Alphabetized Links are Random Links
February 6th, 2006
A bit ago, after my post, Determining Link Order on Intranet Portals, Sarah S. posted this reply:
I disagree that links should be ordered by importance. Importance to whom? To Tom “grape” is more important than “apple.” To Sally “banana” is more important than “apple” or “grape.” In a company of 12,000 employees, we order alphabetically on the intranet because we were constantly being told what was more important than the first thing on the list. Importance is determined by your role in the company and everyone’s role is slightly different. The one item that breaks this policy is the link to the timesheet, which everyone must fill out daily. It is out of order alphabetically on its list because it is most important to everyone. We also offer a portal interface that let’s users have a window of their favorite links, because everyone’s are different. Unfortunately, democracy didn’t work for us in our intranet design. We order alphabetically (in most cases).
This got me thinking. Is it that links shouldn’t be ordered by importance? Or that Sarah couldn’t figure out a way to establish importance for 12,000 employees? (It is interesting that she pulled the link to the timesheet out of the alphabetical group. If she really believed that alphabetical was the best order, she certainly would’ve left it in. So, there’s definitely inconsistency in her thinking regarding these shoulds and shouldn’ts.)
I completely sympathize with Sarah’s position. She’s got 12,000 employees, each with different jobs and priorities. How does one organize links under those conditions?
One option would be to completely randomize the links. You could put all the links into a hat and pull them out, one at a time. The first pulled link would go on the top, the second would go next, and so forth.
Of course, users would have to read every link to find what they want. There would be no way for them to eliminate entire groups at a time. If a user is looking for the link for registering a visitor to the building, they couldn’t easily eliminate all the links that have nothing at all to do with that function, such as the HR links.
Over time, users may remember that their favorite links are in certain positions. For example, they may remember a link that is near the bottom or towards the middle.
But, I think we can all agree that this wouldn’t be ideal.
The next optimization would be to throw away the hat and choose alphabetical order. That would be much better, right?
Well, only if the users knew the wording of the links. The exact wording. This works well for people’s names, states, cities, car models, and sports teams.
Cities, from MyTravelGuide.com, in alphabetical order.
Cars, from Edmunds.com, in alphabetical order.
Where it falls apart is for things where users may not know the exact wording. In that instance, they must resort to the same behavior they needed when we ordered the links based on the hat. They must scan every link to make sure they can see what is relevant and what isn’t.
Cosmetology topics, from DelmarLearning.com, in alphabetical order.
On the Delmar Learning site, which provides training materials for trade schools, the sections are alphabetically ordered. As a result, the related links, Barbering and Professional Barbering are scattered through the list. As are Business, Career Success, Instructor Training, Reference, and School Programs. The user is forced to look through all of the links, just like in the pick-from-hat scenario.
Usability and Accessibility resources, from www.alvit.de/handbook, in alphabetical order.
Alvit.de’s useful list of usability and accessibility resources (albeit missing a critical one) scatters the accessibility topics throughout the list, instead of in any sort of coherent group.
Ordering links by alphabetical order helps the designer because it’s an easy way to make a problem go away. But does the problem really go away? Is Sarah doing her users a favor by alphabetizing (all but one) of the links? Or is it a case of making it equally unusable for everyone?
Is importance the only way to organize things? Certainly not. If Sarah truly can’t determine a unique order of importance, I wonder if she can start grouping into small categories.
Labor statistics, from BLS.gov, not in alphabetical order.
The folks over at the Bureau of Labor Statistics don’t bother with alphabetizing the large number of links on their home page. Would alphabetizing all the links in one large list make this page more usable? I don’t think so.
Instead, they use a divide-and-conquer approach by breaking each set of links into small topic groups. Once broken up, it doesn’t matter what the order of the links are. They can list them in random order, list them from most-to-least popular, or, if their heart so desires, put them in alphabetically. All basically have the same effect. And the design seems to work well for their users.
The moral of the story: Unless you can be absolutely sure that users will know the exact terms in your list, alphabetical order is just random order.
