Slashdot Mirror
Web 'Rules' Changing?
sempf writes "Lots of things have changed since we started this HTML. The IMAGE tag was a nice change, and multimedia with plugins like Flash provide a new look.
What interests me the most, however, is the change in two of the hallowed GUI 'Rules' - the three click rule and the 7 +/- 2 rule. The Three click rule (which states that any page in a site or function in an application should be accessible in three clicks) was just debunked by Josh Porter in an article called Debunking the Three Click Rule. The 7 +/- 2 rule states that a user should never be presented with more than 5-9 choices at any given point in the site or application. James Kalbach has an excellent article debunking that rule at Dr. Dobb's Journal.
Worried that there will be no more 'rules'? Never you mind - the Government has come up with New Rules for us to follow."
The thing about rules like "three clicks", is that they are based on the pre-bubble notion of buzzwords. That doesn't work anymore in the web design field. Now we have to provide tools that the customers want to have, and design stuff so that they can easily access it. Document trees, under the nice standards at w3 are what has really changed with the internet, and not to mention PHP, Perl and free db solutions like MySQL and the other guys.
If users are leaving after 12 clicks now, like it says in the article, that says something about the level of web-smarts of the average user. But what I see in these charts, is a kind of "split the difference" research insight.
For clicking, it's 50/50 that people will go on to get what they want. For the percentage of unsatisfied users, it's 50% who are unsatisfied, according to their research.
What they've said is: "Users weren't any more satisfied with shorter clickstreams than they were with longer clickstreams. The satisfaction of users doesn't depend on the number of clicks."
So that means that in the old days, people were getting used to the infrastructure of web surfing, and things that were far away were annoying people. Today, people are used to the web... some teens have grown up on it, and therefore people as a whole are used to it. Therefore, things like design style and presentation mean more than how far clicks are, and if they know they can get what they want by going there.
I don't think there were ever any rules, just common convention. I mean, any good designer should keep all of these things in mind because they make sense, even some of the ones from the government... In the end the only thing that really matters is that Al Gore invented the internet and we should do as the government tells us! I wish Bush could set some rules, that would really make my day.
I've just made the faux-pas of actually reading* the linked article that claimed that 3-click was debunked, and I don't agree.
/. karma. I promise I won't read the article ever ever ever again, so this should be a one-time problem for slashdotters, since obviously no-one else ever reads articles here.
The 3-click rule says info should be accessible within three clicks.
The article contesting this says they watched over 8000 user clicks, and most users clicked 25 times before 'giving up', when it appeared they were searching for stuff.
The gap that I see is in not more-deeply analyzing how the clicks of users related to depth-of-tree (i.e., 1-click from home, 2-clicks, or 3-clicks, etc.) or perceived website quality. It is possible that people spent 25 clicks wandering but resurfaced to 'home' several times in trying to find the proper 3-click path to their desired target.
My point is that truly debunking this concept would involve:
1 - looking for 'back to home' patterns in click streams.
2 - classifying users a few ways (Some people are too timid/stupid to use the 'back' button!)
3 - validating user satisfaction on usability of sites that honor/ignore the 3-click rule.
All the article does is prove that people are persistent, even in the face of crappy webpage design.
* - My apologies; I hope admitting that I read the article doesn't completely destroy my
Periodically, we hear about the rule of 7 +/- 2 from inexperienced interaction designers: Users can't handle more than 7 bullets on a page, seven items in a form list, or more than seven links in a menu. This has no evidence in reality - on the contrary. The psychologist George Miller's conclusions apply to what we can memorize - not what we can perceive.
Current research strongly supports that broad structures perform better than deep structures. Users can more easily cope with broad structures, they have a greater chance of getting lost in deep hierarchical structures, and new visitors are able to get a better overview of sites offerings from a broader structure.
read more: The Myth of "Seven, Plus or Minus 2"
I use it as a rule of thumb all the time.
