Card sorting for UX

Hey, what's up everyone? Welcome. My name is Joe Formica. I'm the design advocate here at Lyssna, and today we are talking about card sorts. Card sorts are one of the most powerful and one of the most useful research methods for understanding your user's mental models.

Basically, a card sort can help you figure out all the things you need to know to create a site and a navigation that is simple, intuitive, and helps users find the stuff they're looking for quickly.

In this video, I'm going to cover everything you need to know to run a card sort in Lyssna. We're going to start with an overview, talk about what a card sort is. Some of the different variations and when to use them. Then we're going to get into building out a card sort in Lyssna. I'm going to show you how to do that, some of the options available to you, and then we'll walk through recruiting, getting our card sort out there in front of some participants, and finally looking at the results together.

So let's start with the basics. What is a card sort and why and when would you use one? So, as I mentioned, at its core, a card sort is a way to better understand your users' mental models, how they group, name, categorize items, and associate them. So it's most useful for creating a site structure, understanding how your users think things should be organized on your site, and reflecting that in your information architecture and navigation. So a simple example, if you run an e-commerce site selling clothing, and you get a new shipment of hats in.

Should those go under a hat section, a headwear category, an accessories tab. These are the types of questions that come up often and the perfect ones to answer with a card sort.

So here on my screen, you can see what the card sort experience looks like from a participant's perspective.

Just like any card sort, participants are going to be given a set of items or cards, and they're going to be asked to organize those into groups that make sense to them.

So in this example, we're using pieces of furniture, but a card could represent anything. It could be like this example with products on an e-commerce site. It could be features in a SaaS product. Blog posts or content on a news publication, there's no limit to how you can use a card sort. But for this walkthrough, we're going to imagine that we're launching a furniture e-commerce site, and we want to know how users group these items so we can build a navigation that makes it easy for them to find what they're looking for.

So what you're seeing me do here is just drag cards and organize them into groups that make sense to me. Again, acting as the participant, and in this case, I'm giving each group a title, something that also accurately describes the items in there. This is the overall experience that participants are going to have in a card sort, but there are a few different variations worth knowing about that.

The one we're looking at right here is an open card sort. So in an open card sort, participants are not only asked to group items together in a way that makes sense to them, but they're also asked to name those groups. An open card sort is the most freeform variation of a card sort, and that can be extremely valuable.

Now what we're looking at here is a hybrid card sort. Same cards, same items. The experience looks pretty similar, but instead of a completely blank canvas for our participants to group and name these different categories, some groups are already created for them by us as the researchers.

So participants can use those predefined groups. They can drop cards into them. Or they can create their own, or do both. So this is a nice middle ground. Now let's get to a closed card sort. Participants are simply asked to sort cards into groups that are already defined for them, so that's what you're seeing here.

The participant isn't naming anything. They're not creating any new groups. They're just organizing those cards into the category that they think fits best.

So when should you use these card sorts in each variation? Most commonly, you'll start early in the process with an open card sort. It's going to give you the most raw, unfiltered feedback, not only just on how people group things, but especially how they name them.

So if you've got a lot of content that you want to organize, you have maybe no idea or a lot of ideas on how to do it, usually an open card sort is going to be a great first step. Now, let's say you run an open card sort or a couple of them and you start to see some patterns emerge, some common grouping, some recurring category names, trends, and common ways of putting these items together.

That might be a signal to move into a hybrid card sort to validate some of those themes that you're seeing over and over again, but still leave some room for users to surprise you with some new categories.

And typically a closed card sort is going to come last in the process. So once you've established a clear structure and some themes of how people are grouping and naming things, you might want to test whether those final categories actually resonate with users.

All three of these variations are available when you create a study and a card sort in Lyssna. But for this video, we're going to focus on an open card sort with our furniture example. So let's get into it.

All right, so let's build a study featuring our furniture card sort. I've got my Lyssna dashboard open here, and the first thing that we're going to do is create a new study.

So once we do that, we're brought into this edit view. Now, right now we've got a welcome message for our study, but not much else. So our first step is to add a section and we're going to add a card sort. One thing worth noting, Lyssna studies can include multiple sections, multiple research methods, so you can combine different sections and research methods.