Tweet
February 7th, 2006 at 4:43 am
I’m totally with you on this. It’s the difference between known and exploratory information seeking. If the user absolutely knows what they are looking for and how it will be described, alpha is OK. Otherwise, there needs to be some grouping. I’d even suggest that the heading of the groups isn’t so important, but having the right things together is.
February 7th, 2006 at 2:09 pm
An interesting way of looking at an age old problem, addressed with an age old solution
Card sorting!
or have I missed something here?
February 8th, 2006 at 7:24 am
As you have identified on SIGIA, McMaster-Carr’s list of links (www.mcmaster.com) is user-friendly. They are not alphabetical. They are grouped like in BLS.gov. From my analysis, the ordering within the group is by incremental/logical progression which makes for easy searching. The first link could possibly be chosen by importance.
February 8th, 2006 at 12:58 pm
Alphabetized ‘nuggets’ are THE way to go.
February 8th, 2006 at 1:18 pm
Pauric: Card sorting won’t tell you how to order things, once you’ve divided them into groups. (And if you haven’t decided to divide them into groups, you won’t think to card sort.)
Ryan: Why do you think that alphabetized nuggets are the way to go?
February 8th, 2006 at 2:42 pm
I would very respectfully disagree Jared. Well, let me first say I tend to bend the tools to meet my needs at a given time. There are two interfaces I work on. First is a functional noun-verb configuration style interface. Depending on the product we can have anything from 40 features (nouns), each with their own subset of verbs (Summary, detail, modify etc.)
I do a traditional cardsort to make the high level grouping (setup, maintenance, device etc) and alphabetical list the sub items/nouns/features.
The second interface is more of an informational tool, intranet. Created the cards with frequency of use label/tag taken from stats and cross checked with user requirements and tasks (yes, room for error). Traditional Car sort group them and then prioritise the sub group list on frequency. Or any other criterea of your choice.
My point is, there’s a lot more room on that card for tags other than the feature.
But, I do get the feeling I may be missing some part of a bigger picture. I only work on a handful of ui’s but I struggle to see why this approach would fail with something like the the SIGIA-L list, they seem to have not spent enough time thinking about nomenclature and purpose of each category, and just dove in to content.
February 8th, 2006 at 6:34 pm
Jared: Returning users should not be required to scan a long list of options. Breaking that list into sections specifically tailored for different groups will speed repeat visits and simplify the information for new users.
February 8th, 2006 at 9:08 pm
Pauric, I don’t quite get what you are doing. Do you do card sorts with end users or just for yourself to determine groups. I can’t imagine users coming up with summary, detail, modify 😉
And for your intranet, what do you mean by you prioritise on frequency? Frequency of what? How frequent your users put two cards together? I hope not – this would be a particularly poor way of figuring out what should be at the top of a list. All it means is that people think those two items belong together, not that either of them belong at the top.
Jared – I think you could make some inferences about priority from a card sort. Cards that people quickly put together may be more likely to be important, closely related and belong towards the top of the list. You wouldn’t get this from the output of the sort, but from watching as people go through it. But this wouldn’t be a strong measure.
February 8th, 2006 at 10:25 pm
Good points Donna.
I should have explained that I work in closed develoment without the ability to draw real users in, I use personas as well as my knowledge of our products to card sort by myself. I know I know (o; but I’m an ex user, ex developer, ex tester so I feel fairly confident in understanding user’s needs for the stuff I work on.
Thats my main work, the intranet is very infrequent but I measure frequency of use averaged across the users, most popular pages then go to the top. Start with an alphabetical list and release note stating that the menu system will be revised in one month to make the most popular elements easier to access.
February 8th, 2006 at 10:59 pm
Ryan: I get the breaking up into groups part. I was just curious why, within those groups, you felt that alphabetization brought anything to the table?
Donna: Pauric is right that you have a lot of room on the cards. When doing card sorts, we also gather frequency and importance information, as I described here. But, this is a variant that I don’t see discussed when people talk about how they do card sorts.
