Practical framework / 4 MIN READ
Tag content for a real retrieval task
Use categories that help someone find a useful set, with definitions that different editors can apply consistently.
AI-assisted research and analysis for the SNS Creative Fieldbook
Sources checked 2026-11-06
A taxonomy is a set of retrieval decisions. Adding every plausible tag creates noise unless the categories correspond to recognizable questions and have clear boundaries.
What you will leave with: Leave with a small category scheme, application rules and a test using real content.
Use this when
- Editors invent new tags for every article.
- Topic pages mix unrelated levels of detail.
- Readers cannot find a coherent collection on a subject.
This is not a keyword-ranking tactic or a reason to reorganize every existing URL.
Choose the retrieval question
Name what someone is trying to gather: beginner tutorials on a product, examples from a sector or resources for a particular task. Distinguish topic, audience, format and lifecycle stage rather than mixing them into one flat list without explanation. A video is a format; installation is a task. Decide which dimensions readers actually need to filter or browse. Start with a bounded collection and a small set of questions. Do not reproduce the company's reporting hierarchy unless it genuinely matches how the audience looks for information.
Define categories with boundaries
Write a short definition, inclusion rule and exclusion example for each category. If two tags repeatedly apply for the same reason, consider whether they need to be merged or distinguished. GOV.UK's guidance uses topics and collections to organize content; its administrative requirements are specific to that service. The general editorial task is to make a grouping useful and explainable. Give editors a rule for uncertain cases and a route to propose a new term, rather than allowing uncontrolled synonyms to accumulate in the CMS.
Test with real items and readers
Ask two editors to classify a small varied set independently, then compare disagreements. Use the disagreements to refine definitions, not to declare one person careless. Include awkward examples that span topics or formats. Next, ask a reader to find resources for a realistic task using the proposed groupings. Observe whether labels guide them to a useful set. This is a practical diagnostic exercise, not a representative usability result unless the research design supports that claim. Keep the evidence and interpretation separate.
Govern changes without breaking meaning
Assign an owner for category definitions and document how renames, merges and retirements affect existing content. Preserve useful aliases internally where they help editors find the approved term. Review empty, overloaded and rarely useful groupings, but do not delete a category merely because it serves a small specialist need. Check the public collection pages after changes, including titles and explanations. If URLs or access are affected, follow the appropriate migration process. The goal is dependable retrieval, not a perfectly symmetrical classification diagram.
Make the reasoning concrete.
A hypothetical business-learning site mixes case study, finance, beginner and video as equivalent topic tags.
Separate dimensions
Opening: The editor distinguishes subject, format and experience level.
Hypothesis: Readers can ask a clearer retrieval question.
Risk: The interface should expose only dimensions that help actual tasks.
Define a boundary
Opening: The finance category specifies which budgeting and accounting resources belong.
Hypothesis: Different editors can classify consistently.
Risk: A broad label without examples may still collect unrelated material.
The team tests a small content sample and adjusts definitions before changing the full collection.
Take the questions into your next brief.
Use these prompts in your own notes. Nothing is submitted from this page.
Task
What set of resources is the reader trying to find?
Dimensions
Which attributes should remain separate?
Rules
What belongs and what does not?
Test
Where do editors or readers disagree?
Governance
Who approves category changes and updates existing items?
Watch for the wrong conclusion.
- Using tags as an unlimited keyword field.
- Mixing format and topic without a clear model.
- Renaming categories without checking their public collections.
- Mistaking a neat taxonomy for evidence that people can find content.
Your next action
Define five categories for one collection and test them on ten varied items before expanding the scheme.
Sources & scope
The framework, examples and suggested checks are original editorial synthesis. Hypothetical examples are not measured results. Source checking is not an independent audit or a guarantee of business outcomes. No named human reviewer is claimed.
- GOV.UK: Organise and group content →Primary publisher guidance · Supports: Supports organizing published content through topics and collections. The proposed taxonomy testing process is original editorial advice; GOV.UK administrative rules are not universal.