You can add follow up questions, you can even set logic for more complex studies. But for this one, we're going to keep it simple, a single card sort for our furniture store. Now that we've added our card sort section, we've got a few things to configure. So first, we're going to choose our card sort type. Open, hybrid, or closed.

As I said earlier, we're going to go with open for this one. Next, we can add a prompt. This is where you're giving participants a simple scenario or a set of instructions to frame the task and get them to think about how they would organize these items. So we've got a simple one here, asking participants to please organize these into groups that make sense to you, and give each group a name.

Lyssna also includes some standardized instructions that your participants are going to see. That reminds them that there are no right or wrong answers, which is always helpful. So now that we've got our task and scenario set up.

We can get into adding our cards. Now. You can type them in one by one. You can paste in a list, which is what I'll do here. And you also have the option to add a description or even an image to each card. , We're going to keep this simple and just use the item names.

If we were running a closed or a hybrid card sort, this is where we'd also add our predefined categories. Now you'll notice there are a few other options when you're setting up a card sort. You can add follow up questions after the sort.

You can require participants to sort every card or allow some cards to go unsorted. You can even enable think aloud recording, which ask participants to talk through their decisions as they go and why they're putting certain cards together, why they're naming categories certain things. So depending on your study goals, these are all really useful tools at the right time to uncover some of those deeper insights.

We're going to keep it simple, but the one option I am going to turn on is to randomize the card order. So this is just going to make sure that the cards aren't presented in any sequence or order that might influence how participants group them. Essentially just shuffling the deck before we show them to participants to keep things unbiased and get the most reliable feedback.

So once everything looks good in our card sort section, we'll save. We're going to move on to recruiting some participants. Alright, now that our card sort is looking good and ready to go, it's time to get it in front of some participants and to start collecting some results. So in Lyssna, there are two ways to recruit participants.

The first is to self recruit using a shareable link. So you can generate this link, you can share it, and anyone who clicks it will be invited to this study. So this is really useful if you've already got a network, an email list, a community of people in your target audience that you want to share it with.

The second option is to recruit from the Lyssna panel. So this taps into a large network of participants from around the world, and it's often the fastest, and I'd say most efficient way to get high quality responses to your study without needing to do any of the outreach on your own. So you can use either option or you can use a combination of both.

For this walkthrough, I'm going to recruit from the panel, so let's click on that.

Now that we're recruiting from the panel, you've got a couple options to set before launching your study. So first, you can choose your sample size, how many participants you want to include in the study. Recommended sample sizes can vary drastically depending on the type of study and your goals.

But for a card sort and for this demo, I'm going to start with 20 participants. That's a pretty solid starting point. You should be able to see some themes and some common groupings emerge from there. And if not, don't worry about starting too small. You can always place another panel order later if you want to increase that sample size.

So I'm going to start with 20 and scroll down to see our next options, which is who we want to get this study out to. You're going to see a wide range of targeting filters and criteria you can use to make sure your card sort or any study gets in front of the right people. So there's geographic location, employment status, industry, demographics, tons of options, and you can combine these any way you want to closely match your target audience.

Once again going with the theme for this study, I'm going to keep it pretty simple and just target participants in the United States. So once you've set your sample size and you've got your targeting criteria, you're ready to go. Let's launch our study and then we'll take a quick break while we wait for the results to roll in.

All right, it's been only a few minutes and we've already got our results back. All 20 participants have completed the card sort, so now it's time to dig in and see how they grouped and named our furniture items. When you're looking at card sort results, you're generally trying to do two things. First is you're trying to look for items that were commonly grouped together.

So when you see the same items consistently placed in the same group across different participants, that's a pretty strong signal that those might belong together and might be something you want to mirror in your own site's navigation.

So the second thing that you're looking for, especially in an open card sort, is the common names for these groups. What language did participants use? What did they call things? Understanding this is incredibly valuable for naming your navigation categories in a way that actually makes sense to your users.

Now, digging through card sort results, especially open card sorts, can feel a little bit messy at first, and that is okay and totally normal. The goal isn't perfection or 100% agreement on every grouping and category name, especially early on. We're really looking for some patterns. We're looking for those recurring mental models, the groupings and labels that keep showing up so you can use those as a foundation to iterate on and test further.

