused by it? Further, how many average web users know what the 'Cached' link means? Or why there are two areas of 'Sponsored Links?' The web is necessarily relative.
taking the user out of the context of what he's reading bad as well?
He recommends supporting pages in the book. From a section on fundamental errors:
"Content authoring: writing in the same linear style as you've always written. Instead, force yourself to write in the new style that is optimized for online readers who frequently scan text and who need very short pages with secondary information relegated to supporting pages."
I prefer that myself as long as it isn't in a stupid popup, magic div, or new window. It takes me one click and one key press to visit a page and pop back as long as you don't use that navigation breaking crap.
@Stephen - Nielsen does not practice what he preaches (see article: "What is Usability" at fusability.com) but he certainly does have a paycheck.
David Janke seems to have hit the nail on the head here. Nice to see that someone gets it.
I was a Web Developer and a university qualified programmer before I moved into the world of usability. I have been involved in web design for 9 years now, and I have this to say:
"If you can't design an visually pleasing website while also making it usable, then you have no right to call yourself a designer."
Usability is central to, and one of the founding principals of, design.
Jakob is an interesting fellow - the issue of bad information design never seems to register in his research - what makes a web site unusable more than anything is a complete disconnect between user goals and the site content - bring those two closer together and you will be surprised how many of the cardinal sins Jakob espouses fade into oblivion...
#9760;#9760;#9760; Web designers who think they know better than me what size text I find comfortable to read, and whose web page layout doesn't cope when I tell my browser to override their ill-founded opinion.
I'm still trying to resolve the issue of opening new browser windows on external links. I like the idea of making it easier for folks to stay at my site but also explore links that I make available. But I dislike the invasiveness of forcing a new browser opening upon the hapless user.
" Also, in terms of writing for the web, it's important to put the most relevant information at the top of the article or page:"
In other words, *don't bury the lead*.
I still differ with a:visited colors. It was a suitable point when the same colors were applied to each single link on each single page
Interesting article, very good
I'm slightly more inclined to listen to Nielsen than Raskin, but only slightly. It's so easy to criticize when you only deal with the fluff at the top and don't have to care about the tradeoffs involved in real-world design.
Even if one ignores the dull and ugly aesthetics of useit.com, I think it's been accepted for some time now that you just don't use two-column split layouts on a web site. Columns are good for print media, not online content. That's three skulls for violating a web-wide convention. Oh... and it scrolls, that's another two skulls.
Just another preacher who criticizes anything and everything, thereby ensuring that he'll strike a chord eventually.