# Card Sorting for B2B Resource Libraries: Learn How Buyers Group the Content

<a id="main-content"></a>

SEO &amp; Search

October 28, 2026  ·  8 min read  ·  Victor Valentine Romo

On this page

1. [Define the Finding Problem Before Making Cards](<#define-the-finding-problem-before-making-cards>)
2. [Build a Small Fictional Card Set](<#build-a-small-fictional-card-set>)
3. [Choose Open or Closed Sorting for the Question](<#choose-open-or-closed-sorting-for-the-question>)
4. [Give Neutral Instructions and Preserve Uncertainty](<#give-neutral-instructions-and-preserve-uncertainty>)
5. [Use a Recording Sheet That Retains the Reason](<#use-a-recording-sheet-that-retains-the-reason>)
6. [Compare Two Fictional Grouping Patterns](<#compare-two-fictional-grouping-patterns>)
7. [Turn Patterns Into Navigation Hypotheses](<#turn-patterns-into-navigation-hypotheses>)
8. [Keep the Grouping Exercise Inside Its Boundary](<#keep-the-grouping-exercise-inside-its-boundary>)
9. [Sources](<#sources>)

A resource library can mirror the company that published it. Product owns one shelf. Support owns another. Marketing owns the guides.

The buyer arrives with a question about moving a file. Nothing in that question names a department.

Before choosing a new menu, examine how prospective readers connect the material. Give them content items to group. Preserve their explanations beside the finished piles.

[Nielsen Norman Group defines card sorting](<https://www.nngroup.com/articles/card-sorting-definition/>) as a method in which participants group labeled items according to relationships that make sense to them. The exercise below applies that method to a fictional B2B resource library. Its worksheets and decision rules are proposed editorial practices. They describe no customer study or measured business outcome.

<a id="define-the-finding-problem-before-making-cards"></a>

## Define the Finding Problem Before Making Cards

Write one research question: how would someone organize the material needed to evaluate, begin, and maintain a file-transfer service?

Keep the question separate from the proposed solution. “Do buyers like our three departments?” already puts the company structure inside the exercise. Start with the information people need.

For this fictional study, include public buying and operating resources. Exclude internal [procedures](</articles/sop-documentation-b2b/>), account records, contractual negotiations, and unpublished product plans. Nothing in the exercise requires confidential documents.

Specify whose perspective you want to explore. A person evaluating a service and a person operating it may approach the same document differently. Recruit for relevant experience without assuming a job title establishes that experience.

Use a short screening conversation about the person's recent work. Ask whether they have compared transfer options, prepared an import, or investigated a failed exchange. Record the relevant task in general terms. Avoid collecting employer names or examples containing private files.

If the immediate question concerns an existing catalog, use the [information architecture audit](</articles/information-architecture-audit-ecommerce/>) to inspect its current structure. This exercise generates grouping ideas before choosing that structure.

<a id="build-a-small-fictional-card-set"></a>

## Build a Small Fictional Card Set

Use one resource topic per card. Give each item a stable identifier so you can record it without rewriting its label.

The following twelve cards form an intentionally small teaching example. This set illustrates the worksheet. It does not establish an adequate size for a real research program.

| ID | Card label |
| --- | --- |
| C01 | Supported file formats |
| C02 | Example import file |
| C03 | Explanation of file fields |
| C04 | Connection requirements |
| C05 | Required account permissions |
| C06 | Error message reference |
| C07 | First transfer checklist |
| C08 | Recovering a failed transfer |
| C09 | Service agreement terms |
| C10 | What each service package includes |
| C11 | Understanding the monthly bill |
| C12 | Stopping scheduled transfers |

Keep the card labels plain. Remove campaign names, team acronyms, and branded framework terms. Read [each label without its neighboring cards](</articles/descriptive-link-text-b2b-websites/>). Ask whether it identifies a resource or merely sounds like a heading.

Give overlapping resources distinct meanings. In this example, C02 is a sample file, while C03 explains individual fields. If the source library contains only one combined document, represent that document accurately. Do not invent two resources to make the sort look balanced.

