Card sorting results template: how to document and present your findings
Stop staring at a blank document after your card sort. This template structures raw data into a shareable deliverable your whole team can act on.
Card sorting results template: how to document and present your findings
A card sorting results template is a structured document that organizes your raw participant data, grouping patterns, and agreement scores into a shareable deliverable your team can act on. Use one consistently and you’ll cut synthesis time by half while producing cleaner IA recommendations.
This guide gives you the complete template structure, explains what to fill in each section, and shows how to interpret the numbers before you hand findings to design and product.
Why you need a dedicated results template
Card sorting tools like Optimal Workshop, Maze, or Lyssna generate data automatically, but they do not write your report for you. The raw output (dendrograms, similarity matrices, placement percentages) is meaningless to a product manager who has never seen a card sort before.
A results template solves three problems at once:
- Consistency: Every card sort your team runs looks the same, so stakeholders know what to expect.
- Speed: You fill in sections rather than designing a document from scratch after every study.
- Credibility: Structured findings signal methodological rigor, which matters when you are recommending navigation changes that affect the entire product.
The complete card sorting results template
Copy the sections below into your preferred document tool (Notion, Google Docs, Confluence, or similar).
Section 1: Study overview
| Field | Detail |
|---|---|
| Study title | |
| Study type | Open / Closed / Hybrid |
| Date conducted | |
| Researcher | |
| Number of participants | |
| Participant profile | (e.g., B2B SaaS buyers, mid-market, NA/EMEA) |
| Number of cards | |
| Number of predefined categories (closed only) | |
| Tool used | |
| Objective | One sentence: what IA question does this study answer? |
Why it matters: Stakeholders reading the report months later need context. Always include the objective in plain language.
Section 2: Participant summary
Summarize who participated. Even a simple table helps readers calibrate how much to trust the findings.
| Segment | Count | Notes |
|---|---|---|
| Job title or role | ||
| Company size | ||
| Region | ||
| Tenure with product / domain |
If you recruited through a panel or research platform, note the screening criteria used. For B2B studies especially, recording seniority and function helps you spot whether enterprise buyers and power users have different mental models, which they often do.
Section 3: Similarity matrix summary
The similarity matrix is the quantitative core of open card sorting results. Most tools generate a full N x N grid, but that is too dense for a report. Extract the most actionable data.
Top grouping pairs (agreement above 70%)
| Card A | Card B | Agreement % |
|---|---|---|
Weak or contested pairs (agreement below 40%)
| Card A | Card B | Top 2 competing categories | Notes |
|---|---|---|---|
Tip: paste a screenshot of the full matrix as an appendix. Reference specific cells in your written analysis rather than describing every row.
Section 4: Category naming analysis (open sort)
Skip this section for closed sorts. For open sorts, list the most common category names participants created:
| Concept | Top label used | % using this label | Runner-up label | % | Implication |
|---|---|---|---|---|---|
High agreement on both grouping and label signals a strong natural category. High grouping agreement but scattered labels signals the concept is sound but the name needs testing. Low grouping and low label agreement signals the concept may need to be redesigned or split.
Section 5: Placement accuracy (closed sort)
Skip this section for open sorts. For each card, record where most participants placed it:
| Card label | Intended category | % correct placement | Top competing category | % |
|---|---|---|---|---|
Cards below 60% correct placement are your problem areas. They indicate either a poorly labeled card, an unclear category name, or a genuine conceptual mismatch in your proposed IA.
Section 6: Key insights
Write three to five headline findings as bullets. Each bullet should include the supporting data point.
Example format:
- “Settings” and “Account” belong together. 88% of participants grouped account management cards with settings cards. Current navigation separates them, which likely causes drop-off at account setup.
Keep each insight to two sentences maximum. Save elaboration for the recommendation section.
Section 7: Ambiguous items
List every card with low or split agreement. This section is often more valuable than the clean groupings because it highlights where your existing content taxonomy is broken.
| Card | Agreement threshold missed | Competing groups | Recommended action |
|---|---|---|---|
| Relabel / Split / Run follow-up closed sort |
Section 8: IA recommendations
Translate insights into concrete navigation or structure changes. Use a before/after format when possible.
Recommended top-level categories:
| Proposed category | Cards that belong here | Confidence | Basis |
|---|---|---|---|
| High / Medium / Low | Similarity matrix / Naming analysis / Team judgment |
Recommended next step: Note whether you are moving to tree testing, a closed sort validation round, or directly to prototype testing. The card sorting vs tree testing guide covers when each method is appropriate.
