Visited Links -- Your Take
May 03, 2004 | 
33 Comments
Jakob Nielsen today handed down some more “wisdom”, this time about visited links. As usual he offers up a hard rule to base design decisions upon without much context or meat to back that rule up.
He says:
The damage that unchanging link colors cause is one of the most tricky usability problems to identify in user testing. On any given page, users seem to understand the links just fine. Users almost never complain about link colors, as long as they’re distinct from the rest of the text and reasonably legible. Life is good, or so it seems.
Observe carefully, though, and you’ll notice that users frequently move in circles. They’ll visit the same page multiple times — not because they want to, but because they don’t realize that they’ve already been there. Users will give up when they’ve tried most links in a list, even though there’s one link that they haven’t tried; if the links don’t change colors, users don’t realize that there’s only one unvisited link remaining.
I see the underlying logic in this argument, it sounds pretty solid theoretically, but I wonder how it truly translates in the real world. This sound quite a bit like assumption and frankly, I can see this could be looked at in various ways depending on the site. Again, context is important here, we need some specifics to relate this argument to. I’m not saying he’s wrong — just incomplete. As usual.
From what I’ve seen changing the state (doesn’t have to be color, IMHO) of a link that’s been visited in the body of content, under certain circumstances can be useful. However I’ve seen many sites that do this within their navigation structure and I question the value of that.
I really wish Jakob would provide more evidence or data to back up his assertions, and maybe show us some specific examples of how this works or doesn’t work. Be that as it may, I wanted to see what others thought of this, both as users and designers.
Do you find a visited link state helpful? How so? Do you use such a state on your one site? I’d really like to hear from anyone who has user tested for the usefulness of state change on visited links.
Filed under: IA and Usability
Comments
1. Andrei Herasimchuk said:
My answer is rather long-winded.
Posted on May 3, 2004 04:00 PM | #
2. eric said:
If there was some sort of consistent standard for visited links it might be handy, but especially across blogs there is a huge variance in styles. Especially confusing/annoying to me are the strikethrough visited links; why obfuscate the actual link just because it’s been visited? What if I wanted to go back? Bold links which turn normal on click are a bit odd, too, as they reshuffle the text following them (if they’re inline).
I guess what I’m getting at is that things like color change or underline style are preferable to major changes to the text itself.
Posted on May 3, 2004 04:03 PM | #
3. Keith said:
Damn, Andrei, that is a bit long. But it looks like we agree here and it’s nice to note that I wasn’t the only one who was a bit put off by Jakob’s “advice.”
Posted on May 3, 2004 04:04 PM | #
4. DarkBlue said:
I use colour to differentiate between visited and unvisited links on my website. But I have never performed any user testing so I don’t know whether or not it is helpful to my audience.
Personally, I would prefer use the same colour for both states - since I think the colour difference (once a few links have been clicked) looks ugly. But that’s more than likely down to my (lack of) skills in selecting colours.
As a user (on other websites), I certainly prefer to have the colour cue. That is, I benefit from being to distinguish between visited and unvisited links.
However, Andrei is right on the money when he challenges Jakob’s assertions based on the dynamic content of many of today’s websites. The whole link colour issue becomes blurred when the content is dynamic.
Because my content is dynamic and, as a user aid, I recently added a “Changelog” to my website. This lists the top ten additions/changes to the various aspects of my site.
My access logs have vindicated that inclusion. Since I added the Changelog, it has very quickly become one of the most visited pages on my site:
1) blog: 15.3%
2) index.htm: 11.36%
3) changelog: 8.94%
Almost 9% of my visitors look at the Changelog! This suggests to me that users do appreciate being able to determine what’s new (or revised) on my website and that, possibly, my use of colour on links is not helping them as much as Nielsen suggests it should.
Posted on May 3, 2004 04:48 PM | #
5. Chris Vincent said:
I think that visited states are critical to have in every part of the site except the navigation. If your navigation structure is so poor that a user can’t tell if he/she has visited part of it, you probably need to redesign it.
