Latest / 5 Minute UX / Card Sorting: Running and Analyzing a Study
Transcript
- 0:08This lesson is Card Sorting: Running and Analyzing a Study.
- 0:11Let's get started.
- 0:18Preparation starts by defining your study scope and selecting
- 0:21the right sort type for your specific goals. You must decide
- 0:23whether your project requires an open sort where users create
- 0:26their own groups and titles, or a closed sort where they place
- 0:28items into pre-defined categories. This choice dictates
- 0:30how you structure the session and what kind of data you will
- 0:33ultimately collect. Next, audit your content source to create a
- 0:35comprehensive list of items for your cards. These content items
- 0:38might include articles, documents, videos, or photos
- 0:40that need organizing within your information architecture. Once
- 0:42your list is ready, you choose your card format, which means
- 0:45preparing either physical cards for in-person observation or
- 0:47digital equivalents for remote sessions. Your environment
- 0:50setup depends entirely on whether you are facilitating
- 0:52locally or using a remote tool configured for digital sorting.
- 0:54You also need to ensure your participant count is sufficient
- 0:57to allow common patterns to emerge statistically across
- 0:59multiple sessions. Without enough participants, you won't
- 1:01see the recurring groupings necessary to build a reliable
- 1:04structure. The tangible output of this phase is a ready-to-use
- 1:06deck of cards and a clear understanding of whether the
- 1:08session will be open or closed. This preparation sets the stage
- 1:11for data collection and ensures you are ready to observe genuine
- 1:13user behavior. When you walk into the facilitation phase, you
- 1:16will be watching how users group items and label categories to
- 1:18reveal their mental models.
- 1:25Let's say you have a participant sitting in front of a deck of
- 1:28cards, ready to begin the sorting process. Your primary
- 1:30job is to guide them through the task while remaining completely
- 1:33neutral so their natural behaviors can surface without
- 1:35influence. Whether they are working with physical cards or
- 1:37digital equivalents, you must resist the urge to offer
- 1:40suggestions or correct their choices. You need to observe
- 1:42grouping logic by watching how users cluster items together to
- 1:45understand their mental models. This means paying close
- 1:47attention to which content sources, like articles or
- 1:49videos, they place in the same pile. Because these groupings
- 1:52reveal how users mentally organize information, your
- 1:54observational notes become the raw data for your future site
- 1:57map. Simultaneously, you must monitor labeling to listen to
- 1:59how users name their groups and reveal their specific
- 2:02vocabulary. When a participant creates a title for an open sort
- 2:04or selects a label in a closed sort, you are hearing their
- 2:07expectations directly. This specific language is crucial
- 2:09because it tells you exactly what terms your final content
- 2:12hierarchy should use to make sense to them. The most
- 2:14valuable insights often come from identifying friction where
- 2:16users hesitate, disagree with a category, or express confusion.
- 2:19Do not smooth these moments over, as they highlight exactly
- 2:22where the current structure fails the user. Instead,
- 2:24actively document these areas of disagreement because they
- 2:26represent common areas of confusion or disagreement that
- 2:29need resolution. The session concludes only when the
- 2:31participant has finished sorting all items and labeling their
- 2:34groups. At this point, you have a set of sorted cards and
- 2:36detailed notes on user behavior that are ready for the next
- 2:39phase. These observations will directly feed into your pattern
- 2:41recognition and conflict resolution steps to build a
- 2:43robust information architecture.
- 2:51Pause and think about the last card sort you ran. Did you
- 2:53notice where users hesitated, or did you smooth over their
- 2:56confusion? Those moments of friction are actually your most
- 2:58valuable data points, revealing exactly where your current
- 3:01structure fails the user. Now, look at your observational notes
- 3:03to begin pattern recognition. You need to review all sessions
- 3:06to find recurring groupings and labels across different
- 3:08participants. This isn't about one person's opinion; it's about
- 3:11finding the collective mental model that emerges from the
- 3:14data. Next, move into conflict resolution by identifying common
- 3:16areas of disagreement. Did users struggle to decide where
- 3:19specific items belonged? Documenting these points of
- 3:21confusion is essential because it highlights the gaps in your
- 3:23current information architecture. Once you have
- 3:26resolved these conflicts, proceed to hierarchy
- 3:28construction. Use the identified patterns to draft a site map
- 3:30that organizes your content sources into logical categories
- 3:33and subcategories. This step transforms your raw observations
- 3:35into a tangible structural design. Finally, translate
- 3:37these insights into a proposed information architecture that
- 3:40reflects the collective mental models of your user base. Your
- 3:43final site map should prioritize the most common user patterns
- 3:45while directly addressing the areas of confusion you
- 3:47identified. This ensures your new structure actually reduces
- 3:50friction for the people using it.