The thing you need to think about though, especially on the web, is this:
It's not about having only 7 links on a page. It's about grouping. You can group links using colors, a box, a header or just placement.
The reason site maps are useless on most sites is because if you have a web site with a good gui, it is actually mentally cheaper to click a few times and wait for pages to load, than be overwhelmed by hundreds of links at the same time.
Will code a sig generator for food
Whenever I need information about a product or application, I very much appreciate having access to a PDF version. I can take it with me on my laptop when I'm in the field or at a customer site, and I can archive it on CD in the event that the product is discontinued (or the company goes tits-up, leaving me with the maintenance issues.)
I've noticed that many companies have taken to presenting product data as HTML-only. I find that annoying because, assuming I'm interested, the first thing I end up doing is printing the HTML page to a PDF file so I can archive it. Usually I need to futz with the page formatting before I get a useful output, and that futz-time costs me time and aggrivation. I'm not advocating that all content should be PDF'd, but I do believe it has substantial value. Balancing the amount of HTML and PDF content presented is the tricky (and subjective) part.
I think the 3-click study is inherently flawed, since they studied the results of tests where people were asked to complete specific tasks; naturally they would *work harder* to complete them.
Now analyze a bunch of random people, who are not privy to the study in their everyday web habits, and see how the 3-click rule holds up.
At least I can get to anywhere on Slashdot in ThreeClicks. :-)
Does anyone know if this was intentional design?
PS: I would get pretty pissed if I couldn't get to where I wanted on Sashdot within 3 clicks. This is more due to speed than clicks, though. Also, it does not apply as strongly to sites I use less often.
Don't waste your vote! Vote for whoever you want, unless you live in a swing state it won't matter anyways
But these rules were created before many advances were made.
For instance, you aren't likely to find everything you want at an online store in three clicks. If you are looking for jewelry or specialty blank CD media, you may get to your category in three clicks, but there are still a dozen clicks beyond that to see the full contents of the category.
I would be interested in seeing what kind of tasks users were asked to perform and rate their "three-clickability" (terrible term). Almost anything involving a store, inventory, or selection process voids that "rule" for the end result, but not for the category.
Perhaps it should be rephrased that the user should be able to get to any content-space in three clicks instead of a page.
Pricewatch gets you to content in two clicks.
Outpost has three clicks to content on the sections I checked - one click, really - two for refining.
ice.com has one click to content, and then two for refining.
Barnes & Noble has three clicks to content.
Even eBay has three clicks to content.
That what was all this school was for... to teach us how to solve our own problems. -- janeowit
That's why it's a good idea reuse the same graphics as much as possible on many pages of a web site, e.g., place a banner that identifies your company at the top of each page. Modem users will already have the graphics in their cache, and won't have to wait for them to load again.
What really frustrates me is sites like Apple's, where you can't even tell what's on the page or how to navigate it until you wait for a megabyte of jpegs to load. Thirty three-second clicks is heaven. Three thirty-second clicks is hell.
Find free books.
There are actually good reasons to both of those rules, and the 7+-2 article did a better job of mentioning these than the 3-click article did.
The research that the 7+-2 rule is based on has to do with short-term memory, not how many people can read through. The point of this rule is that if people are "browsing" when they come to the site, meaning that they are not sure what they are looking for, they have to look through all the options and choose one. If there are more options than they can store in short-term memory, they have to do multiple browses to find what they want. As an example, if the site has 20 links, and the most appropriate link is link #10, the person needs to browse the whole list once, ask themself if any of those were appropriate (which they may or may not remember), then rebrowse from the top for that choice, or start over. Since they might not remember even seeing an appropriate one, they may have to do this multiple times to move more of the list into long-term memory so they can analyze it better, or just make a choice that doesn't take into account all the options. If the list had been 7+-2 in length, they could have made that determination in their short-term memory much more quickly.
If, on the other hand, every user coming to your site knows what they are looking for and where it is, they can look through 100 or more links to find it and as soon as they see it, they will click on it. They are not browsing, but searching for a specific thing.
