Online card sorting: The comprehensive guide

When it comes to designing and testing in the world of information architecture, it’s hard to beat card sorting. As a usability testing method, card sorting is easy to set up, simple to recruit for and can supply you with a range of useful insights. But there’s a long-standing debate in the world of card sorting, and that’s whether it’s better to run card sorts in person (moderated) or remotely over the internet (unmoderated).

‍

This article should give you some insight into the world of online card sorting. We've included an analysis of the benefits (and the downsides) as well as why people use this approach. Let's take a look!

‍

How an online card sort works

‍

Running a card sort remotely has quickly become a popular option just because of how time-intensive in-person card sorting is. Instead of needing to bring your participants in for dedicated card sorting sessions, you can simply set up your card sort using an online tool (like our very own OptimalSort) and then wait for the results to roll in.

‍

So what’s involved in a typical online card sort? At a very high level, here’s what’s required. We’re going to assume you’re already set up with an online card sorting tool at this point.

‍

1. Define the cards: Depending on what you’re testing, add the items (cards) to your study. If you were testing the navigation menu of a hotel website, your cards might be things like “Home”, “Book a room”, “Our facilities” and “Contact us”.
2. Work out whether to run a closed or open sort: Determine whether you’ll set the groups for participants to sort cards into (closed) or leave it up to them (open). You may also opt for a mix, where you create some categories but leave the option open for participants to create their own.
3. Recruit your participants: Whether using a participant recruitment service or by recruiting through your own channels, send out invites to your online card sort.
4. Wait for the data: Once you’ve sent out your invites, all that’s left to do is wait for the data to come in and then analyze the results.

‍

That’s online card sorting in a nutshell – not entirely different from running a card sort in person. If you’re interested in learning about how to interpret your card sorting results, we’ve put together this article on open and hybrid card sorts and this one on closed card sorts.

‍

Why is online card sorting so popular?

‍

Online card sorting has a few distinct advantages over in-person card sorting that help to make it a popular option among information architects and user researchers. There are downsides too (as there are with any remote usability testing option), but we’ll get to those in a moment.

‍

Where remote (unmoderated) card sorting excels:

‍

• Time savings: Online card sorting is essentially ‘set and forget’, meaning you can set up the study, send out invites to your participants and then sit back and wait for the results to come in. In-person card sorting requires you to moderate each session and collate the data at the end.
• Easier for participants: It’s not often that researchers are on the other side of the table, but it’s important to consider the participant’s viewpoint. It’s much easier for someone to spend 15 minutes completing your online card sort in their own time instead of trekking across town to your office for an exercise that could take well over an hour.
• Cheaper: In a similar vein, online card sorting is much cheaper than in-person testing. While it’s true that you may still need to recruit participants, you won’t need to reimburse people for travel expenses.
• Analytics: Last but certainly not least, online card sorting tools (like OptimalSort) can take much of the analytical burden off you by transforming your data into actionable insights. Other tools will differ, but OptimalSort can generate a similarity matrix, dendrograms and a participant-centric analysis using your study data.

‍

Where in-person (moderated) card sorting excels:

‍

• Qualitative insights: For all intents and purposes, online card sorting is the most effective way to run a card sort. It’s cheaper, faster and easier for you. But, there’s one area where in-person card sorting excels, and that’s qualitative feedback. When you’re sitting directly across the table from your participant you’re far more likely to learn about the why as well as the what. You can ask participants directly why they grouped certain cards together.

‍

Online card sorting: Participant numbers

‍

So that’s online card sorting in a nutshell, as well as some of the reasons why you should actually use this method. But what about participant numbers? Well, there’s no one right answer, but the general rule is that you need more people than you’d typically bring in for a usability test.

‍

This all comes down to the fact that card sorting is what’s known as a generative method, whereas usability testing is an evaluation method. Here’s a little breakdown of what we mean by these terms:

‍

Generative method: There’s no design, and you need to get a sense of how people think about the problem you’re trying to solve. For example, how people would arrange the items that need to go into your website’s navigation. As Nielsen Norman Group explains: “There is great variability in different people's mental models and in the vocabulary they use to describe the same concepts. We must collect data from a fair number of users before we can achieve a stable picture of the users' preferred structure and determine how to accommodate differences among users”.

