Understanding and refining choreography for your information architecture

When I first heard the word ‘taxonomy’, I had no idea what it meant. I remember enthusiastically nodding my head at my boss about how awesome it is while frantically Googling it under the table. We’ve all been there early on in our careers. Although, what I found left me feeling even more confused — something about classifying animals? Whether you’re as confused as I was all those years ago or just in need of a refresher, this quick guide to all things taxonomy will sort you out.

‍

What is a taxonomy in information architecture?

‍

In information architecture, taxonomy refers to how information is grouped, classified and labeled within a shared information environment. The overarching structure of that shared information environment is the information architecture (IA) and we find our way around it using the navigation. Think of an IA as a house. The taxonomy determines which pieces of furniture belong in each room and we navigate around the house via doorways and hallways. It all fits together to create one shared environment.

‍

For a website architecture example, think of an online shoe store. The shoes might be organized and labelled by color, size, style, season or collection — that’s the taxonomy. The overall picture of where those groups of shoes live is the IA and in our pursuit of new shoes, we might navigate that structure via a navigation bar at the top of the page. In the amazing Grand Taxonomy of Rap Names visualization below, we can see how the information is categorized, connected and labeled through the lines and the colors. There’s no structure or hierarchy to it yet; that would be the next step in the process to build the IA.

‍

‍

Creating a taxonomy

‍

There are so many different ways to carve information up into a taxonomy and the key drivers for determining that are your content and, of course, your users. Your taxonomy needs to make sense to your users.You may be starting from scratch with a new website or you may have inherited a taxonomy that for whatever reason just isn’t fit for purpose. The first step when creating an initial taxonomy is to do a comprehensive audit of your content. Ask yourself, is your content relevant? Is it up-to-date? Is it all necessary? Are there opportunities to delete or condense content? Once you have your content sorted, you’re ready to move on to the next step of running a card sort with users.Running a card sort early in your taxonomy creation process will allow you to build it up from an evidence based foundation. There’s no point guessing then testing and potentially going back to square one, when you can co-create with your users and then test that informed approach to validate and further evolve your thinking.When you’re designing your card sort, you’ll need to decide if you’re going to do an open, closed or hybrid sort. Here’s a very high level look at what each type involves:

‍

• Open: participants sort cards into groups and name their own categories
• Closed: participants sort cards into categories determined by you
• Hybrid: participants sort cards into categories determined by you AND they can also make up their own.

‍

This early in the taxonomy creation process, it’s best to start out with an open card sort. Not only will this tell you how your users expect your content to be grouped, but will also provide insight into the language and labels that they would expect that content to be associated with. You never know, an open card sort may even surface something you hadn’t considered. At this stage of the process, it’s important to be open to ideas and new possibilities and an open card sort will do just that.Once you’ve settled on the type of card sort you’ll be running, you’ll need to test which can be done through a tool such as Optimal Workshop’s OptimalSort. OptimalSort enables you to run unmoderated card sorts remotely (or print out cards for a moderated/in-person card sort!). After your participants have completed your card sort, you can access the benefits of OptimalSort’s powerful result analysis functions.

‍

Learn more about running a card sort and more through our 101 guide.

‍

After you’ve run your initial open card sort with users, you should have everything you need to create the first iteration of your taxonomy. Consider everything you learned during the card sort and cross reference that with your business goals and any tech constraints you might be facing. Don’t stress too much about nailing it this time around — remember this is the first iteration and as you test more and learn more, you can make changes. Build out your taxonomy in Post-it notes with a team and then whack it into a spreadsheet to make future testing and iteration activities easier.

‍

How to test a taxonomy

‍

Now that you have the first iteration of your taxonomy, it’s time to have a go at structuring those groups into an IA and running a tree test. A tree test works like a card sort but in reverse — it allows you to test your thinking by working backwards. Optimal Workshop’s Treejack is an online tree testing tool that helps you assess the findability of your content without any visual design elements. All you need are clear objectives for what you’d like to learn more about and a spreadsheet version of your draft IA (told you it would come in handy! ).

‍

Learn more about Treejack and tree testing through our equally handy 101 guide.

‍

Another way to test your taxonomy thinking is to run another card sort. However this time, a hybrid or a closed card sort might be more suitable. A closed card sort would be useful if you’ve got evidence to suggest that your group labels are making sense to users but you’re not 100% sure what belongs in each group. A hybrid sort will let you go one step further and tell you if your content does in fact fit within those labels and if not you’ll also pick up some new ideas to iterate your taxonomy further.