Section 9: Appendices
- Full similarity matrix (screenshot or exported table)
- Dendrogram screenshot
- Raw category data export (CSV link or attachment)
- Participant screener used
- Card list used in study
How to analyze results before filling in the template
The template only works if you have actually analyzed your data. Here is the workflow:
Step 1: Export your tool’s data. Optimal Workshop, Maze, Lyssna, and most other card sorting tools export CSV similarity matrices and category data. Download both.
Step 2: Set your agreement threshold. For most studies, treat 70% or above as strong agreement, 40 to 70% as moderate, and below 40% as weak. Adjust down to 60%/35% if you have fewer than 12 participants.
Step 3: Cluster the matrix manually or visually. Look for groups of cards that all score highly with each other. These clusters become your candidate categories. Color-code cells in a spreadsheet if your tool does not generate a dendrogram.
Step 4: Cross-reference with naming data. For open sorts, check whether participants who grouped cards similarly also used similar labels. If they did, you have a natural category. If they did not, naming the category will require separate testing.
Step 5: Flag ambiguous items before writing. Ambiguous cards distort insight quality if you ignore them. Surface them explicitly in Section 7 so stakeholders understand the limits of your recommendations.
This analysis workflow pairs well with affinity mapping when you want to layer qualitative session notes on top of quantitative card sort patterns.
Segment your results when possible
A single aggregate report hides important differences. If your card sort included both enterprise buyers and SMB users, or both developers and non-technical users, run your analysis twice and compare.
Segmented results often reveal that different user types have fundamentally different mental models for the same product. Ignoring this produces a navigation structure that confuses everyone equally rather than serving any group well.
When recruiting for card sorting studies, platforms like CleverX let you filter participants by job title, company size, industry, and seniority, so you can build clean segments from the start rather than trying to reconcile mixed data afterward.
Common mistakes when documenting card sort results
Reporting only strong agreement. The weakest groupings and most contested cards are often the most important findings. Always include ambiguous items.
Skipping the participant summary. A 90% grouping agreement from five participants means something different than the same score from 20 screened users who match your actual customer profile.
Presenting the dendrogram without explanation. Dendrograms are visually compelling but confusing to non-researchers. Always write a plain-language summary of what the dendrogram shows before including the image.
Treating card sort output as final IA. Participants reveal mental models; they do not design navigation. Always validate your proposed structure with tree testing before committing to a build.
Losing raw data. Store exports from your card sorting tool alongside the finished report. If questions arise later, you need the original data to re-analyze, not just your written synthesis.
Frequently asked questions
What should a card sorting results document include?
At minimum it should cover study metadata (type, participant count, card set), a similarity matrix or agreement score table, category naming analysis, key groupings with confidence percentages, outliers and ambiguous items, and IA recommendations. Adding a visual dendrogram screenshot and a next-steps section makes the report actionable for design and product teams.
How many participants do you need for reliable card sorting results?
Most researchers recommend 15 to 20 participants for quantitative open or closed card sorting. This number typically surfaces the main grouping patterns without requiring excessive analysis time. For exploratory open sorts where you only want qualitative themes, 5 to 8 participants can reveal useful signal, but patterns will be less statistically reliable.
What is a similarity matrix in card sorting?
A similarity matrix is a grid showing how often each pair of cards was placed in the same group across all participants. If 18 out of 20 participants grouped ‘Billing’ and ‘Invoices’ together, the cell for that pair shows 90%. High-scoring pairs are strong candidates for the same navigation category. Most card sorting tools generate this automatically.
How do you handle ambiguous cards in your results?
Flag cards that show low agreement (typically below 40-50% with any single group) in a dedicated ‘ambiguous items’ section. Note the top two or three competing categories participants used for each card. Common fixes include rewriting the card label for clarity, splitting the concept into two separate items, or running a follow-up closed sort with cleaner category options.
What is the difference between open and closed card sorting results analysis?
Open sort analysis focuses on category naming (what labels participants invented), grouping similarity, and emerging mental models. Closed sort analysis focuses on placement accuracy (what percentage put each card in its intended category) and which categories caused confusion. The two analyses answer different questions: open sorts inform structure, closed sorts validate it.
How do you present card sorting findings to stakeholders?
Lead with the three to five strongest grouping insights, each supported by an agreement percentage. Follow with a visual similarity matrix or dendrogram screenshot. Then present your IA recommendation with a before/after navigation comparison if possible. Keep the document to one or two pages for executives; attach full data appendix for the design team.