‍

Evaluation method: There’s already a design, and you basically need to work out whether it’s a good fit for your users. Any major problems are likely to crop up even after testing 5 or so users. For example, you have a wireframe of your website and need to identify any major usability issues.

‍

Basically, because you’ll typically be using card sorting to generate a new design or structure from nothing, you need to sample a larger number of people. If you were testing an existing website structure, you could get by with a smaller group.

‍

Where to from here?

‍

Following on from our discussion of generative versus evaluation methods, you’ve really got a choice of 2 paths from here if you’re in the midst of a project. For those developing new structures, the best course of action is likely to be a card sort. However, if you’ve got an existing structure that you need to test in order to usability problems and possible areas of improvement, you’re likely best to run a tree test. We’ve got some useful information on getting started with a tree test right here on the blog.

"Hi Optimal Work,I was wondering if there are some best practices you stick to when creating or sending out different UX research studies (i.e. Card sorts, Prototyye Test studies, etc)? Thank you! Mary"

‍

Indeed I do! Over the years I’ve learned a lot about creating remote research studies and engaging participants. That experience has taught me a lot about what works, what doesn’t and what leaves me refreshing my results screen eagerly anticipating participant responses and getting absolute zip. Here are my top tips for remote research study creation and launch success!

‍

‍

Creating remote research studies

‍

‍

Use screener questions and post-study questions wisely

‍

Screener questions are really useful for eliminating participants who may not fit the criteria you’re looking for but you can’t exactly stop them from being less than truthful in their responses. Now, I’m not saying all participants lie on the screener so they can get to the activity (and potentially claim an incentive) but I am saying it’s something you can’t control. To help manage this, I like to use the post-study questions to provide additional context and structure to the research.

‍

Depending on the study, I might ask questions to which the answers might confirm or exclude specific participants from a specific group. For example, if I’m doing research on people who live in a specific town or area, I’ll include a location based question after the study. Any participant who says they live somewhere else is getting excluded via that handy toggle option in the results section. Post-study questions are also great for capturing additional ideas and feedback after participants complete the activity as remote research limits your capacity to get those — you’re not there with them so you can’t just ask. Post-study questions can really help bridge this gap. Use no more than five post-study questions at a time and consider not making them compulsory.

‍

‍

Do a practice run

‍

No matter how careful I am, I always miss something! A typo, a card with a label in the wrong case, forgetting to update a new version of an information architecture after a change was made — stupid mistakes that we all make. By launching a practice version of your study and sharing it with your team or client, you can stop those errors dead in their tracks. It’s also a great way to get feedback from the team on your work before the real deal goes live. If you find an error, all you have to do is duplicate the study, fix the error and then launch. Just keep an eye on the naming conventions used for your studies to prevent the practice version and the final version from getting mixed up!

‍

Sending out remote research studies

‍

‍

Manage expectations about how long the study will be open for

‍

Something that has come back to bite me more than once is failing to clearly explain when the study will close. Understandably, participants can be left feeling pretty annoyed when they mentally commit to complete a study only to find it’s no longer available. There does come a point when you need to shut the study down to accurately report on quantitative data and you’re not going to be able to prevent every instance of this, but providing that information upfront will go a long way.

‍

‍

Provide contact details and be open to questions

‍

You may think you’re setting yourself up to be bombarded with emails, but I’ve found that isn’t necessarily the case. I’ve noticed I get around 1-3 participants contacting me per study. Sometimes they just want to tell me they completed it and potentially provide additional information and sometimes they have a question about the project itself. I’ve also found that sometimes they have something even more interesting to share such as the contact details of someone I may benefit from connecting with — or something else entirely! You never know what surprises they have up their sleeves and it’s important to be open to it. Providing an email address or social media contact details could open up a world of possibilities.

‍

‍

Don’t forget to include the link!

‍

It might seem really obvious, but I can’t tell you how many emails I received (and have been guilty of sending out) that are missing the damn link to the study. It happens! You’re so focused on getting that delivery right and it becomes really easy to miss that final yet crucial piece of information.

