Taxonomy and Tag Audit (Consolidation and Value Review)

Description

Audit an existing tag set or taxonomy: consolidate sprawling terms into a controlled list and check what its term relationships and orderings imply about whose knowledge is default.

Free tagging saves effort at first but sprawls into misspellings, near-duplicates and one-off tags that split content streams and weaken findability and measurement. The audit exports every tag with its usage count, flags unused, single-use and duplicate terms, classifies the rest by role (topic, series, author, category, joke) and merges topics into a controlled vocabulary while moving other roles into their own metadata fields. Martin pairs this with a value review: every term-to-content relationship is a judgement, so for contested terms the team lists all plausible matches, asks what the draft ordering privileges (culture, language, gender, discipline) and neutralises it, for example with relevance tiers and randomised order within each tier. The rigour comes from working from the full usage data rather than impressions, and from recording the rationale for each relationship so it can be revisited with diverse reviewers after launch.

Attributes

Loss type
Active
Effort
Days

Steps

  1. Export every tag or taxonomy term with its count of use and the content it is attached to.
  2. Flag unused, single-use, misspelled and near-duplicate terms; in Martin's example 6,152 of 8,182 tags had been used once and over 400 never.
  3. Classify each surviving term by role (topic, series, author, category, humour or other) and move non-topical roles into dedicated metadata fields.
  4. Merge topical terms into a controlled list and retag the content.
  5. List the term-to-content relationships for contested terms, note which matches the draft favours and question what that ordering implies about culture, language, gender or discipline.
  6. Group results into relevance tiers and randomise or otherwise neutralise order within tiers where no group should be privileged.
  7. Set up governance for proposing, adding and retiring terms, with regular editorial review and a post-launch check by diverse reviewers.

Sources

  • Everyday Information Architecture (Martin, 2019)

Also known as

  • Tag audit and folksonomy consolidation
  • Tag clean-up
  • Tagging system audit
  • Taxonomy value and bias review
  • Taxonomy relationship bias check
  • Tiered and randomised result ordering

Categories

  • Content
  • Information Architecture & Content