February 9th, 2006 at 5:44 am
Thanks Pauric for clarifying. It is great that you do try to do the best for your users, but this is not really card sorting. You could do it just as easily in your head 😉
Jared: I do collect frequency and importance in some card sorts – sometimes on the cards, more often as a part of a debrief discussion.
February 9th, 2006 at 7:46 am
Donna, you’re right. I have called it card sorting for lack of a better term. All the individual features were grouped, the groups were given names (and yes.. ‘system’, ‘device’ etc are user driven beleive it or not!) then within these groups you can sort on relevence, frequency, alphabetical or any combination. Although I avoid relevence as it causes the odd ‘debate’. Thats just extra information that can be put on cards. ‘Card sorting’ to me as I play the user.
Can this be done in your head, yes. As I’m not working with real users why go to the effort? Some very good reasons I wont bore you with. But I should have been more careful using a term that means something else to ui prefessionals – thanks for picking up on that.
February 9th, 2006 at 9:32 am
Ryan: I agree with your line of thinking. “Alphabetized nuggets” are an efficient way to organize information, especially if the nugget contains >7-9 items.
For example, it’s faster for users to scan this list:
Computers
Furniture
Gifts
Jewelry
Kitchen Appliances
Music
Pets
Phones
vs. this list …
Phones
Furniture
Music
Kitchen Appliances
Computers
Gifts
Pets
Jewelry
In general, I’ve discovered that users accomplish their goals faster if they readily understand how the “nuggets” are organized (either by price, by location, alphabetized, etc.). So, if “alphabetized nuggets” are not appropriate for your site, then use other methods for communicating the site’s structure quickly and effectively to users … because users will want to know.
February 9th, 2006 at 10:25 am
Agreed, if the brain is prepared for an alphabetical list then it will start searching based on those rules.
Using your example list above if the user was looking for a Necklace the eye first trains in on N > miss: no such item > Scans the list filtering out items not related to Necklace > user will either stop on jewelry or continue scaning and return > click and repeat steps if necessary
It would seem to me that this is most appropriate for interfaces where the user enters at the top level of the menu looking for an item buried within, with no knowledge of the path. Would it be a generalisation to say this is likely to need seperate search functionality as good practice? If so then this brings in to question the choice of a menu system which may require search processing on behalf of the user when there is search functionality on the system.
For example, how often do you drill down through the menu on amazon versus enter a search item. I just tried drilling for ‘toaster’ and gave up.
I’m making some sweeping generalistions. There are times when alphabetical nuggets are the only option, windows for one.
February 9th, 2006 at 12:50 pm
Jarod & Bob: Bob’s example is spot on. Grouping content and then leveraging an additional and instantly recognizable convention of organization helps the user sort information faster.
February 9th, 2006 at 2:43 pm
I kind of share Ryan’s view too. A blend of functional grouping and then “Alphabetized nuggets” for long lists would probably work well.
I didn’t find many examples of alpabetical listings on sites like Target.com or Kmart.com, instead there’re usually two tiers of links in the left navigation : the functional groups then the subcategories, all ordered by criteria not so obvious to me.
Target
Home
Furniture
Baby
Kids
Women
Men
Electronics
Sports
Toys
Entertainment
Kmart
Apparel
Baby & Kids
Jewelry
Electronics
Toys
Health & beauty
For the Home
Sports
Take Target for example, under “Home” category we have:
Bedding
Kitchen + Housewares
Patio + Garden
Home Decor
Lighting
Rugs
Window Coverings
Slipcovers
Appliances
Bath
Dining
Home Improvement
Health + beauty
Pets
See All
For such long lists, wouldn’t it be nice if both tiers of links are listed alphabetically? I never succeeded in remembering the positions of these links because they’re all different across websites..
February 9th, 2006 at 10:53 pm
Any suggestions on how to organize a category list like the following on the home page of an antique show related website?
