Competency systems: linking who is qualified to what the schedule allows them to run

A wall matrix is a picture; gating is a control

Most plants have a grid of names and tasks with coloured squares, maintained by somebody with other priorities. It informs planning at best. The step that changes behaviour is connecting qualification to work assignment, so an unqualified operator cannot start a restricted operation at a terminal and a scheduler cannot allocate them to it. Sites resist this because it exposes how thin coverage really is on certain processes, and because it will stop production on a bad day. That exposure is the argument for doing it, not against it.

Defining competence so it can actually be assessed

Competence has to be written at the level of a task somebody can be watched doing, not as a course title or a job description. That means a statement of what the person must be able to do, the conditions under which they must do it, the standard applied, and who is qualified to make the judgement. Assessor competence is the part most often skipped, and it determines whether the record means anything. Where every supervisor can sign off any skill, the assessment records converge on attendance, which is a different and much weaker claim.

Expiry and the triggers that force reassessment

Time-based renewal is the crudest trigger and the only one most systems implement. More useful ones are event-driven: the process or its instruction changed, the person has not performed the task for an extended period, a quality escape or incident occurred on their work, or a trend in their output suggests drift. Building those triggers in requires links to change control, to scheduling data and to quality records, which is why competency systems deliver far more when connected than when run as a standalone training register.

The evidence an auditor actually asks for

Auditors do not ask whether training exists. They pick a batch record, identify who performed a critical operation, and ask to see that person's qualification for it on that date, together with the instruction version they were trained against. Two links fail regularly: the record shows training against a document that has since been revised with nobody reassessed, and the assessment is a signed attendance sheet rather than a demonstration of capability. Both are fixable in the data model and neither is fixable retrospectively once the audit has started.

Keeping it current as people move around

Records rot through movement. Transfers between lines, promotions, long absence, agency staff who return under a different reference, contractors working under their employer's training system, and cross-site recognition of qualifications that were assessed to different standards. Each needs a rule decided in advance rather than a judgement made under pressure by a supervisor short of people. The maintenance load is real and continuous, so name an owner per area and accept that a competency system nobody maintains is more dangerous than a wall chart, because people believe it.

Frequently asked questions

Should the system physically prevent an unqualified person starting a job?

For operations where an error causes harm, a regulatory breach or an expensive scrap event, yes, and accept that it will occasionally stop production. For everything else a warning with supervisor override and a logged reason gives most of the benefit at far lower operational risk. What matters is deciding which category each operation falls into deliberately, rather than applying hard blocks everywhere and watching supervisors acquire a habit of overriding them.

How do we handle agency and contract workers?

They need the same task-level qualification records as employees for any work they perform, held by you rather than assumed from their agency. Practically this means an induction and assessment path short enough to be workable for short assignments, and a record that persists so a returning worker is not reassessed from scratch. The common failure is a temporary worker performing a restricted operation because the shift was short-handed, with no record existing anywhere afterwards.

Does completing a course prove competence?

No, and treating it as proof is the most widespread weakness in these records. Attendance shows exposure to information. Competence is demonstrated performance of the task under normal conditions, judged by somebody qualified to judge it. Keeping both in the system is right, but they should be distinguishable fields rather than merged into a single completion flag, because an auditor will ask which one a given entry represents and the answer needs to be visible.