A skill taxonomy is the controlled vocabulary that AI HR applications use to map workforce skills, candidate qualifications, learning outcomes, and role requirements to a shared reference frame. The taxonomy is the buyer’s institutional asset across every AI application in HR.

The structural definition

The taxonomy is a hierarchical structure of skill entries, each with a unique identifier, a primary label, a set of synonyms and aliases, a parent category, and a competence level. The structure follows information science conventions — every entry has exactly one parent, synonyms are not duplicates, and the version history is preserved. The 5 AI Applications Revolutionizing HR & Recruiting — Complete 2026 Guide expands the role of the taxonomy across the five AI applications.

What the taxonomy enables

The taxonomy enables five outcomes — consistent skill extraction across resumes, consistent skill mapping in learning content, consistent skill measurement in performance reviews, consistent gap analysis in skill analytics, and portability across AI vendors. The compound effect of shared vocabulary across applications is the strategic argument for taxonomy ownership.

Where the taxonomy lives

The taxonomy lives in version control as a structured file (JSON or YAML). Every change runs through a pull-request workflow. Every release ships with a version number, a changelog, and a disparity-impact review. The taxonomy is treated as code, not as a content artifact. The 12 essential HR integrations guide covers the integration patterns that consume the taxonomy.

The base taxonomies

ESCO covers European hiring and most technical roles globally. O*NET covers US service, trade, and clinical roles. Lightcast and SkyHive license commercial maintained taxonomies. The organization picks one base and extends with 200 to 1,500 role-specific entries. The 8 HR metrics guide covers the workforce analytics layer.

Who owns the taxonomy

Recruiting operations owns the day-to-day maintenance — adding skills, retiring obsolete ones, reviewing synonyms. The data team owns version control and validation tests. The recruiting director owns final sign-off on each quarterly release. Legal owns the disparity review when taxonomy changes are proposed. The report design for strategic impact guide covers the reporting on taxonomy health.

The taxonomy lifecycle

A new skill enters through a proposal — a recruiter or hiring manager submits the skill with a definition and example resume language. The data team validates against existing entries and proposes the mapping. The recruiting director approves. The entry ships in the next quarterly release. The full cycle runs 30 to 60 days. The Make.com HR productivity guide covers the orchestration that supports the lifecycle.

FAQ

How is the taxonomy different from a competency framework?

A competency framework defines what good performance looks like in a role; a skill taxonomy provides the controlled vocabulary that maps to performance, hiring, and learning data. The two complement each other — the competency framework is the strategic artifact; the taxonomy is the operational data structure.

Can two business units share a taxonomy?

Yes — the recommended pattern. The base taxonomy is shared; each business unit extends with role-specific entries. The shared base maintains consistency; the extensions handle specialization.

What if we change AI vendors?

The taxonomy transfers cleanly to the new vendor. The change is straightforward because the taxonomy is owned, not rented. The 9 disaster recovery technologies guide covers the broader architecture.