Latest / 5 Minute UX / Using Card Sorting to Discover Information Patterns
Transcript
- 0:08This lesson is Using Card Sorting to Discover Information
- 0:11Patterns. Let's get started.
- 0:20Card sorting reveals how potential users categorize
- 0:22information, moving beyond internal departmental structures
- 0:25like legal versus human resources. This technique helps
- 0:28you create site maps or hierarchies of content,
- 0:30categories, and subcategories for articles, documents, videos,
- 0:33or photos. It is particularly effective for company intranets,
- 0:35which often default to department-based navigation that
- 0:38does not match user mental models. By observing these
- 0:40patterns, you ensure the final architecture reflects user
- 0:43expectations rather than organizational charts. The goal
- 0:45is to discover how users naturally group items, ensuring
- 0:48the final architecture reflects user expectations rather than
- 0:51organizational charts. When you let patterns form on their own,
- 0:53you avoid forcing structure prematurely. This approach
- 0:56uncovers common areas of confusion and higher-level
- 0:58categories such as goals, mental models, or feature requests.
- 1:01You’ll see how people debate rationale in group sorts,
- 1:03providing insight into organization thinking. Once
- 1:06you’ve identified these emergent patterns, you can use affinity
- 1:08diagramming to bring together various trends for the project
- 1:11team. This moves card-sorting results into additional levels
- 1:14of structure, bridging the gap between raw data and design
- 1:16decisions. Now that you understand why card sorting
- 1:19transcends departmental lines, let’s look at preparing
- 1:21materials, specifically the forty to sixty item range and
- 1:24terminology choices.
- 1:32You start by gathering forty to sixty items for the sort,
- 1:35because this specific range allows for a large number of
- 1:38groups without overwhelming participants or the subsequent
- 1:40analysis. If you exceed this limit, you risk analysis
- 1:43overload, which means the data becomes too complex to interpret
- 1:46effectively. So when you prepare the materials, you stick
- 1:49strictly to this forty to sixty item range to ensure clarity.
- 1:52You then choose items that are easy to understand and free from
- 1:54unnecessary jargon to ensure clarity during the session. It’s
- 1:57crucial that the content is accessible, so you include
- 2:00subject-matter terms likely known by user groups, but you
- 2:03avoid excessive insider terms that may confuse participants.
- 2:05You print each item on a separate card to allow for
- 2:08physical manipulation and grouping during the exercise.
- 2:10This tangible format helps participants engage with the
- 2:13content directly, which means they can move items around
- 2:16freely as they discover patterns. By identifying the
- 2:18forty to sixty item range and jargon-free criteria for card
- 2:21selection, you set the stage for meaningful results. Once these
- 2:24cards are ready, you can move on to deciding whether to use an
- 2:27open or closed sort process.
- 2:35You choose the sorting process based on whether you need to
- 2:38discover new structures or validate existing ones, which
- 2:40determines how participants interact with the cards. In an
- 2:43open sort, participants make their own groups and title each
- 2:46group themselves, revealing emergent categories that reflect
- 2:49their unique mental models. This approach is essential when you
- 2:51are building information architecture from scratch and
- 2:54need to understand how users naturally categorize content
- 2:57without preconceived constraints. Conversely, a
- 2:59closed sort requires participants to group items into
- 3:01categories provided beforehand, which is useful for validating
- 3:04an existing site map or hierarchy. You prefer closed
- 3:06sorts when adding large amounts of content to an existing
- 3:09information architecture, because results are easier to
- 3:12analyze with a small set of possible categories. The
- 3:14analysis becomes straightforward when the potential categories
- 3:17are fixed, allowing you to quickly identify where items fit
- 3:20best. You must also decide between individual sorts, where
- 3:22participants work alone, or group sorts conducted as part of
- 3:25a focus group activity where participants work together.
- 3:28Individual sorts reveal personal mental models, while group sorts
- 3:31provide insight into collaborative rationale by
- 3:33observing how people debate the placement of each item.
- 3:35Selecting the right variation ensures you capture the specific
- 3:38insights needed for your design goals.