Developing a taxonomy is like any other design process. Bring users into your process as early as you can and never stop iterating, improving and learning.Oh, and about those animals — I wasn’t entirely wrong. The way we classify animals (e.g., vertebrates and invertebrates) is a taxonomy. There are taxonomies everywhere and they’re not all digital. From libraries to supermarkets, we are immersed in taxonomies. It's the role of information architects to determine how these taxonomies are presented to us and how we navigate through them — the possibilities are truly endless!

‍

Further reading

• "What is information architecture" — a guide we created to help people learn more about information architecture, including stuff for beginners and those who are more advanced.
• "An introduction to taxonomies" — an article from UXBooth's Sarah Khan that details what taxonomies are, taxonomy displays and more.

“Questions are never indiscreet, answers sometimes are.”Oscar Wilde

‍

Frequently asked question pages. Love them or hate them, I don’t think they’re going anywhere anytime soon. This debate has been going on for quite some time and there is an equal number of opinions on both sides of the FAQ fence. Nielsen Norman Group’s Susan Farrell says FAQs can still add value to a website when done properly, and Gerry McGovern says FAQs are the dinosaurs of web navigation.

‍

So, how do we really know for sure if they will or won’t add value to a design? Like anything in UX, you have to test it! I don’t know about you, but I’m a shake-it-and-see-what-falls-out kind of UXer, so naturally I decided to run a Treejack study. Scouring the web one fine day, I came across Sainsbury’s Active Kids. Its FAQ page was unlike any I had ever seen and I knew I’d found the one. I was also curious to see how it would test with people who don’t use the website — after all, anyone should be able to use it. Since Active Kids is an active lifestyle program for UK schools and sports clubs, I recruited my participants entirely from the US.

‍

Pull up a chair and get comfy because what I found out should serve as a lesson to us all.

‍

Why Active Kids? 🤸🏼

‍

First of all, why did I choose this in the first place? The Active Kids FAQ page caught my attention for three main reasons:

‍

• structure
• labels
• content

‍

The structure of this FAQs page is quite deep, complex and very different from the rest of the site — almost like another information architecture (IA) had been built within the main structure. Imagine you have a large warehouse with hundreds of shelves, and then somewhere in the middle of it, someone builds a house — that’s how it felt to me.

‍

There are two ways to get to it: through the “Help” label on the top navigation bar and the “FAQ” label in the footer. It also uses a combination of drop-down filters that the user needs to apply, but it also has automatic filter options and confusing labels that can send you down a path you don’t necessarily want to take.

‍

I also found it very interesting that most of the information contained within the FAQs section cannot be located anywhere else on the website and most of this is essential to gaining a fundamental understanding of what Active Kids actually does. Adding to the house in the warehouse analogy, it’s like the house holds all the key information the warehouse needs to function, but no one knows which room it’s kept in.

‍

‍

‍

Setting up the study 🛠️

‍

Treejack was the perfect choice for testing the findability of information on the Active Kids FAQ page and I decided to test the IA of the website as a whole — this means both the warehouse and the house. I couldn’t just test the house in isolation because that’s not how a user would interact with it. The test needed the context of the whole site to gain an understanding of what’s going on. Creating a Treejack study is quick and easy and all you have to do is build the structure out in a basic Excel spreadsheet and then copy and paste it into Treejack.

‍

My next job was to determine the task based scenarios that my participants would use during the study. I decided to choose nine and all were derived from content located in the FAQs section and related to tasks a user might carry out when investigating or participating in the program. Once I had my tree and my tasks, all I had to do was set the correct answers based on where the information currently sits on the Active Kids website and I was ready to launch.

‍

Recruiting participants for the study🙋🏾

‍

In my experience,recruiting participants for a Treejack study is quick and easy. All you have to do is determine the screener criteria for your participants and Optimal Workshop takes care of the rest. For this study I requested 30 participants and they all had to reside in the US. I ended up with 31 completed responses and it was all over in less than two hours.

