# Give a guide collection a usable starting point

Complete original constructed teaching cases. No real people, studies, source datasets, customers, accounts or external publications.

Define the reader’s finding task, then propose clear groups, specific link labels and a route to the needed guide. Keep a collection, a sequential procedure and search as different aids. Editorial grouping is a hypothesis until suitable real evidence supports it; a successful link check does not establish that readers can find the answer.

## Write a concrete finding task

A creator with a library of articles can ask a useful question: where would someone start to check a report’s period, prepare an accessible export or publish the intended file? These tasks do not require the same first guide. List what the person needs to reach and the information they already have. Do not define the need as “use our new navigation”; that presupposes your preferred solution.

Separate the proposed task from evidence that anyone has it. Relevant existing questions, permitted research and actual support patterns can inform the plan; an editor’s intuition can only supply a hypothesis. GOV.UK’s user-need guidance stresses the action and reason before the format. Its government task examples do not establish the needs of every LinkedIn creator.

## Choose groups, cross-links and sequences deliberately

Use a group to help readers explore related references. Use a genuine step-by-step explanation when the job depends on a sequence, and keep any prerequisite visible. A “Publishing” folder is not a complete procedure just because its links happen to be numbered. Give each guide one stable destination and cross-link when the same resource genuinely helps more than one task.

GOV.UK distinguishes topic tagging, document collections, manuals and step-by-step navigation. Its collection rules explicitly reject using a collection as a substitute for guiding a task. These are that publishing system’s policies; the transferable decision is whether the reader needs a set of references or an explained route. No native LinkedIn folder, category permission or automatic related-link feature is asserted here.

## Build and inspect one complete proposed map

The original exercise has six cards: A metric meaning, B reporting period, C export inspection, D document publication, E image alternatives and F disclosure/privacy review. A draft map puts A/B under understanding a report, C/E/F under preparing a public copy and D under publishing. D explicitly links to C and F as preparation; those links do not create duplicate guides or prove the sequence works for readers.

Check each proposed task against the map: a period question starts at B; a publishing task reaches D and can find C/F; a chart explanation can find A and E. Save unclear labels and missing routes rather than forcing every card into one folder. If a generic “Tools” label hides all six tasks, changing only its icon does not resolve the wording problem. This is an editor-constructed route inspection, with no participants or measured findability.

| Card | Task | Proposed entry |
| --- | --- | --- |
| A | Interpret metric | Understand report |
| B | Check period | Understand report |
| C | Inspect export | Prepare public copy |
| D | Publish document | Publish; C/F preparation |
| E | Image alternatives | Prepare public copy |
| F | Disclosure and privacy | Prepare public copy |

## Resolve three navigation choices

N1 — Constructed duplicate route: the same privacy guide is copied into two categories with different text and no stated owner. Decision: retain one intended source and use two descriptive links where both tasks need it; resolve existing factual differences before consolidation. Outcome: a proposed ownership and navigation repair, not a verified live merge or proof that one category always fits every reader.

N2 — Constructed prerequisite gap: a “Publish the PDF” link reaches a native-upload guide but gives no route to review the actual exported file or private details. Decision: name those prerequisite checks beside the publishing entry and retain the upload guide’s own complete steps. Outcome: a technically inspectable route proposal; actual reader task success remains unmeasured.

N3 — Constructed language gap: an English label says “Check the reporting period,” while its German counterpart says only “Analyse.” Decision: write a specific German task label and check that it reaches the same intended topic in German. Outcome: aligned editorial intent and a destination check, not independent localization validation or observed language preference.

## Use real grouping and finding evidence appropriately

Card sorting can explore how willing relevant participants group and name existing cards; an open sort lets them propose groups, while a closed sort uses supplied categories. Victoria’s method notes the dependence on the prepared card set and warns that large sets can overwhelm. Results inform a structure rather than dictate the correct one. An automated persona sorting cards is not evidence of real readers’ mental models.

A June 2026 Department for Education design history reports an open sort with nine people, a draft structure and tree testing as a next step. That is the publisher’s reported process, not our study or a completed validation of this journal. For your own eligible future check, use neutral finding tasks and record actual routes, completion, confusion and uncertainty. A click alone does not establish comprehension or preference.

## Maintain the routes with the content

The record preserves task and supporting evidence, cards/source destinations, proposed labels in each language, groups, cross-links, prerequisites, ambiguous items and observed versus pending checks. Test actual links, keyboard reachability, mobile wrapping and empty-search recovery separately from reader finding success. Search, categories and an alphabetical list can complement a few task entry points without making the collection’s entire content one forced sequence.

Primary guidance was read on 1 November 2026. The map and all N-cases are original constructions. This journal’s editor-selected starting links are a maintained proposal, not a card-sort outcome or newly completed reader study. No person was recruited or contacted, and no preference was inferred. Keep real later findings separate from an editor’s completed link check.

1. Define finding task and existing evidence.
2. Draft stable cards and descriptive labels.
3. Inspect groups, cross-links and prerequisites.
4. Actually check links, keyboard and mobile layout.
5. Keep future real-reader evidence separate from editor assumptions.

## Primary sources

- [GOV.UK: Task needs and supporting evidence](https://guidance.publishing.service.gov.uk/writing-to-gov-uk-standards/plan-manage-content/identify-user-needs/)
- [GOV.UK: Grouping, collections and sequences](https://guidance.publishing.service.gov.uk/writing-to-gov-uk-standards/plan-manage-content/organise-group-govuk-content/)
- [Victoria: Card-sort method and limitations](https://www.vic.gov.au/card-sorting)
- [DfE: June 2026 first-party card-sort design history](https://design-histories.education.gov.uk/deliver-good-services/using-a-card-sort-to-understand-how-users-group-information)