The 3-click rule is almost related to the above, and it involves browsing vs. searching. If a browser makes a choice at the top they feel is appropriate (again not sure if they're in the right spot), if they don't find what they're looking for in 3-clicks they probably determined they chose incorrectly initially and will back up and start again. If they have definite progress towards their destination, they will go dozens of links deep to find it.
A searcher who knows what they are looking for is more confident about their initial choice and will keep digging to find it. The 3-click rule doesn't really apply to them.
The 3-click rule is much more of a guideline, and should really be that they need to see progress to their goal after 3 clicks or they'll turn back. It was also created because you must have created a mess if someone has to dig through 25 steps to find what they're looking for; I would call that failed site design even if people were willing to go that far. The article referenced was generally pretty poor as far as a study goes, they didn't give any information about what these people were doing, if they knew what they were looking for, etc. It doesn't really prove anything, and certainly doesn't "debunk" the guideline, which is pretty much based on common sense.
Apparently he can't deal with the idea of a document that isn't contained within a single file. That's about the only archival advantage PDF has over HTML. Personally, I prefer the HTML.
If a job's not worth doing, it's not worth doing right.
Don't you have your google preferences set to retrieve the first 100 links??
.. I like to scroll up & down.
.. just let me scroll the fuck down dammit. Or AT LEAST display 150 results .. dont make me click "Next" for every ten results .. that's annoying.
I hate clicking "Next" to see more search results
Why do pages insist on having "Next" buttons for results
These are not rules, these are guidelines. Even the link that /. included in the article description is
http://usability.gov/guidelines/index.html
There is a big differene between a rule and a guideline. The web-site is from the National Cancer Institute and it appears they wanted to share some lessons learned with the community. I for one appreciate that they took the time to formalize their findings on how to make the web easier to navigate. Unlike some rules, there is an address provided if you feel they have missed something. See their about page: http://usability.gov/guidelines/about.html
'IMAGE' is not an element in HTML 4 (check for yourself). Maybe it should be. Maybe it should stand for inline, base64 encoded images. But it doesn't.
Makes you wonder when the submitter of the article last wrote a page of HTML...
I personally don't mind longer clickstreams, as long as they make sense!. What really gets me going are two things: Flash-only navigation, and pages/contents that make no sense based on your task at hand. Take Comcast's web site for example- a prime example of UI nastiness.
First off, if I don't have access to a flash-enabled browser, I can't do anything associated with my account, or locate any contact information. Even after I have access to flash, there's even more trouble. Instead of providing you a list of service contacts so that you can easily scan through and locate the one in your area, you first have to know which of five or so regions you reside in (county incorporated, county unincorporated, etc). How the hell am I supposed to know this? Who cares? I only want some help with my problem, and suddenly I've got a whole new issue to deal with.
Even worse, this is the exact same request screen that appears when you're looking to BUY service from comcast, so it almost looks like they're inter-mingling their support and sales functions- quite confusing, because you're never sure if you're on the right page. My only feedback in situations like this would be to hire a competent web designer/design firm who is well aware of UI issues, and come up with a better solution.
I teach my students that, but in the context of the number of major elements to have in a system. I also tell them 3-15 is the range to be in. My point is that a system should have that number of subsystems to be 1) grokable, and 2) sufficiently complex to be worth defining.
James Kalbach's article points out how poorly understood the "7 +/- 2" "rule" is in general, but he seems to ignore that since its publication in 1956 psychologists have learned quite a bit about this "limitation" on information processing abilities. His suggestions are old news on this front and, instead of debunking 7 +/- 2, confirm its importance.