In actual content, however, they are most useful. Especially since I read a lot of blogs in this “echo chamber” of ours. It’s especially handy when a lot of the blogs I read link to the same page, and this happens pretty frequently. Of course, I can look at the URL in the status bar, but with a visible state, I won’t even have to hover over the text.
Which brings up a small irritation I have with MovableType’s new anti-spam redirection scheme: it breaks the visited state AND my ability to see where it’s going via the status bar. The URL should be placed in the title attribute or something so I can see where the link is going without actually clicking it. (And I’m not a dial-up user, mind you; it’s just annoying. Besides, my cable connection seems to crawl now and then, which I secretly think is a ploy by my ISP to promote their new “premium” service. :P)
Eric, I actually like the strike-through convention. What The Daily Flight does is even better, IMHO.
Posted on May 3, 2004 04:51 PM | #
6. Eric TF Bat said:
I’m not the sort to let old “Do As I Say” Nielsen dictate style, but I have to say I do find it useful to have visited links display in a different colour. I’m likely to make a change on my personal site (thank the gods it’s a Small Matter Of Templating) and advise that we do the same on the sites I develop for work. I’ll even quote the Nielsen rant^H^H^H^Harticle to my boss.
However, I think he’s missing one of his own earlier points, the one I call the Scruffy Trashcan Syndrome. Quick explanation: trashcans/recycling bins on your desktop are smart because you can move files to them and retrieve them if you did so by accident; this is a quantum leap beyond DEL *.* and UNDELETE (even when it works). Unfortunately, I like a tidy desktop. When I see a full trashcan, my impulse is to Empty Trash to make it neat. Thus the net effect of the trashcan is not to save me from myself, but merely to force me to do an extra couple of clicks. Result: I usually switch the Recycling Bin off in Windows, and if I delete something by accident, well… I learn not to.
Similarly, different-coloured links mess up a nice neat page. This drives designers nuts, I’m sure; it does me too. We need a non-ugly way to achieve the same usability ideal; from the look of his website, I’m guessing that’s not an argument Jakob is mentally equipped to deal with.
Posted on May 3, 2004 05:00 PM | #
7. Sage said:
I think Andrei hits it head-on… it simply matters what kind of data your site projects.
I’m currently building a an “online magazine” of sorts, and I’m considering adding :visited link states so that users know which articles they’ve already visited. I’m still mulling over it though.
Posted on May 3, 2004 05:08 PM | #
8. Mike P said:
“it simply matters what kind of data your site projects.”
So true. Today a client calls “Hey Mike, why do our links change in the list of search results when they’ve been visited? So-and-so doesn’t like it. Also, can we disable the right click?”
Yikes.
Anyway, the site has a searchable database, and on each result (10 per page) you can ‘request more info’ or click thru to see product details. A great place to use a different *color/font-weight/etc.* for visited links versus fresh ones.
I think - though I too have no data to support it!
Posted on May 3, 2004 05:44 PM | #
9. i1277 said:
I find specific colours for visited links to be very helpful. On some sites it saves me trouble (ill-spent time going back and forth). On sites that has lists with a lot of similar looking entries I will forget which I have visited and which I have not. Common sense, really.
However. Lots of different colours in the text hardly is the graphic designers dream. I find my blog to be easier on the eye without it -it can a bit of a struggle finding 3 type colours (text, links, visited) to fit in with the general colour scheme. I’ve still chosen to have seperate colours because I believe in this as a navigation principle.
But it’s a trade-off (“it depends”). On some kind of links the user will be likely to know what he needs without any colour hints.
If Nielsen’s dream of a fully standarized web (all blue and purple links, very little graphics etc.) was implemented, it would be a dream to navigate it. It would be a dead boring dream though, and one amazing thing about the web as we know it is that it’s the biggest information bank in the world but you don’t have to be a librarian to enjoy using it. The diversity of styles and the special “identity” that design can give a website is part of what attracts millions.