Keep a separate inventory mapping card IDs to the public resources they represent. The researcher uses that mapping. Participants receive the cards without departmental ownership or the current menu path.

Pilot the labels before the main exercise. If someone cannot explain a card's meaning, record the ambiguity and revise the set. Preserve the version used in each session.

<a id="choose-open-or-closed-sorting-for-the-question"></a>

## Choose Open or Closed Sorting for the Question

In an open sort, participants create and name their groups. In a closed sort, they place cards within supplied categories. [NN/G distinguishes these approaches](<https://www.nngroup.com/articles/card-sorting-definition/>) and recommends [tree testing](</articles/tree-testing-b2b-navigation/>) for evaluating navigation categories.

Use an open exercise here to explore possible relationships. Ask what belongs together and why. Keep a closed exercise for a narrower placement question once candidate categories exist.

For example, suppose a draft already contains “Evaluate,” “Set up,” and “Operate.” A closed sort can explore where participants place the billing explanation under that constraint. It cannot tell you whether those three categories would emerge without prompting.

Record the supplied categories as part of the study materials. A grouping made within those boundaries answers a different question from a grouping made freely. Keep those records separate.

Do not use a closed exercise to stage approval of an internal decision. If the categories cannot change, state the fixed constraint. Examine ambiguous placements honestly. Leave the broader navigation decision open.

<a id="give-neutral-instructions-and-preserve-uncertainty"></a>

## Give Neutral Instructions and Preserve Uncertainty

For the proposed open exercise, read these instructions in the same form each time:

> Arrange these resource cards into groups that make sense to you. Make as many groups as you need. Groups can contain different numbers of cards. Move cards whenever you want. Set unclear cards aside. After grouping, name each group and explain the relationships you used.

These are original sample instructions. They are not a quotation from a participant or a required research script.

Keep the participant's screen or workspace free of the current menu. Offer a way to handle the materials that the participant can use. If the sorting interface itself causes difficulty, record that obstacle separately from content grouping.

When someone pauses, leave room for thought. Avoid suggesting a department or correcting a placement. Ask what they are considering. If they ask what a card means, record the question before supplying a prepared factual explanation.

After the sort, ask which cards were difficult to place. Ask whether any item could belong elsewhere. Keep the answer attached to the card rather than flattening it into a general satisfaction score.

Record a changed placement as a change. A final screenshot alone omits the earlier alternative that may explain the uncertainty.

<a id="use-a-recording-sheet-that-retains-the-reason"></a>

## Use a Recording Sheet That Retains the Reason

Start each session record with a participant code, relevant task experience, card-set version, study type, and any assistance provided. Keep identifying information outside the analysis worksheet.

Use one row per group, then additional rows for uncertain cards. The explanation field contains a paraphrase unless an exact quotation was [captured with permission](</articles/b2b-research-consent-recording-scope/>).

| Record | Cards | Label | Explanation or open question |
| --- | --- | --- | --- |
| Group | \[IDs\] | \[Participant's words\] | \[Reason for grouping\] |
| Uncertain card | \[ID\] | Unplaced | \[Unknown meaning or competing home\] |
| Revised placement | \[ID\] | \[Old and new group\] | \[Reason for change\] |

For each explanation, [distinguish observed action from interpretation](</articles/b2b-usability-findings-triage/>). “Moved C05 beside C04” is an action. “Wants a technical menu” is an interpretation requiring more evidence.

Add a researcher note for possible follow-up. Keep it outside the participant's label field. The worksheet must not make your preferred wording look like participant language.

Before ending, review the group memberships with the participant. Confirm that the record reflects the finished sort. Preserve the uncertain pile even when it complicates the analysis.

<a id="compare-two-fictional-grouping-patterns"></a>

## Compare Two Fictional Grouping Patterns

Consider two invented sessions using the same twelve cards. Participant A evaluates services for a small team. Participant B prepares recurring file transfers. Their task backgrounds are fictional descriptions, not representative buyer segments.