‍

Treejack results 🌲

‍

So, what fell out of that tree when I tested a website aimed at parents and teachers of kids in the UK with 31 Americans? I’ll be honest with you: it wasn’t pretty. Here’s what I discovered in this study:

‍

‍

• 81 per cent were unable to find out if home educators were eligible to apply (number 1 on the graph)
• 65 per cent were unable to find out what a Clubmark accreditation is (number 2 on the graph)
• 68 per cent were unable to find out how to share their wishlist with friends and family (number 3 on the graph)
• 64 per cent could not find the information that would explain the purpose of the £1 fee mentioned in the terms and conditions (number 4 on the graph)
• 97 per cent could not locate the information that would tell them if they could use a voucher from 2014 in 2015 (number 5 on the graph)
• No participant was able to determine if students from a middle school would be able to participate in Active Kids (number 8 on the graph)
• 58 per cent of participants in this study were unable to find out what the program is even about (number 9 on the graph)

‍

‍

On the flip side, 68 per cent of participants in this study were able to locate a phone number to contact Active Kids directly (number 6 on the graph) and 97 per cent were successfully able to work out how to redeem vouchers (number 7). Overall, it wasn’t great.In addition to some very useful quantitative data, Treejack also provides detailed information on the pathways followed by each participant.

‍

Understanding the journey they took is just as valuable as discovering how many found their way to the correct destination. This additional level of granularity will show you where and when your user is getting lost in your structure and where they went next. It’s also handy for spotting patterns (e.g., multiple participants navigating to the same incorrect response).

‍

I always set my studies to collect responses anonymously and when this occurs, Treejack assigns each participant a numerical identifier to help keep track of their experience without the participant having to share his or her personal details. For task 6, the paths chart below shows that participants numbered eight to 20 were able to navigate directly to the correct answer without deviating from the correct path I defined during setup.

‍

‍

‍

For Task 3 (below) , the story told by the paths was quite different. Participant number five navigated back and forth several times through the structure in their attempt to locate information on how to share a wishlist. After all that effort, they were unable to find the information they needed to complete the task and nominated to contact Active Kids directly. Not only is this a bad experience for the user but it also puts unnecessary pressure on the call centre because the information should be readily available on the website.

‍

‍

‍

Treejack also provides insights into where participants started their journey by recording first click data. Just like Chalkmark, this functionality will tell you if your users are starting out on the right foot from that all important first click.In this study I found it interesting that when looking for information regarding the eligibility of home educators in the Active Kids program, 42 per cent of participants clicked on “Schools & Groups” and 19 per cent clicked on “Parents & Community” for their first click. Only 6 per cent clicked on “Help”, which happens to be the only place this information can be found.

‍

‍

‍

‍

‍

I also found the first click results for Task 9 to be very interesting. When looking for basic information on the program, more than half (52 per cent) of the participants in this study went straight to “Help”. This indicates that, for these participants, none of the other options were going to provide them the information they needed.

‍

‍

‍

‍

What can be learned from this study? 🎓

‍

I mentioned earlier there was a lesson in this for everyone, and rather than dwell on how something tested, it’s time to move on to some lessons learned and constructive ideas for improvement. Based on the results of this Treejack study, here are my top three recommendations for improving the Active Kids website:

‍

Rethink the content housed in the FAQs section

‍

Most of the key information required to master the basics of what Active Kids is all about is housed entirely in the FAQs section. FAQs should not be the only place a user can find out basic information needed to understand the purpose of a product, program or service. I believe this website would benefit from some further thinking around what actually belongs in the FAQs section and what could be surfaced much higher.

‍

Another idea would be to follow the lead of the Government Digital Service and remove the FAQs section altogether — food for thought. Frequently asked questions would not be frequently asked questions if people could actually find the information on your site in the first place. Figure out where the answers to these questions really belong.

‍

If you’re using Treejack, just look at the fails in your results and figure out where people went first. Is there a trend? Is this the right place? Maybe think about putting the answer the user is looking for there instead.

‍

Restructure the FAQs section

‍

If you must have an FAQs section (and believe me I do understand that they don’t just disappear overnight! Just try to keep it as an interim solution only) please consider streamlining the way they are presented to the user. Ditch the filtering and display the list on one page only. Users should not have to drill down through several layers of content and then navigate through each category. For further reading on getting your FAQs straight, this Kissmetrics article is well worth a read.

‍

Review the intent of the website

‍

Looking at the Active Kids website and the results from this study, I feel the intent of this website could use some refining. If we come back to my warehouse and house analogy, the main chunk of the website (the warehouse) seems to be one giant advertisement, while the house (the FAQs) is where the action-oriented stuff lies. The house seems to hold the key information that people need to use the program and I think it could be displayed better. Don’t get me wrong, Active Kids does some incredibly good work for the community and should absolutely shout its achievements from the rooftops, however a sense of balance is required here. I think it’s time for the house and the warehouse to join forces into a solution that offers both rooftop shouting and usable information that facilitates participation.