‍

To avoid this irritating mishap, I always complete a checklist before hitting send:

‍

• Have I checked my spelling and grammar?
• Have I replaced all the template placeholder content with the correct information?
• Have I mentioned when the study will close?
• Have I included contact details?
• Have I launched my study and received confirmation that it is live?
• Have I included the link to the study in my communications to participants?
• Does the link work? (yep, I’ve broken it before)

‍

‍

‍

General tips for both creating and sending out remote research studies

‍

‍

Know your audience

‍

First and foremost, before you create or disseminate a remote research study, you need to understand who it’s going to and how they best receive this type of content. Posting it out when none of your followers are in your user group may not be the best approach. Do a quick brainstorm about the best way to reach them. For example if your users are internal staff, there might be an internal communications channel such as an all-staff newsletter, intranet or social media site that you can share the link and approach content to.

‍

‍

Keep it brief

‍

And by that I’m talking about both the engagement mechanism and the study itself. I learned this one the hard way. Time is everything and no matter your intentions, no one wants to spend more time than they have to. Even more so in situations where you’re unable to provide incentives (yep, I’ve been there). As a rule, I always stick to no more than 10 questions in a remote research study and for card sorts, I’ll never include more than 60 cards. Anything more than that will see a spike in abandonment rates and of course only serve to annoy and frustrate your participants. You need to ensure that you’re balancing your need to gain insights with their time constraints.

‍

As for the accompanying approach content, short and snappy equals happy! In the case of an email, website, other social media post, newsletter, carrier pigeon etc, keep your approach spiel to no more than a paragraph. Use an audience appropriate tone and stick to the basics such as: a high level sentence on what you’re doing, roughly how long the study will take participants to complete, details of any incentives on offer and of course don’t forget to thank them.

‍

‍

Set clear instructions

‍

The default instructions in Optimal Workshop’s suite of tools are really well designed and I’ve learned to borrow from them for my approach content when sending the link out. There’s no need for wheel reinvention and it usually just needs a slight tweak to suit the specific study. This also helps provide participants with a consistent experience and minimizes confusion allowing them to focus on sharing those valuable insights!

‍

‍

Create a template

‍

When you’re on to something that works — turn it into a template! Every time I create a study or send one out, I save it for future use. It still needs minor tweaks each time, but I use them to iterate my template.What are your top tips for creating and sending out remote user research studies? Comment below!

"Dear Optimal Worshop

I have two huge content structures I would like to A/B test. Do you think Treejack would be appropriate?"

— Mike

‍

Hi Mike (and excellent question)!

‍

Firstly, yes, Treejack is great for testing more than one content structure. It’s easy to run two separate Treejack studies — even more than two. It’ll help you decide which structure you and your team should run with, and it won’t take you long to set them up.

‍

When you’re creating the two tree tests with your two different content structures, include the same tasks in both tests. Using the same tasks will give an accurate measure of which structure performs best. I’ve done it before and I found that the visual presentation of the results — especially the detailed path analysis pietrees — made it really easy to compare Test A with Test B.

‍

Plus (and this is a big plus), if you need to convince stakeholders or teammates of which structure is the most effective, you can’t go past quantitative data, especially when its presented clearly — it’s hard to argue with hard evidence!

‍

Here’s two example of the kinds of results visualizations you could compare in your A/B test: the pietree, which shows correct and incorrect paths, and where people ended up:

‍

‍

‍

And the overall Task result, which breaks down success and directness scores, and has plenty of information worth comparing between two tests:

‍

‍

Keep in mind that running an A/B tree test will affect how you recruit participants — it may not be the best idea to have the same participants complete both tests in one go. But it’s an easy fix — you could either recruit two different groups from the same demographic, or test one group and have a gap (of at least a day) between the two tests.

‍

I’ve one more quick question: why are your two content structures ‘huge’?

‍

I understand that sometimes these things are unavoidable — you potentially work for a government organization, or a university, and you have to include all of the things. But if not, and if you haven’t already, you could run an open card sort to come up with another structure to test (think of it as an A/B/C test!), and to confirm that the categories you’re proposing work for people.

‍

You could even run a closed card sort to establish which content is more important to people than others (your categories could go from ‘Very important’ to ‘Unimportant’, or ‘Use everyday’ to ‘Never use’, for example). You might be able to make your content structure a bit smaller, and still keep its usefulness. Just a thought... and of course, you could try to get this information from your analytics (if available) but just be cautious of this because of course analytics can only tell you what people did and not what they wanted to do.

‍

All the best Mike!