Let's start off with an example from where the research was perhaps first applied -- telephone numbers (George Miller, the researcher who "discovered" this number, worked for Bell Labs). US telephone numbers, since 1947, have followed the 3-3-4 format: 3 numbers for the area code, 3 for the exchange and 4 for the line number. Add the 1 in front of any number for dialing long distance and you've got an 11-number sequence. Does this violate the 7 +/- 2 "rule"? Not really, for a number of reasons:
Given these factors, a local phone number can have a demand on your STM as little a 5 "bits" of data for a local call. Still, you might think that with auto-dial features of phones these days, does this format really matter anymore? Well, maybe not to the technology in our phones that stores the information for us, or to the telephone switching technology that accepts and routes and connects our calls, but if someone gives you a phone number to remember you'll have a much easier time of it if you at least recognize the area code, even if all you need to do is walk to the phone and dial (as opposed to memorizing it). That 3-3-4 pattern helps us cluster the data and retain it in STM longer than if we'd try to hold a ten-digit sequence without any clustering or recognizable pattern.
The point being that 7 +/- 2 is not a design "rule" that has anything to do with the underlying technology but, rather, how human brains work. Kalbach and others either have forgotten or never knew that the "7 +/- 2" pieces of info have nothing to do with what the technology can handle and everything to do with what one person can juggle in STM while trying to do something meaningful with that info.
Chunking or clustering data is something we do naturally, without conscious effort, to reduce demands on our information processing. Use of cultural conventions (like requiring the 1 for long distance) that everyone familiar with a task can learn can also reduce these demands. By reducing these demands, you can help people
They should be able to. If it's not exactly a standard, it's 100% standards-based. And if they can't, you can easily write a plugin or app to do it.
funny munging
Is this a sigs-optional kind of place? 'Cause I am totally down with that if you know what I mean.
You forgot the most important rule: Never assume people will view the site exactly like it appears on your screen.
I don't like webdesigners controling my browser, so I have Firebird set not to display blinking text, status-bar tickers, banner ads, or custom scrollbars; flash doesn't play unless I click on it; I have toolbar buttons to resize text; automatic pop-ups don't work, and links that are supposed to open in pop-ups go straight to a tab. There are lots of sites that look laughable (if not unusable) because the designer added all sorts of bells and whistles on the assumption that everyone uses IE with the defaults set.
Les Miserables Volume 1 now up with my reading of
The 7+/-2 rule can also be attacked by clumping. Arrange the data in blocks, create an obvious hierarchy, use multiple columns. A well organized navigation bar with 5 global destinations, along with a table of half a dozen main categories with a handful of brief sublinks, and a sidebar with four context-sensitive links... that's fine.
Similarly...
The three click rule shouldn't be based on the number of clicks spent on the site, but the number of clicks spent without making any obvious progress. If you reward the visitor with an indication that they're narrowing down the goal, and don't force them to backtrack unnecessarily, or let them backtrack easily when they have to, they'll keep resetting the "click counter".
It's not that hard to devise a site that'll do this, if you think about it. You do have to think about it, not just copy things you've seen without understanding what they meant.
For one example: A lot of sites have "breadcrumb trails". There's two kinds of these, one useful, one pointless.
If you take the "breadcrumb" analogy too literally, you track where they go and provide links back to previous pages they've been on the site. That's pointless, they already have that information in their browser's backlinks.
But if you think about how people are going to use them in combination with their browsers, then what you do is show them how they would get there from the base of the site... now the trail is a guide to related information... much more useful.
Then, add more cross references.
There's support sites I've given up on after half a dozen clicks because the search engine was the only index to the site, no connections from a document to related documents, and I could see I'd never find what I wanted following random searches. Others I've been happy to spend half an hour digging through because they were effectively cressreferenced... every link rewarded me and reset the "click counter".
Clustering, crossreferences, partial results, progress, rewards. That's how you apply these rules, and that's what you have to figure out how to measure to see if these rules are actually useful.
You don't. Big e-commerce sites are always improperly designed, and this is an inherent property of any site that has more than 343 pages in a strict tree structure. It's amazing that anyone manages to buy anything online; you'd think that the rules would keep them away from e-commerce sites.
Or, we could realize that, in order to make a good web site, you should just make a good web site.
One web site please, hold the suck.