• Advertising & Displays
• African Art
• American Art Pottery
• Americana
• Architectural & Garden
• Art & Paintings
• Art Glass
• Baskets & Woodenware
• Black Americana
• Books & Manuscripts
• Bottles
• Brass-Copper-Bronze
• Breweriana
• Buttons & Smalls
• Cameras
• Cast Iron
• Ceramics
• Children’s Items
• Civil War
• Clocks & Watches
• Coin Op & Mechanical
• Coins
• Comic
• Costume Jewelry
• Country
• Deco & Decorative Arts
• Decoys
• Dolls & Accessories
• Eclectic
• Ephemera
• Fiesta & Depression
• Fishing & Sports Related
• Flow Blue & Majolica
• Folk Art
• Forged Iron
• Furniture
• Furniture Country
• Furniture English/Euro
• Furniture Oak
• Furniture Painted
• Furniture Period
• Furniture Victorian
• Games, Toys & Trains
• General Line
• Glass, China & Crystal
• Guns, Armour & Arms
• Hats & Hat Pins
• Jewelry & Gemstones
• Kitchen Related
• Lamps & Lighting
• Lithographs & Engravings
• Maps
• Marbles
• Marine & Nautical
• Maxfield Parrish
• Medical
• Memorabilia
• Metalware
• Military
• Miscellaneous & Other
• Musical
• Native American
• Optical
• Oriental / Orientalia
• Paper
• Pewter
• Photographica
• Postcards
• Pottery & Porcelain
• Primitives
• Prints & Posters
• Quilts
• Radios & Phonographs
• Rugs-Oriental & Indian
• Scientific Instruments
• Staffordshire
• Steiff & Teddy Bears
• Sterling & Silver Plate
• Stoneware & Yelloware
• Tools & Hardware
• Transportation
• Victorian
• Vintage Clothing
• Vintage Textiles & Linens
• Wallace Nutting
• Western Memorabilia
• Wicker
• Wrought Iron
February 10th, 2006 at 11:28 am
You have a mixed batch of very high level grouping and low level descriptions. E.g. I would advise against both Western Memorabilia and Coin being at the same level in a menu tree.
I would revise the names of some of the items, e.g. Fishing and Sports related . Even though you may have more fishing related items than the rest of sporting items combined, fishing is part of the sport group. Sports and Fishing would be a slightly better name.
I would try grouping items such as Wrought Iron & Cast Iron together, find a general name then start grouping the general names together to build a hierarchy.
For fun, you could build a yes/no search engine based on 20 questions (o;
February 11th, 2006 at 5:30 pm
Are web sites getting smart enough to be able to look into statistics on a given link and say this month – link A is more popular or has more clicks than link B so deserves more attention?
Now this may create a whole series of new problems potentially. But its the equivalent of moving old stock to the back of the store and moving the stock that is moving well up front.
February 13th, 2006 at 2:15 am
Daniel –
Yes, it is very easy to look at link statistics and have your website modify itself accordingly, but in my opinion this is a bad bad idea. Like you’ve hinted at, this creates a problem for returning users looking for consistent navigation. One unbreakable rule, if there ever was one, is to keep your navigation consistent – and that means across pages and across time.
Changing it up every month (or any time period really) is going to drive your returning visitors up the wall.
February 13th, 2006 at 2:48 am
Sorting Lists 101
No, not that kind of sorting. We’re talking about how to sort lists of links (or any objects, really) such that users can find what they’re looking for faster, better, harder, stronger. Jared over at UIE has a great article on how to arran…
February 13th, 2006 at 1:42 pm
There’s another reason why you want to be careful when using site analytics to make design decisions:
Are the analytics showing a link is unpopular because nobody wants it or because nobody can find it (or understand it)?
If the case is the latter, then you may just reinforce the problem by making the stuff that’s already easily findable more easy and the stuff that isn’t more difficult.
February 14th, 2006 at 4:02 pm
Brian –
Consistent navigation is important in so many arenas. I live and work in a rural area of a small state where a certain Super retailer is king. The most frequently heard remark about said retailer is, “I hate shopping there. They change stuff around all the time, and I can never find what I’m looking for.”