Posted on May 3, 2004 05:52 PM | #
10. pixelkitty said:
Instead of changing the colour of my visited links, I remove the underline. So the link is still visible and the same color as all the other links, it just isn’t underlined anymore.
Seems to work, people don’t get confused.
I don’t like the idea of changing colours for different link states - I think that is more confusing to a site visitor than anything else. Especially if they haven’t visited for a few days.
Posted on May 3, 2004 07:40 PM | #
11. Christiaan said:
I remember reading an intersting highlight of the BBC homepage a while back. Basically links on the homepage gradually became darker to highlight frequently used areas of the site.
More information I found here.
Also search the web for “theglasswall.pdf” for a fascinating 80 odd page read of the design processes the BBC used. Regarding their visited links…
Posted on May 3, 2004 07:58 PM | #
12. zinckiwi said:
I tend to stick with the same colour for all the link states, but change the saturation/brightness to indicate state. Usually:
link - normal colour
active/hover - a bit brighter
visited - a bit darker and/or less saturated
I figure that gets the point across, doesn’t “busy” up the page as much as wholly different colours, and also avoids any colour-blindness issue (as far as I know).
Posted on May 3, 2004 09:28 PM | #
13. Steven Streight said:
No one so far has mentioned User Tests as Keith requested.
I have not test links changing color vs. links not changing color.
But I have watched users in User Observation Tests struggle with trying to remember what links they visited, and what ones they had not yet visited, when trying to find information in the Task Assignment portion of a web site usability evaluation.]
Sometimes they would click on an already visited link because they forgot they already tried it…other times they said, “I’m sure the information must be here. I’m going to look again. It must be there. Where else could it possibly be?” and they revisited a link to no avail.
There may be exceptions to usability/design guidelines, but web norms, what users tend to experience at most other sites, should be kept in mind and conformed to, not for beautiful design, but for maximum ease of use.
So I basically agree that in many cases, this blue/purple link color change is a good idea.
The design vs. usability strife reminds me of how so much great design is found in music CD packaging, the covers and inserts. But a common annoyance, from a usability point of view, is how often the text is unreadable: too tiny and colors like orange text against lime background, or worse.
I know what generally causes this. From my experience in ad agencies, the designer thinks the text is just another design element, not a means of communication. So he/she reduces it to microscopic size, and uses reverse white against a silver background and proudly proclaims the finished piece a “work of art.”
Meanwhile, interesting or important information, like song titles, names of band members, lyrics, etc. are unreadable, or a pain to read.
Just remember, people don’t just adore or lovingly behold a CD insert or web site. They also use/read them.
Changing link colors is especially vital in long lists of articles, when users are doing research, or trying to quickly find important info.
Posted on May 3, 2004 09:57 PM | #
14. Stu said:
Well, as a person whose “Blogs” bookmarks folder has been growing at an exponential rate, I’d say that I find the :visited pseudo-class to be ridiculously helpful. Here’s two situations off the top of my head in which they have saved me a click:
The long and short of it is that differentiating visited links from links that have yet to seen the light of day (as far as I’m concerned) is immensely useful.
At least to me.
Posted on May 4, 2004 12:03 AM | #
15. Sam Newman said:
This is a highly contextual issue. If your webpage mirros what the inventors of the Web thought the Internet would be used for (namely displaying large amounts of scientific text) then showing link state is handy. The target of said links is static, it shows you what parts of the site you have read, it makes sense.
In a web application? Link state has no meaning - a user might perform several searches a day, all showing a change in link state does is make the UI look inconsistent.
You also get some grey areas - what if the link target is dynamic? The link state is supposed to be saying to the user “You looked at that”, but what if the page they looked at could of radically changed without changing its URL? Comments for example - you might of read the comments to this post when there were only 5. You go back, now there are 10 but the link state shows a visit. Will you remember that there were 5 before? In such a circumstance you should probably change the URL to show a change in state (perhaps an unused comment count in the URL) assuming your system allows this of course.