Participant A builds three groups:

| Participant label | Cards |
| --- | --- |
| Decide whether it fits | C01, C02, C09, C10, C11 |
| Get ready | C03, C04, C05, C07 |
| Keep it working | C06, C08, C12 |

Participant A places the example file beside formats because they want to inspect the shape of a supported import. They place the field explanation with preparation. They see the bill as part of evaluating the purchase.

Participant B makes four different groups:

| Participant label | Cards |
| --- | --- |
| File preparation | C01, C02, C03 |
| Connection setup | C04, C05, C07 |
| Transfer problems | C06, C08 |
| Account administration | C09, C10, C11, C12 |

Participant B groups stopping a schedule with administration. Participant A groups it with ongoing operation. Both groupings remain plausible within their stated explanations.

Each pattern accounts for all twelve cards once. Neither participant leaves a card unplaced in this [worked example](</>). That complete accounting verifies the example's transcription. It says nothing about whether a real card set would be clear.

The same pairings can hide different reasoning. Both participants put C01 beside C02, but one groups them by a buying decision and the other by preparation work. Preserve that difference before naming a category.

Two fictional sessions [establish no market percentages](</articles/small-b2b-customer-survey-interpretation/>). Even two real sessions would leave recruitment, role coverage, and task context unresolved. Use the patterns to write questions worth investigating.

<a id="turn-patterns-into-navigation-hypotheses"></a>

## Turn Patterns Into Navigation Hypotheses

Write a candidate structure as a hypothesis with a reason and a planned challenge. Avoid voting a menu into existence from the group labels.

| Candidate | Reason to examine it | Unresolved question |
| --- | --- | --- |
| Organize by stage of work | A uses evaluation, preparation, and operation | Can a returning user locate account information? |
| Organize by resource subject | B separates files, connections, problems, and administration | Does a new buyer understand the boundaries? |
| Add a second route to stopping schedules | C12 has two plausible homes | Can both paths remain clear without duplicating the content? |

Keep source documents distinct from navigation entries. Exploring two routes to one resource does not require [publishing duplicate articles](</articles/keyword-cannibalization-content-consolidation/>). Record the proposed destination once and list each candidate access path separately.

For the first hypothesis, draft plain category names. For the second, draft subject labels. Test the routes with realistic finding tasks before choosing a menu. The participant's original group name stays in the evidence record even if the candidate label changes.

NN/G cautions that [card-sort labels need not become final navigation labels](<https://www.nngroup.com/articles/card-sorting-definition/>). Treat each candidate as an interpretation that still needs evaluation.

<a id="keep-the-grouping-exercise-inside-its-boundary"></a>

## Keep the Grouping Exercise Inside Its Boundary

Do not ask this worksheet to diagnose a search engine. A resource can have a reasonable category and still fail to appear for a submitted query. Use [internal site search analysis](</articles/internal-site-search-analysis-b2b/>) for that separate investigation.

Keep content planning separate too. The [content hub architecture article](</articles/saas-content-hub-architecture/>) covers relationships among published resources. The card sort supplies possible reader groupings for those resources. It does not supply keyword demand or a publishing forecast.

Prepare a handoff containing the research question, card inventory, session records, candidate structures, unresolved placements, and the next evaluation task. Name which perspective remains missing from the sample.

If the team [cannot agree on a candidate](</articles/conflicting-client-feedback-consulting/>), return to the disputed relationship. Ask what evidence would distinguish the options. Another discussion about personal menu preferences will not fill the missing record.

Keep the uncertain cards visible.

<a id="sources"></a>

## Sources

- [Nielsen Norman Group: Card Sorting: Uncover Users' Mental Models for Better Information Architecture](<https://www.nngroup.com/articles/card-sorting-definition/>). Method, open and closed sorting, and interpretation limits. Checked 2026.10.28. Worksheets and all examples above are original [proposals](</articles/proposal-templates-b2b-services/>).