While that remark is not necessarily true, the perception is. Enough displays change on a frequent basis to make shopping more frustrating than rewarding. Here, there are few, if any, choices and shoppers are locked in. Traveling 35 to 65 miles for basic household items makes the cost of change too high. On any website (at least, any website that is not 100% unique), the user’s cost to change is negligible, navigation must be consistent, and users need advanced notice for any major changes.
February 14th, 2006 at 4:42 pm
To address those design concerns
1)Unless you have an extremely large and ill-designed menu system, any one user’s inability to find a particular item an an alphabetical list would be compensated for statistically across all user interactions.
2)A review of search items, if such functionality exists on your interface, would highlight items users were looking for that existed in the menu tree. Thus flagging an issue with that item’s placement.
However… I’m begining to question the usefulness of developing and implementing some form of Importance list, no matter what way it was arrived at.
The reason you would create one is to reduce the user’s time spent looking down long lists searching for best matches. Alphabetical lists work when the user knows what they are looking for, no matter the size of the list. If they are searching for best matches then why ask the user to do the hardwork when you can do it for them. Search functionality better addresses their goal than putting what you think might be what they’re looking for at the top of the list/tree. To my mind that’s a type of recommendation system without any understanding of the user’s specific goal or needs.
Amazon combines Alphabetical, Importance and Search paths to products. Of the three I find the Importance list “See all 32 categories” the most time consuming interaction. For direct product look-ups, where I knew exactly which item I wanted, the Search was the quickest. For browsing I found a combination of Alphabetical list drilling followed by Search as part of a single product look-up to be the most efficient.
After some thought I feel this negates the need for anything other than an Alphabetical menu. That is, in future if I build something where I find users hunting around a menu tree I’ll emphasize the search functionality.
Here’s a question for the UIE team. At what level of interface complexity (groups & items or breath and depth of menu tree) do you feel alphabetical listing becomes random search for users who do not know exactly what they are looking for?
February 14th, 2006 at 5:58 pm
On an intranet, the system has a great deal more information than on the Internet (at least for the moment). A user logs in to an intranet, so the system most likely has the workgroup and role information about that user. It would be possible to start with an arbitrary list (possibly in alphabetical order) and create smaller personallized groupings for each workgroup and another for each role based on usage patterns.
Links can appear in multiple listings.
Such quick links sections and a search facility *might* be a good solution.
February 15th, 2006 at 12:57 pm
The only hitch I see with personalised importance is cross user group consistency.
– Annoucements about new features… “All, we have added a tools section to our intranet. Accounting, you can find yours under Financial>Planning>Software Tools. Sales dept, your new tools are under Go to Market>Toolkit. ” Plus supporting this might be an issue, new user calls the helpdesk looking for something that is group specific.
I was thinking of a more static architecture. A high level grouping with general links bringing the user in to Summary pages, I guess these pages could have Importance or Recent listing (NOT menu items) and for anything that wasnt in immediate sight you provide search & browsing.. but definitely do not try to provide a menu tree right down to individual pages if your site is large.
I like the structure of http://www.dmxzone.com/ with General grouping across the top, some verbs in the top left. Search functionality in both places. A main summary on important announcements and another Grouping/Recent Items list on the RHS. Its a little overkill but extremely efficient > ‘fast’.
February 25th, 2006 at 10:46 am
[…] […]
February 28th, 2006 at 7:54 am
Jared Spool, thanks for your work. I’ve added your resource in my catalog. A link will be online after the next update.
November 1st, 2011 at 3:39 pm
jewelry box…
[…]Alphabetized Links are Random Links » UIE Brain Sparks[…]…
December 9th, 2011 at 3:18 am
Ravinder Tulsiani (Family Budget Articles), Ravinder Tulsiani (Family Budget Blog)…
[…]Alphabetized Links are Random Links » UIE Brain Sparks[…]…