So let's take a look at what came back for our card sort. Now there are a couple ways to look at card sort results and a couple views offered in Lyssna. The first one is the cards view. So on the left you'll see each of your cards, and in our case, each piece of furniture next to it.

It's a breakdown of the categories participants placed it in along with the percentage for each. So for sofa, for example, you might see that 50% of participants placed it in a category that they called living room. Now here's something important to note, because this is an open card sort.

Participants named their own categories. So you might see living room with the first letters capitalized and a space between it. You might see living room, all lowercase, living room, one word, all showing up as separate categories, even though it's pretty clear to us that all of those participants were thinking of the same thing.

This card view is great for getting a granular look at how individual items were sorted, but making sense of the bigger picture will be a little bit easier in some of the other views. So let's move over to the categories view. This essentially just flips the view and the information that we just saw.

Instead of starting with the cards, now we're looking at all of the categories that participants created with the cards they placed inside them. So in our results, you can see there are 51 different categories, but a lot of those are just slight variations or different spellings of the same idea. This view is a great place to do a little cleanup and kind of condense some of the names that we saw.

You can take all of those variations and standardize them under a single name. So for example, you might see outdoor, outside, outdoors, all showing up as separate categories. We can group those together and give them one standardized name, something like outdoor furniture. That way, despite the small differences in how participants phrased it or typed it out, we're capturing the underlying intent. We're making it much easier for ourselves to understand the real groupings that emerged. The cards and categories views give us a solid understanding of how participants named their groups.

The agreement matrix and the similarity matrix are going to take that a step further and show us which items were most consistently grouped together. So the agreement matrix shows how consistently participants placed each card into the same categories.

So for example, we can see that patio dining set and outdoor lounge chair were placed together by a large majority of participants. And after the cleanup we did in that last step, we can see all of those fall under some outdoor furniture title.

We can also see strong alignment between items like a bed frame and a nightstand, and some consistent pairing across some of the other items as well. So this card sort was pretty straightforward. You're not going to see wildly different groupings here, but the agreement matrix confirms that participants were largely on the same page.

Now let's go to the similarity matrix, which kind of takes this a step further. It shows the percentage of participants who sorted each pair of cards together. So the darker the color, the higher the number is the strongest association between those two items. So, for example, that 95% down here at the bottom tells us that nearly every participant put desk and office chair together.

On the other end of the spectrum, some of the lighter cells with lower numbers show weaker associations like sofa and patio dining set, which only 5% or in our case, one participant grouped together, which makes a lot of sense.

Now, as your card sort gets more complex, or as you start to see some of the more varied responses, these results can take some time and some patience to work through.

But for our example here, the insights are pretty clear. Participants consistently grouped bed frame and nightstand into a bedroom category, dining table and dining chairs into a dining room category, desk and office chair into a home office category, sofa and coffee table into living room, and patio dining set and outdoor lounge chair into an outdoor furniture category.

The one interesting note is that a handful of participants grouped patio dining set with dining room table. Worth keeping an eye on, but not a major outlier. So overall I'd feel pretty confident moving forward with these category names and groupings. There aren't any drastically different mental models or major surprises being surfaced in this card sort, but what I'd probably do next is run a closed card sort just to confirm.

So we could use these category names to validate the structure and figure out if it's ready to go live or if it needs another round of testing and discovery.

Card sorts are an incredible research method for understanding your users' mental models, how they group, categorize, name and associate different information, and how they'll expect to find them on your site.

So at this point, you are ready to launch your first card sort. I've linked some additional resources and tips below, just in case you wanna dive deeper.

If you've got a site that's feeling a little cluttered and needs some organization, if you've got a new line of products and you're wondering where to put them. If you just want to put your information architecture to the test and see how easily users can find what they're looking for, a card sort is going to be a great way to get there.

So start simple. Build your first card, sort, launch it, see what comes back. Whatever the results look like, you'll be building your navigation, not on guesswork and assumptions, but your actual users, their feedback, and where they expect to find things on your site. Happy testing and I'll see you in the next one.