Would you care to back up this claim? Alhtough it's not a counter-proof I know of two groups who have implemented from-scratch PDF renderers that are yet to encounter documents using undocumented features. The PDF specification is quite large (1000+ pages) and some things are relatively obscure so it can sometimes appear that documents are doing things not in the spec.
It sounds like the thing you like about PDF is that you can save a local copy. But you can do that with html, too:
wget -r -l0 http://site.com/
Even easier would be if site owners would provide you with a tar file, so you can save a copy with just one click.
I dislike having to download a HUGE PDF, and then load up my PDF viewer every time I want to look at the docs. Plus, all the page breaks are annoying, as are the sized-for-print fonts.
"It sure was strange to see something on Usenet about me that didn't involve Klingon gang rape." -- Wil Wheaton
She's talking about Dimitry Sklyarov. And yes, he was arrested for describing a process that could enable a program to read encrypted PDF files to blind users. I was there. I attended his talk. He stated specifically that it was about usability, not cracking. One guy got up and stormed out because Dimitry refused to tell him how to "crack PDFs".
Dimitry was arrested the next day as he was leaving his hotel room to catch a plane home to Russia. He wasn't even an American citizen.
If you are interested, email me (bford (at) eecs (dot) wsu (dot) edu) and I will send you a recording of his talk.
The point was "content". Most pages have no content to find so naturally users will be disappointed. I spend the last week looking for some very specific information and I would have been glad to find it _anywhere_, even if it was an annoying flash animation. But the content I was looking for was just _not_ there.
One factor I didn't see in the article is bandwidth. What does a "click" mean? Normally it means navigating to (ie transferring from the server to their PC) new content. As bandwidth has increased, which includes everything from server performance and internet infrastructure to the final mile, the delay until that new content is available at the client has decreased, meaning that clicks are lest costly time-wise now.
So as the penalty of clicking on a link has reduced, the tolerance to clicking has gone up.
This should be a huge factor in the 3-click rule, which I don't remember seeing in the article.
Dan East
Better known as 318230.
That a single set of guidelines will satisfy all users of all sites is, imho , ludicrous. Some people like web sites with "wallpaper" or patterns; me, I hate web sites with wall paper - what is the point of having text if you cant read it.
Now it is true that a lot of companies have a std layout, at least in my industry, you get to the home page, and there are tabs for home, products, technology, contact, investors. And that std format is a bit help, esp when they follow thru on the lower layers. But this sort of std format is a lot less important then clear english.
IMHO, what matters is the clicks to get what you want. If the site can get its message across with 1 choice, per screen, great; if you need 20, ok.
But the choices/page is far less important then clear layout and a little thought; e.g., if you have, in your hand, a product and the label gives a name and a model number, why does a search of the site turn up nothing ? Or the links to documentation are not on the product page (yes, virginia, some people are stupid enough to have separate, nonlinked tech info and product trees). Untill you pass some minimal level of intelligence, it is not important to worry about design criteria for clicks/page.I have even seen company sites where u cant skip the flash intro (true !)
I, on the other hand, hate PDF files. I would much rather have the same info in HTML files. And I can't imagine why you feel you can archive PDF but you can't archive HTML.
But, to me, the Government website was a mess... WAY too many headings on the one page for my liking... why couldn't they have the headings on one page which linked through to the sub headings? Or, to keep it in one page, click on the heading to expand out the sub headings?
Reducing the number of clicks to get somewhere just to reduce the number of clicks is ridiculous when the tradeoff is an actually harder to immediately grab site.
If you recall an earlier discussion here about ternary computing (base 3 instead of base 2) there is a scientific proof that the optimal balance between width-oriented menus (lots of choices at each level, decreasing the number of levels) and depth-oriented menus (few choices at each level, deeper levels) is to have e (~2.7) choices. Obviously you can't have .7 choices, but if the number of choices per level averages to e and you group your choices logically, you'll have a solid argument that your layout is optimal.