‍

The value of fresh eyes 👀

‍

This study goes to show that regardless of where you are in your design process, whether that’s at the very beginning or a few years post-implementation, there is value to be gained from testing with a fresh set of eyes. I’m still undecided on which side of the FAQs debate I belong to — I’m going to sit on the fence and stand by the “if in doubt — test it” school of thought.

‍

Further reading:

• "Strategic design for frequently asked questions" - This ebook from Nielsen Norman Group explains how to design world-class FAQs.
• "Infrequently asked questions of FAQs" - An article from R. Stephen Gracey published on A List Apart, discussing FAQs and their place on the web.
• "Anatomy of a website: website architecture" - An article on our blog explaining website architecture and how to test your own structure.

What is information architecture (IA) without its complementary elements? On this blog we’ve already discussed the information architecture elements of taxonomy and ontology, and now it’s time to take a look at choreography. While taxonomy refers to how information is grouped, classified and labeled and ontology refers to the meaning behind our words, choreography describes the rules for how all those parts should interact.

‍

Choreography is essentially the nuts and bolts that bring organization, structure and meaning together to form one well-oiled machine that supports, enables and delights users on their path to task completion within a specific context.An often overlooked element of information architecture, nailing your choreography is essential to the delivery of a seamless user experience because it is present in everything your users will do on your website.

‍

Understanding choreography and IA

‍

Coming back to our house analogy from my taxonomy blog, choreography would be present throughout the entire house determining how everything works together. It’s in the positioning of the door handles and hinges to enable easy access and use of rooms and so you don’t swing an open door into a shower screen or into the knees of a toilet user (who should have locked the door anyway). Choreography appears in open plan living environments seamlessly flowing from one space to the next. The way that the laundry is usually a room that includes an external door allowing an easy workflow from storing the dirty clothes, to washing them to hanging them outside on the line to dry. These are all examples of choreography.

‍

In the context of a website, choreography is present in a number of different website elements. These might include: the rules for how a menu might respond or behave under specific circumstances, the way the user interface adapts to suit different devices, the hierarchical relationship between content and much more.

‍

Choreography examples

‍

To help you understand how choreography might appear in a website IA context, I’ve scoured the internet and have found 3 interesting examples I’d like to share with you.

‍

Mental Floss

Mental Floss is an online publication designed for curious minds — sounds awesome! But the thing I find most interesting is the choreography that is present in the surfacing behavior of the top navigation menu. There are two different ways to access its content: you can either click on the big orange menu button in the top right hand corner and you’ll get everything in one hit — including the footer links (see two images below) — or you can scroll down or up the page and it will automatically drop down and become sticky giving you just the level 1 IA headings and that sticky orange menu button (see third image below).

‍

‍

Mental Floss as it appears when I first open the home page.

‍

‍

When I click on the big orange button

‍

‍

Now look what happens when I scroll a bit.... Ta da!

‍

If you go with the scrolling option, you’re not going to miss out because Mental Floss appears to have been designed to be explored — might have something to do with that curiosity thing! All the links that live under the big menu button are distributed throughout the long scrolling page, so when you’re exploring the content, you also get everything! If you get really stuck, that sticky orange menu button is always there to help. If I scroll I get one thing and if I click I get something else, but no matter how I choose to consume the content I get access to everything without having to think about it. This website’s choreography supports both exploratory and direct navigation styles and brings all the parts together.

‍

IKEA US Inspiration page

‍

Part blog, part catalogue, part how-to hub, the Inspiration page on IKEA’s US website is a fascinating IA example because it exists outside the patterns used on the rest of the website. It’s like an IA within an IA. While the rest of the website has a wide rigid structure typical of large IAs, the Inspiration page (also titled ‘IKEA Ideas’) is more organic and is organized by content tags and is navigated by filtering (see below).

‍

The tag cloud-like thing displays the content tags in alphabetical order. Each piece can be assigned up to four tags and users can filter by clicking or just peruse the whole lot by scrolling. Upon clicking a content tag, the piece selection below is automatically updated to only include pieces with the selected tag.Choreography on this part of IKEA’s US website is present in the relationship and behavior of the content tags and how that affects the information on the page.

‍

Do nothing for 2 minutes

‍

Do nothing for 2 minutes has a completely flat IA and its choreography is present in the way it responds to user behavior and the way it presents visual and audio content. The website opens to a single page (see below image) showing an image of a beautiful sunset at the beach accompanied by the soothing sound of waves gently crashing and a group of seagulls fighting over a pickle someone tossed them from a McDonald’s cheeseburger (maybe that was me). It gives me a very clear instruction to “do nothing for 2 minutes” and presents a timer that immediately starts counting down.

