The Skills Genome is the structure everything else in Workforce Intelligence stands on. A well-built one produces ratings that survive a disciplinary hearing. A vague one produces numbers nobody trusts.
This guide covers the decisions that matter when you build it.
The shape of the genome
Organisation
└── Framework (versioned, with a provenance record)
└── Domain (weighted grouping — e.g. "Mechanical Diagnostics")
└── Competency
└── Behaviour anchors (one per level, 1–5)A competency is then attached to a role with a required level, and that pairing is what produces a gap.
Write competencies as capabilities, not topics
The most common mistake is writing the training catalogue back out as a competency list. “Hydraulics course” is not a competency. “Diagnoses faults in hydraulic systems” is.
A good competency name completes the sentence “This person can…”. If it does not, it is a topic, and topics cannot be observed in a workplace.
Behaviour anchors are what make a rating defensible
For each competency you can define five behaviour anchors — what each level actually looks like in the workplace. This is what turns supervisor assessment from
“Score this person 1 to 5 on problem solving”
into
“Which of these have you actually seen them do?”
An example set for Problem Solving:
| Level | Behaviour |
|---|---|
| 1 | Recognises when something is wrong and escalates it accurately. |
| 2 | Solves routine issues using established procedures. |
| 3 | Diagnoses unfamiliar problems independently and to standard. |
| 4 | Resolves complex cross-functional problems and improves the process afterwards. |
| 5 | Creates organisational approaches that measurably improve how problems are solved. |
Anchors dramatically reduce rating drift between supervisors, and they give an employee something concrete to disagree with — which is the point.
Provide all five or none. A partial set leaves a supervisor’s rating form with missing rungs, which is worse than a form with none at all. The import will flag a partial set as an error for exactly this reason.
Evidence standards
Each competency can declare what evidence is acceptable, and how many independent sources are needed. This is where you decide, up front, what “we know this person can do it” means for your organisation.
For a safety-critical competency that might be a formal practical assessment plus a supervisor observation within the last twelve months. For a behavioural one it might be two independent observations.
Setting this now is much easier than arguing about it after the first cycle produces a number somebody dislikes.
Half-life: how fast the evidence ages
Not all capability ages at the same rate. A welding certification from eight years ago tells you less than a leadership observation from eighteen months ago.
Each competency has an evidence half-life. Defaults by type:
| Type | Half-life |
|---|---|
| Digital, Safety, Compliance | 12 months |
| Technical | 24 months |
| Behavioural, Leadership, Cross-cutting | 36 months |
| Values | 48 months |
Crucially, decay reduces confidence, not competence. Old evidence makes the module less certain of its conclusion; it does not invent a lower score. If a score does decline because of ageing evidence, the module says so explicitly rather than implying performance has dropped.
Recertification: the hard line
Half-life is soft. Recertification is hard.
If you set a recertification interval, evidence older than that is excluded from scoring outright — not merely down-weighted — and the competency is flagged as expired. This is the correct and defensible behaviour for a safety or compliance competency where an out-of-date certificate genuinely means nothing.
Use it deliberately. Applying recertification to a behavioural competency will simply erase your data every year.
Mark critical competencies
A critical competency escalates gap priority and will feed the capability risk index. Reserve it for capabilities where a shortfall is an organisational problem, not a development opportunity — safety response, regulatory sign-off, single-point-of-failure technical skills.
If everything is critical, nothing is.
Mapping: one skill, many frameworks
This is where integrating Workforce Intelligence with skillSYMS beats buying a standalone performance product. A single employer competency can be mapped simultaneously to:
- a skill in the platform knowledge graph,
- a unit standard,
- a QCTO module or programme outcome,
- an OFO code,
- an external assessment provider’s item.
Once mapped, verified competency scores project into the knowledge graph automatically, so the skills map and career pathway views light up without any extra work. Mapping is also what makes it possible to say “this qualification covers 72% of this role’s requirements” — the foundation of a bridging programme.
The Suggest mappings action does deterministic name matching against the skills taxonomy and unit standards. It suggests; a human confirms. Nothing is written until you do.
Bringing an existing framework
Large organisations usually arrive with a competency dictionary already. Use Imports → Competency framework rather than retyping it.
An imported framework always lands in its own framework record, marked as imported, with its source recorded. Provenance is never ambiguous: a licensed or customer-supplied framework must never read as one skillSYMS authored.
Nothing is written on upload. Rows are validated first, errors are listed per row, and only the valid rows are applied when you commit — so you can fix and re-upload just the problem rows rather than re-running the whole file.
Version and activate deliberately
Frameworks start as draft. Build the whole thing out, then activate. Activation is blocked while a framework has no active competencies, because an empty genome silently produces zero gaps for every role attached to it — the most misleading possible result.
Once roles are built on a framework and people have been assessed against it, treat changes the way you would treat changing an assessment instrument mid-cycle: version it rather than editing under people’s feet.