Posted on May 4, 2004 01:37 AM | #
16. Rimantas said:
I prefer some clue that link has been visited.
Especially if it the list of many links. I can live with pages having only few links, then it is easy to remember what was visited and what was not, but normally having visited links differ a bit is quite useful.
Posted on May 4, 2004 02:56 AM | #
17. Drew McLellan said:
I’d agree with the comment that main navigational links shouldn’t change state - after all these are not usually the destination, merely part of the journey.
To my mind, it doesn’t matter too much which device is used to indicate that a link has been visited, but it is important that there is some level of indication. One thing that makes the web so friendly is the ease of following links. You see a link, you can click on it and off you go - if you didn’t like it, you hit the back button and you’re back to where you were. What’s not included in that process is the bit that says, “hey, let me just make a mental note of the link I’m about to click, as I’ll need to know in case I come back to this page and want to click another”.
By failing to indicate the visited state of a link, the ease of use is harmed. Setting the visited state to the same color as the regular link color is (it could be argued) disabling an element of the browser’s inbuilt functionality, and is akin in that respect to disabling the right-click context menus.
Posted on May 4, 2004 05:59 AM | #
18. waylman said:
I bet we all can’t wait till next time when he tells us all what colors we should use. Thats just the thing, He seems to give absolutely no consideration to the ‘look’ of a site. Its all about function/usability. There has to be a balance somewhere.
While, visually identifying visited links has helped me in the way he describes at times, I see no need for such a hard and fast rule.
Consider tabbed navigation. The tab is visable on every page. The user should always know where he is in relation to the other tabs without much help. And generally, there are not very many links to keep track of when using tabs because there is just not enough room to fit that many across the page. Jakob’s arguments just don’t seem to fit here.
However, if we have 20+ links in a sidebar where the user has to scroll to see them all, the value of some indication of where he’s been becomes much more apparent. As others have said, it depends on the situation.
Posted on May 4, 2004 06:49 AM | #
19. Keith said:
Good comments all. One thing I’m surprised no one brought up was the obvious (to me anyway) accessibiliy problem of using color to convery any kind of useful information.
It’s one of the reasons why I also change the state (from bold to normal) of my visited links, which if you’ll notice I only do in the content area of my site.
After reading everyone’s comments I may make a change to show a visited state in the side bar as it seems like some of you would find that useful.
Posted on May 4, 2004 08:14 AM | #
20. Dave Scott said:
Good point about accessibility, Keith.
Also, what about color blindness? Using color alone isn’t enough I’d imagine. I don’t find too much value in visited link state, but I occassionally do notice it and find it helpful, especially on blogs who all seem to link to the same thing.
Posted on May 4, 2004 08:32 AM | #
21. Robert Lofthouse said:
Nielson only likes to cater for the technical “scientific” world, which is where he came from.
I have studied a lot of HCI (Human Computer Interface) and usability issues. I don’t think there is much use for the visited pseudo class selector. If someone deletes their cookies etc every night like me or has them switched off, then the next time you visit the same web site, the browser will think you have never visited the links. The behaviour state will go back to “link” rather than “visited”.
I believe if you have a technical site (such as this one), then there is no point in using the visited pseudo-class selector, because most people who are average (and above) at IT clear their cookies etc.
As someone else mentioned, there’s that small problem that some people can’t differentiate between the colours anyways, so it wouldn’t make much difference. I can’t remember about screen readers, but surely the colour of a link isn’t going to help people who use that technology.
Overall: There aren’t many benefits to using the visited pseudo-class selector, as there are so many things that render it useless.
Posted on May 4, 2004 09:33 AM | #
22. Steven Streight said:
Sam Newman brought up the good point that a main navigation category, like “Comments” should not have the visited vs. unvisited link color change.