‍

‍

In the centre of the screen in that sunlight created highlight are the the words “Just relax and listen to the waves. Don’t touch your mouse or keyboard.” It’s a bit hard to read and I didn’t instantly notice it but when I don’t follow its guidance, the timer resets and tells me to “try again”(see below image).

‍

‍

When I comply with its instructions, the clock ticks down and when there are just 40 seconds to go something magical happens — the wave sounds stop. It’s not jarring or surprising, in fact it actually felt like I dropped to a deeper state of relaxation. Like an experienced lead dance partner, this website’s choreography pushes and pulls the follower where it wants them to go. The rule here is: if I touch my mouse or keyboard, the clock resets, and if I follow the instructions, I get to access an additional layer of relaxation when the sound drops off. It forces me to take 2 whole minutes of out my day in a strict but supportive way.

‍

When I reach the end of the 2 minutes, I’m congratulated and shown an advertisement for a book to help me further explore this technique. I usually hate seeing things like that but I don’t mind in this case because it doesn’t come across as pushy and it’s already given me something.

‍

Refining the choreography of your IA

‍

When you break it down, choreography is about behaviors and relationships between all the pieces of the IA. It’s about which parts go together, what they do and how that fits in with everything else. Think content pairings in a card sort and the hierarchical position of content within the IA’s tree structure — why do those cards go together and why does label Y appear directly before label X? These choices are deliberate; it’s not just a case of “Oh let’s just put them wherever,” or “I’m going to whack a certain label at the very top because my boss told me because that’s where he thinks it should go.” Choreography exists throughout your entire IA and like everything else, all refinements must be determined by user research.

‍

When running a card sort, pay close attention to:

• card pairings
• cards that are consistently not paired
• the hierarchy of cards within each group (card order)
• the conversation between your participants (moderated card sorts only) for insights into the logic behind the pairings and hierarchical positioning

‍

Cards that are paired — or not — as well as their hierarchical placement can not only provide insight into your users’ taxonomical expectations but can also help you identify relationships (or lack thereof) between content and elements. Say you were running a card sort on an intranet and two cards labeled ‘Annual Leave’ and ‘Public Holidays’ were consistently being grouped together under ‘Human Resources’. What’s their expected relationship and interaction beyond the subject matter connection? Does one appear before the other? Is one linked to the other one’s page? Or are they together on one page and if so how do they interact? (if at all). For moderated card sorts, listen closely to the conversation for insight into this and be sure to ask your participants about anything you’re not sure of.

‍

If your IA is a bit further along in its development or you’d like to evaluate an existing IA, tree testing can help you understand the choreography that will best support your users.

‍

Keep an eye on:

• the location of the first click
• the pathways followed

‍

The pathways followed by participants in a tree test will help you determine the right sequence of interactions required by a user to reach their goal — the ultimate step by step flow to task completion. What order do the labels need to be in? And what lives underneath them? And again, in what order?Think of it like a path through the woods to a lake. Your IA’s choreography should enable your users to be seamlessly guided along a smooth pathway made of big stone steps. They shouldn’t be running into trees or bushwhacking to create their own pathways.

‍

When looking at a tree test, consider how many of your participants did not follow the pathway/s you defined as correct. Where did they go instead? What does the right path look like to your users? Also look at where the all important first click landed. If users start out on the correct first click, they are almost 3 times as likely to reach their goal. If your participants started out on a different first click, you’ll need to explore why that is. It could indicate ambiguity in the labels, it could also be a sign of an expectations mismatch, a hierarchical issue or it could be something else entirely! The tree test will help you identify where the issues are, but you’ll need to go and have a conversation with your users to understand why it’s happening.

‍

Unless more than say 80% of your participants achieved direct success in reaching their goal (meaning they never strayed from the big stone path that you defined) you’ll need to check these things anyway to resolve findability and usability issues. Choreography related insights are an extra thing you can pull from what you’re already doing. And of course always remember that any choreography related data has to be considered in conjunction with whether or not the labels are even correct. Choreography is just one piece of the puzzle; it sets the rules for how all the parts interact and isn’t going to be much help if the parts aren’t even right in the first place!

‍

Choreography in information architecture might be one of the most overlooked elements but it’s not hard to give it the time and consideration it deserves and your users will thank you for it!