- 3:57A frequent error is failing to distinguish between open sorts
- 4:00and closed sorts, which leads to inconsistent data where some
- 4:02users create categories while others use provided ones. You
- 4:05must clarify instructions by reiterating the specific rules
- 4:08of the sort before the session begins to ensure every
- 4:10participant understands their task. This simple step prevents
- 4:13the confusion that derails your entire analysis. Watch out for
- 4:16the tendency to smooth over moments of hesitation, because
- 4:19those friction points are often your most valuable insights.
- 4:21Instead of helping them move on, focus on friction by actively
- 4:24documenting where users struggle to decide how to categorize
- 4:27specific items. These disagreements highlight exactly
- 4:29where your current structure fails the user. Another
- 4:32critical breakdown occurs if the items you prepared do not
- 4:34accurately reflect the actual content source, such as a
- 4:37company intranet. You need to revisit content sources to
- 4:40verify that every card represents a real article,
- 4:42document, or video found in your system. Without this accuracy,
- 4:45your resulting site map will not match the reality users face
- 4:47daily.
- 4:54In your next project, start by auditing your content source to
- 4:57create a comprehensive list of items for your cards. This step
- 4:59ensures your deck accurately represents the actual articles,
- 5:02documents, and videos you need to organize. Without this
- 5:05foundation, your data will lack the relevance needed to inform a
- 5:08real solution. Decide if your current project requires an open
- 5:11sort to explore new structures or a closed sort to validate
- 5:13existing categories. An open sort lets users create their own
- 5:16groups, while a closed sort tests how well they fit into
- 5:19your predefined buckets. Choosing the right type
- 5:21determines whether you are discovering fresh patterns or
- 5:24confirming what you already suspect. Conduct your sessions
- 5:26with a focus on observing where users struggle, then synthesize
- 5:29these observations into a draft site map. Prioritize the most
- 5:32common user patterns while addressing identified areas of
- 5:35confusion to build a hierarchy that truly works. By translating
- 5:38friction points into structural changes, you turn raw sorting
- 5:40data into a functional information architecture. You
- 5:43now have the full workflow to execute a card sort study and
- 5:46synthesize results into an information architecture. This
- 5:48process bridges the gap between user mental models and the
- 5:51tangible site maps that guide your design.
- 5:56Welcome to The Take. I'm joined by a fellow practitioner and
- 5:59together we're going to dig into Card Sorting: Running and
- 6:02Analyzing a Study — what it means in practice and what's
- 6:04worth remembering.
- 6:06[HOST] I'm still thinking about that bit on friction points
- 6:09where we are told not to smooth over user hesitation. [LEARNER]
- 6:12That's the hardest part to execute because our instinct is
- 6:15to help them move forward so the session doesn't drag. [HOST]
- 6:18Exactly, but that hesitation is the only data proving your
- 6:21current categories don't match their mental model. [LEARNER]
- 6:23I've seen researchers gently suggest a label just to keep the
- 6:26flow, which kills the validity of the whole analysis. [HOST]
- 6:29And if you do that, you lose the conflict resolution step
- 6:32entirely since you never actually identified the gap.
- 6:34[LEARNER] It's tempting, but I think AI tools are changing how
- 6:37we capture those moments without us needing to intervene. [HOST]
- 6:40How do you see that working in practice without the human
- 6:43element guiding the observation? [LEARNER] Some platforms now
- 6:46flag hesitation duration and vocal tone automatically, so we
- 6:49get the data without breaking the user's focus. [HOST] That's
- 6:52useful, but can an algorithm really distinguish between a
- 6:54thoughtful pause and genuine confusion about a label?
- 6:57[LEARNER] Not yet, which is why we still need to sit there and
- 7:00document the specific items that caused the stumble. [HOST]
- 7:03Right, the tool gives us the signal, but we have to interpret
- 7:06the noise to build the right hierarchy. [LEARNER] So the real
- 7:09skill shifts from facilitating the sort to interpreting the
- 7:11friction data we collect. [HOST] Precisely, because the site map
- 7:14only works if we resolve the exact points where the structure
- 7:17failed.