How to Build a User-Centric Information Architecture Without Endless Debate

Paul Boag

Creating a site information architecture is a dangerous business. Too often it ends being structured around organisational silos rather than user’s mental models. But there is a process that can prevent that from happening.

So many things can go wrong when creating a site’s information architecture. Often it turns into a political battleground between rival departments each seeking to have their area of responsibility highlighted at the highest level of the site.

Then, of course, there is the tendency for organisations to structure their information architecture around departmental silos. After all, this makes it easier to divide up the site between these related departments.

Finally, many of us cram our information architectures with jargon or product names that the majority of users do not know.

In short, organisations often ignore user needs when it comes to the creation of information architecture. Fortunately, it doesn’t need to be this way. We just need to start from a different premise.

1. Collect Questions and Tasks

Most websites begin life with the question “what do we want to say?” However, to ensure a site is user-centric, we should start with the question “what do users want to know?” This shift in emphasis will make all the difference in the world to both site architecture and content.

Next time you kick-off a project, start by gathering the questions and tasks that users wish to complete on the website.

Some internal brainstorming will be enough to get you started, but do not stop there. Also look at site analytics and in particular search terms that people use on your site. These will help uncover further questions users are asking.

Running a simple survey on your existing website is also a useful way for gathering insights into what users wish to do or what questions they need to be answered. You could also try looking at your social media channels.

A simple survey is a great way of collecting user questions and tasks.

Finally, running workshop sessions where you ask users to brainstorm questions or tasks they might have, will further expand the list.

The result of this research should be an impressively long list of potential things somebody might want to do, or know, when visiting your website.

However, not all questions or tasks are equal. Some things will be more important to users than others, and it is essential to identify which these are, as that will inform our visual hierarchy and site information architecture. That is where a top task analysis can help.

2. Do a Top Task Analysis

Top task analysis is a process developed by Gerry McGovern to help identify the things that users most care about on your website.

Gerry discovered that the majority of users only care about a small subset of the content that exists on your website. However, they often find it hard to locate that content because they are distracted by the plethora of other information. To avoid this problem, we need to identify the top tasks. The things that users most care about and want to know.

The first step in a top task analysis is to gather all of the potential tasks that a user might wish to complete. That is what we did in step one. Following that we need to reduce that list down to something a little more manageable.

Create a Shortlist

Creating that shortlist of potential tasks and questions is principally about removing duplicates and combining overlaps. Also, try and avoid things that aren’t a task or question. A typical example might be “knowledge base”, which is too broad to be meaningful.

Finally, keep your tasks universal. Don’t start adding tasks and questions specific to individual audiences at this stage.

Your final shortlist will vary in length, but try and keep it under the hundred item mark if at all possible. This is because the next step is to survey users to find out which ones they care about most.

Getting Customers to Vote

Typically this involves either posting a survey via your website and social media or sending out an email. In either case, you are asking participants to rank their top five tasks by giving five to the most important one, four to the next most important and so on.

Getting users to vote on their top tasks helps inform the information architecture.

At face value, this seems like madness. Asking users to select between potentially a hundred different options and rank them seems like too much. In a sense you are correct.

The reason that this approach works is the very fact that it overwhelms users. There is too much information for them to go through each result individually and assess whether it matters to them. Instead, they are forced to ask themselves what they want to do, and then scan the list to find it. That prevents the tasks from influencing them too much.

What you will find when you look at the results is that a small proportion of the tasks gets the vast majority of votes. These are your top tasks, and they are the elements that you will build your information architecture around. They are also what we will use when we carry out a card sorting exercise.

3. Carry Out Open Card Sorting

With our prioritised list of tasks, it is now time to start structuring an information architecture around them. But instead of us trying to do this we are going to involve the user in the process through a card sorting exercise.

If you haven’t done card sorting before, don’t worry. It is very straightforward. The first step is to print out no more than 30 top tasks on separate index cards. Use the top 30 from the list you created in the top task analysis.

In a card sorting exercise, users organise information into stacks that make sense to them.

Next, we are going to ask users to sort the cards into groups. They can sort the cards in any way that makes sense to them. You can do this online using a tool like OptimalSort, or you can do it in person with people individually or in groups.

Once they have sorted the cards into piles, ask them to label those groups with whatever titles makes the most sense to them. That will give you some inspiration for the labelling of sections in the information architecture.

Regarding how many people you do this exercise with; well that depends on available time. Even doing this activity with five or six people is better than nothing, but the more people you survey, the better.

An open card sorting session won’t leave you with a perfect top-level information architecture, but it will leave you a lot more informed about how users view things and give you some ideas about how sections could be labelled.

4. Create an Information Architecture and Validate It

With the open card sorting done, you can now create a draft information architecture based on what you have learned. That is not going to be an exact science, but the open card sorting exercise should help you avoid the common pitfalls of making the information architecture to organisation centric. It will also give you some ammunition to combat any political pressure.

Where possible, use the labelling and groupings created by users in the card sorting exercise, but don’t feel constrained by it. You will have more tasks and questions to organise than you showed to users, so some changes will be required.

A tool like OptimalSort can help run both open and close card sorting exercises.

Once you have created a draft information architecture, we can now validate its effectiveness with users through another card sorting exercise, but this time we will take a slightly different approach.

Instead of our previous approach we are going to do a closed card sorting exercise. That differs from the previous approach we used.

This time we are going to print out more cards. Feel free to print out as many as 60 cards, each with a task or question printed on them.

Now we are going to ask the user to organise the cards again. But this time we are going to ask them to arrange those cards into a series of groups you pre-defined at the start of the exercise. Each ‘bucket’ will relate to a top-level section on the website.

Closed card sorting encourages users to organise content into pre-defined categories.

By doing this, you can validate whether your top-level architecture matches the user’s mental model. If the users put their index cards in the same groups as you intended, then you know your information architecture works. If they don’t, you know you have some work to do.

A User Centric Information Architecture

Admittedly the process I have outlined here will take longer than sitting down and taking a guess at the best approach. But in the long term, it pays dividends. Not only will it save time in endless debates about what should appear on the top level of the site, but it will also ensure users can find answers to the questions they have about your products and services.