HR software · Updated October 2026
A skills taxonomy is a controlled list of the skills your organisation cares about, grouped into families and sub-families, with one agreed name and description for each skill. It gives your HRMS, learning system and recruiters a shared vocabulary, so “Python”, “Python 3” and “python scripting” stop living as three separate entries.
These three get mixed up in vendor decks, so it helps to pin them down:
| Term | What it is | Question it answers |
|---|---|---|
| Skills taxonomy | A hierarchy: family, sub-family, skill | What do we call this skill, and where does it sit? |
| Skills ontology | The taxonomy plus typed relationships: prerequisite of, close to, used in role X at level Y | If someone knows SQL, what else are they close to? |
| Competency framework | Behaviours and capabilities expected at each role or level | What does good look like for a team lead here? |
Usage is not standardised, so ask any vendor to show you its actual data model. A competency framework tends to be small and behavioural (“drives accountability”). A taxonomy is large and granular (“GST return filing”, “CNC lathe setting”, “Java Spring Boot”), and it is what software needs to match people to work.
Do not build from a blank sheet. Public frameworks give you a skeleton: the EU’s ESCO classification lists about 13,900 skills against some 3,000 occupations, and in India, Sector Skill Councils publish Qualification Packs made up of National Occupational Standards, each tied to an NSQF level. A plant HR team can borrow NOS wording for shop-floor roles; an IT services firm usually starts from its own project staffing tags.
Then keep it lean. A few hundred well-defined skills that managers actually use beat ten thousand imported ones nobody maintains. Name an owner per skill family (often the function head), merge synonyms every quarter, and add proficiency levels only where they change a decision. The usual failure is a taxonomy that is accurate on launch day and stale a year later, because nobody owns new skills such as GenAI tooling, or because it drifts away from your job families.
Once the taxonomy exists, measuring who lacks what is the job of a training needs analysis; the taxonomy only supplies the labels.
Only if you want to match people to specific work, such as staffing projects, running an internal talent marketplace or screening on skills; a behavioural framework is too coarse for that.
Many HR platforms ship a library you can switch on, which saves months, but you will still prune it heavily and add company-specific skills, so budget HR and manager time either way.
Let employees suggest, but have a named curator approve, otherwise duplicates creep back in within weeks and your skills-based hiring filters start missing people.
Granular enough that two people with the same tag could swap on a task, and no finer; “Excel” is too broad for a finance analyst, “VLOOKUP” is too narrow.
A taxonomy also feeds development planning, so see how it plugs into an individual development plan for each employee.