If it did, the first time you visited something like “Comments” would be the last time. This is not what Nielsen means, as far as I can ascertain. But if there is a long list of comments, it helps to know which ones you’ve already read. Of course you know things like “News” and “Comments” will have dynamic content and archived items.
One thing to bear in mind, and it’s easy to forget, is that principle of Jakob Nielsen: your users spend most of their time at other people’s sites.
Except for your loyal groupies who spend all day every day devouring your site, the rest are busy at other web sites, for business, pleasure, or curiosity.
To say, “users should remember what links they visited” is to perhaps forget that users shouldn’t have to remember anything. They are speeding down the information highway, barely pausing to read or examine anything, they skim and scan and more like hunter/gatherers than studious scholars.
Some users will go to favorite blog and news sites every day, but many will visit a site more sporadically. How many sites to you visit religiously every single day? How many pages do you visit? How much time do you spend there?
Posted on May 4, 2004 11:45 AM | #
23. Andrei Herasimchuk said:
This is not news.
It is in fact this special case, and many other special cases where visited links make little sense, that make Nieslen’s Alertbox column inadequate as a piece of design or usability advice to take hook, line and sinker.
Posted on May 4, 2004 12:48 PM | #
24. Eric TF Bat said:
Joel Spolsky came up with a clever idea for comments on his feedback pages. The URL for a particular thread is something like “comments.cgi?thread=123&count=42”. (Actually he’s a tragic Windoze geek so he uses ASP, but the idea’s the same.) When someone adds a comment, the number changes, but in fact the CGI ignores the number when you click on the URL. The result is that the link colour reverts to a standard unvisited blue as soon as the linked page changes. You can see at a glance which threads you don’t need to read. Clever!
Posted on May 4, 2004 04:35 PM | #
25. Steven Streight said:
On the question of universal, ironclad design rules, are there any rules that apply 100% of the time to 100% of web sites?
If not, why get upset when an expert proclaims a generally valid principle, based on actual observation of users grappling with web sites?
It might be nice if all web design guidelines were accompanied by a non-exhaustive list of potential exceptions, so everyone would know that few, if no, rules are rigid, universal commandments etched in stone.
Are there any web design absolutes? One that comes to my mind is: a web site should be satisfying for its users, gratifying for its designers, and productive (in whatever manner that may be defined) for its owner/organization.
In most cases, web sites are not merely works of art to gaze at, but have jobs to perform.
A blog site is probably one of the applications that’s closest to being a free web art form, but even in a personal blog site, users want to navigate quickly and perform various tasks easily.
A Search Site function is often helpful on a blog site, and some more informative method of archiving past articles might also help.
Some blog sites archive by month/year, which is very nebulous, and by topic category, which is more specific.
Posted on May 4, 2004 06:51 PM | #
26. Keith said:
Steven – You might be interested to read my article The One Magic Rule of Web Design
The reasons people get upset at Jakob is because he does proclaim things to be rules that should be followed all the time and this in general his evidence, style and tone leave quite a bit to be desired.
That’s why he bugs me. He does have some good things to say at times, and his history is good as far as I’m concerned, but lately it’s been a little too much theory (and guru-tude) and not enough practical application.
Posted on May 4, 2004 07:23 PM | #
27. Andrei Herasimchuk said:
Nielsen is not stating his advice as a guidline or principle. He is stating it as a design rule that should be followed. As near as can be taken from the article, followed without exception.
You also bring up a point many usability folks need to pay more attention to, whether you realize it or not.
It is one thing to tell designers what users are grappling with and the problems they experience. I am always attentive to hear this sort of information. It’s good information. No one questions the motives of anyone who performs such a task or duty.
But it is quite another to tell designers what the design solution should be, especially in as emphatic terms as Nielsen does. It is especially problematic to do this when that person pays little attention to real technical issues with the solutions they propose, as they affect interaction, scalability of the design, and contextual issues depending on the task. (A sort of deafness to anything other than user problems is what I call it.)
