Step-by-Step Guide: How to Map a Skills Inventory Without Job
Skills-based hiring is the headline. Skills architecture is the work. Here's how to map a usable skills inventory across your existing workforce — without a six-figure consulting engagement…
Eighty-five percent of progressive employers say they've moved to skills-based hiring. Ask the middle managers responsible for the actual hiring, and they'll tell you the same thing: nobody has shown them how to actually build the skills inventory the strategy assumes exists.
This is the practical version. No big consulting deck. Just the steps that produce a working skills map you can use for hiring, mobility, and L&D inside a quarter.
Step 1 — Pick a taxonomy, don't build one
Do not invent your own skills language. Adopt an existing taxonomy (Lightcast, ESCO, or a vendor's) as a base. You can add 10–15% custom skills. Building from zero is where this project usually dies.
Step 2 — Extract skills from what you already have
Use an AI mapping tool to pull skills from existing job descriptions, performance reviews, project documentation, and Slack channels. This gives you a first-pass current-state map without surveying anyone.
Step 3 — Validate with a self-assessment, kept short
Ask employees to confirm or correct ~15 skills the system inferred about them. Not 80. People won't do 80. Fifteen, twice a year, is sustainable.
Step 4 — Add proficiency, not just presence
A skill on a profile means nothing without a level. Use a simple 4-level rubric (aware → practicing → proficient → expert) with one-line behavioral anchors per level. Don't over-engineer.
Step 5 — Map adjacency, the unlock for mobility
Skill X is adjacent to skill Y if 60%+ of the underlying tasks overlap. AI mapping tools do this well. This is what powers "you're 70% there for this role" — the killer feature for internal mobility.
Step 6 — Tag every open role with required vs. nice-to-have skills
If a role has more than 8 required skills, you don't have a role — you have a wishlist. Force the prioritization conversation.
Step 7 — Open the internal marketplace
Once you have profiles, role tags, and adjacency, you can stand up an internal mobility marketplace. Start with stretch projects, not full job moves. It's lower risk and builds the data you need.
- Hiring by title and years of experience.
- Promotion by tenure.
- L&D by role catalog.
- Mobility blocked by org chart.
- Hiring by demonstrated skill at a defined level.
- Promotion by proven skill acquisition + impact.
- L&D recommended by skill gap to a target role.
- Mobility driven by adjacency, not lineage.
Michael Polanyi's 1958 work distinguished explicit knowledge (what you can write down) from tacit knowledge (what you can only demonstrate). Job titles are explicit shorthand for tacit bundles of skills — and the bundle never matches across two people with the same title. A skills inventory that tries to capture all the tacit knowledge is doomed; one that captures the explicit, transferable subset is high-leverage. The trap is consultants selling 7-figure 'enterprise taxonomies' that try to do the impossible.
Add the half-life of skills research (IBM, 2024): the median half-life of a technical skill is now ~30 months and shrinking. A taxonomy built for a 5-year planning horizon will be 50% obsolete by year 3. Build for evolution, not perfection.
Unilever's 'Skills Cloud' is the textbook case — but the textbook misses the iteration. They started in 2017 with a single function, used AI to extract skills from existing CVs and project records, and only expanded to the enterprise after 3 years of refinement. Companies that copied the announcement without copying the patient build hit the wall at year 1 and quietly shelved the platform.
- Pick ONE business unit (50-200 people). Resist the urge to start enterprise-wide.
- Use AI to extract skills from existing artifacts: CVs, project descriptions, GitHub, JIRA.
- Validate with a 5-min self-attestation cycle per employee. Don't ask for more.
- Refresh quarterly, not annually. Skill drift is the failure mode.
- Wire the inventory into one concrete decision (internal mobility, project staffing) on day 1.
- Publish a quarterly 'inventory health' score: % current, % validated, % wired to a real decision.
- Plan to expand to BU #2 only after 6 months of demonstrated value in BU #1.