- 3:46You begin by distributing the printed cards to participants,
- 3:49who work either individually or in small groups depending on
- 3:52your research goals. The next step is to ask them to sort
- 3:55these items into groups that make sense, using pre-provided
- 3:58categories for closed sorts or self-created groups for open
- 4:01sorts. As the clusters take shape, you label the groups to
- 4:03provide structure, but only after the patterns start to form
- 4:06naturally rather than forcing premature organization. If
- 4:09specific cards belong to multiple categories, you handle
- 4:12overlaps by writing duplicates and placing them in each
- 4:14appropriate group to reflect complex user mental models.
- 4:17Throughout this process, you observe the emerging patterns
- 4:20and resist the urge to impose structure, allowing the data to
- 4:23reveal the true information architecture. You now have the
- 4:25raw sorting data ready for analysis to identify common
- 4:28areas of confusion and higher- level categories like goals or
- 4:31feature requests.
- 4:39Once the sorting concludes, you look for emergent patterns and
- 4:42common areas of confusion rather than just counting groups.
- 4:45You’ll identify higher-level categories like Goals, Mental
- 4:47Models, or Feature Requests that reveal how users truly think.
- 4:50For instance, a comment like “I use this tool as my briefcase”
- 4:53exposes a mental model you might otherwise miss. These insights
- 4:56are raw data that need structure before you can present them to
- 4:59the project team. That is where affinity diagramming becomes
- 5:02essential for bringing together various trends and attitudes.
- 5:05You cluster similar cards to visualize the underlying logic,
- 5:08turning scattered observations into a coherent map for your
- 5:10stakeholders. This process transforms individual behaviors
- 5:13into a shared understanding of the information architecture.
- 5:16It’s the bridge between user behavior and design decisions,
- 5:19making the invisible patterns visible and actionable. Be wary
- 5:22of overwhelming participants or creating analysis overload by
- 5:24exceeding the recommended range. Strictly sticking to the forty
- 5:27to sixty item range prevents cognitive fatigue and keeps the
- 5:30data manageable. If you include too many insider terms, you risk
- 5:33jargon confusion that distorts the results. Choose items free
- 5:36from unnecessary jargon to ensure clarity during the
- 5:38session. Finally, resist the urge to force structure
- 5:41prematurely by labeling groups before they form naturally. Let
- 5:44the patterns emerge on their own, then apply labels once the
- 5:47clusters are distinct. This discipline ensures the final
- 5:49hierarchy reflects user mental models, not your assumptions.
- 5:52You now have the framework to execute a card sorting session
- 5:55that uncovers genuine categorization patterns.
- 6:00Welcome to The Take. I'm joined by a fellow practitioner and
- 6:03together we're going to dig into Using Card Sorting to Discover
- 6:06Information Patterns — what it means in practice and what's
- 6:09worth remembering.
- 6:11[HOST] The forty to sixty card limit feels arbitrary until you
- 6:14hit analysis paralysis. [LEARNER] I know. I tried
- 6:17sorting eighty items last quarter and the data became
- 6:20noise. [HOST] Exactly. The cognitive load spikes and the
- 6:23mental models blur into each other. [LEARNER] It’s harder
- 6:26when stakeholders insist on including every obscure feature.
- 6:30[HOST] You have to push back. Jargon-heavy items distort the
- 6:33natural categorization process. [LEARNER] True. I strip out
- 6:36internal acronyms to keep the focus on user intent. [HOST]
- 6:40That discipline is vital. Otherwise, you’re mapping the
- 6:43org chart, not the user. [LEARNER] But open sorts are
- 6:46messy. How do you handle participants creating duplicate
- 6:49groups? [HOST] You let them. Duplicate cards reveal complex
- 6:52mental models that closed sorts miss. [LEARNER] I worry about
- 6:56the time cost of analyzing those overlaps manually. [HOST] That’s
- 6:59where AI is changing the workflow. Pattern recognition
- 7:02tools speed up affinity mapping. [LEARNER] So the AI handles the
- 7:06clustering while we interpret the rationale? [HOST] Precisely.
- 7:09You still need to read the comments to understand the why.
- 7:13[LEARNER] I like that. It keeps the human insight central while
- 7:16removing the drudgery. [HOST] Just remember the tool is only
- 7:20as good as the clean cards you feed it.