My wife has a background in user research and usability. Then she learned more about design the past two years. She coded web sites, designed workflow, dealt with interaction, the whole nine yards. She now completely understands the problems usability professionals can cause when they operate in the manner of providing design recommendation while not understanding design both at the technical and aesthetic level.
You continue to bring up side points or issues that have very little to do with the actual debate at hand. No one has or is suggesting users should be burdened in their experience.
In this particular case, people have raised very valid reasons where using colors for visited links breaks down, and as such, breaks the rule the Nielsen generalizes and proclaims as solid design advice. Why is that so hard to acknowledge?
Posted on May 4, 2004 07:37 PM | #
28. Robert Lofthouse said:
The only “main” rules I stick to are:
Make the interface familiar - i.e Don’t put the navigation in the footer, or in the bottom left hand corner etc. That frame of mind comes from me doing HCI and analysing GUI’s.
Ensure the text stands out from the background (quite a simple design guideline).
and
Ensure the navigation on a web site is easy to find, also make the website easy to navigate. If you click a link that says something, make sure that “something” is there when the link is clicked.
I dislike anyone who thinks they are the “guru” of “insert subject here”, there is always someone who knows more than you. Nielson should just stop force-feeding his opinions. Then again, I never go on his web site to be affected by his opinions.
Posted on May 5, 2004 12:38 AM | #
29. Andy Smith said:
As a web user I certainly appreciate it when I can see that I’ve visited a link. As a web developer I sometimes wonder how many visitors will notice the difference, particularly those who don’t use the web as much as me, but I still think it’s worthwhile making the difference visible because it’s useful for visitors who do notice it and hopefully not harmful to people who don’t. (I think the only situation where it could impair understanding of a site is when people notice the difference but don’t grasp what’s causing it; since links change in direct response to the visitor’s actions, I think most people will understand what’s causing it).
What I often do is use a more muted version of the unvisited link colour with the same hue, somewhere in between the normal text colour and the unvisited link colour. Most forms of colour vision deficiency lead to problems distinguishing between different hues, so a change in saturation and/or brightness will hopefully still be noticeable. This also reduces the visual disruption you can get with completely different link colours.
I think there are exceptions, though. It depends on the purpose of the link - in HTML, it would probably be useful to distinguish between different types of link with class or rel attributes and use that to remove the visited link difference when it’s not so useful.
For example with links in a web application that perform an action, it’s often not much use to indicate whether the action has already performed. It may even be harmful if it makes users think they can follow the link again and just go back to something they’ve seen before, when in fact it will perform the action again.
Posted on May 6, 2004 08:00 AM | #
30. Colly said:
Ah, weeks after this thread died, I’ve finally managed to detail the method I use, which utilises background images and an alternative link style to produce ‘ticked’ results. It’s valid code and seems to aid usability.
Here’s the tutorial:
Ticked Off?Visited Links How-To
Hope it’s of use to someone…
Posted on May 18, 2004 02:01 PM | #
31. Colly said:
Ah, weeks after this thread died, I’ve finally managed to detail the method I use, which utilises background images and an alternative link style to produce ‘ticked’ results. It’s valid code and seems to aid usability.
Here’s the tutorial:
Ticked Off?Visited Links How-To
Hope it’s of use to someone…
Posted on May 18, 2004 02:03 PM | #
32. Colly said:
Hmm. First post and that happens. So much for “couldn’t find server”. Sorry Keith…
Posted on May 18, 2004 02:06 PM | #
33. Random Terrain said:
I hate web sites that don’t show that you used a link. There are a ton of them out there and many of them are supposedly professional web sites. Why would anyone want to hide which pages you visited? I often find myself going around in circles as Jakob Nielsen mentioned and when I visit a web site like that, I never return. I’m not going to waste my time on frustrating, poorly thought out web sites.
Posted on September 20, 2005 09:07 PM | #
Comments are now closed