Competence Matrix: Build It in 30 Days

ISO 45001:2018 requires organizations to determine worker competence, yet many plants still treat the training file as proof that a high-risk task is under control. This guide shows how an EHS manager can build a 30-day competence matrix that separates attendance from verified task capability.

What you need before starting

A competence matrix is a live control document that connects each safety-critical task to the person authorized to perform it, the evidence proving competence, the verifier responsible for the sign-off, and the review date. Before the 30-day build starts, collect the task inventory, the legal and standard requirements that apply, the training records, the field verification notes, incident history for the last 12 months, and the names of supervisors who can confirm real work conditions.

The thesis is direct. A training record answers whether someone attended something, while a competence matrix answers whether the organization has verified capability at the task level. HSE explains that training must be supported by information, instruction and supervision, which is why a matrix that stops at classroom completion is still weak for maintenance, isolation, confined space, hot work or mobile equipment.

Across 25+ years leading EHS in multinationals and more than 250 cultural transformation projects, Andreza Araujo has seen the same trap in different sectors. In her Portuguese title A Ilusao da Conformidade, or The Illusion of Compliance, the useful warning is that compliance can look complete while work remains fragile. The competence matrix exists to close that gap between the file and the field.

Step 1: Define the safety-critical task list

Start with a task list, not a person list, because competence belongs to work exposure before it belongs to an employee profile. In the first 3 days, identify 20 to 40 tasks that can create serious harm if performed badly, such as energy isolation, confined-space atmospheric testing, forklift operation, scaffold acceptance, chemical transfer, line breaking, hot work fire watch, emergency eyewash inspection, and manual handling assessment.

This is where many matrices fail. They begin with generic course names, which makes the document look complete but leaves the riskiest work undefined. If the matrix says "maintenance safety training" instead of "verify zero energy before opening guarded equipment," no supervisor can tell whether the worker is competent for the task that can injure someone today.

Use existing risk material to build the list. Your risk register, JSA library, PTW system, incident reports and procedure usability reviews should all point to tasks where poor clarity or weak skill changes the exposure. The first deliverable is a spreadsheet with one row per task, not one row per course.

Step 2: Who is allowed to perform each task?

By day 6, every task in the matrix should have named roles that are allowed, conditionally allowed or not allowed to perform it. The decision should use 3 levels: authorized, authorized with supervision, and not authorized. A binary yes-or-no field hides too much, especially for new hires, contractors, temporary workers and people returning after long absence.

The counterpoint is important because organizations often confuse job title with competence. A mechanic may be fully competent for routine guarding removal and still not competent for a complex line break with chemical residue. A supervisor may be competent to approve a low-risk JSA and still need EHS support for a permit involving stored energy, height and simultaneous operations.

Write the rule in operational language. "Authorized with supervision" should name the supervisor level, the type of job allowed, and the limit that triggers escalation. This connects naturally to a field risk escalation matrix, because competence limits only protect people when the crew knows when to stop and ask for a higher decision.

Step 3: Convert training records into evidence categories

Between days 7 and 10, convert every record into an evidence category. Use at least 4 categories: formal training, field demonstration, supervisor observation, and recent task performance. A 2024 training certificate may prove exposure to content, but it does not prove that the worker can apply the content during night shift, production pressure or abnormal conditions.

This step prevents the matrix from becoming a prettier training archive. Andreza Araujo's work on safety culture treats behavior as something shaped by context, not only by intention. That matters here because a person can answer every classroom question correctly and still make a poor field decision when the procedure is unclear, the tool is missing or the schedule is compressed.

For each task, define the minimum evidence mix. Forklift authorization may require formal instruction, practical test, observed operation and a 12-month refresh. Confined-space gas testing may require instrument calibration knowledge, bump-test demonstration, atmospheric reading interpretation and a supervised entry simulation. The evidence must match the consequence.

Step 4: Set the verification method for each task

By day 14, every task should have a verification method that a competent verifier can repeat. Use 3 evidence types when possible: document review, field demonstration and scenario questioning. ISO describes ISO 45001 as an occupational health and safety management system standard, and the management-system logic is useful here because competence needs planned criteria, not informal confidence.

The common trap is letting the same person train, verify and approve without a check on quality. That may be acceptable for low-risk skills, but it is weak for critical tasks. A verifier should know the task, the failure modes, the procedure limits and the field cues that show whether the worker is applying the method or merely repeating words.

Build a short verification script for each high-risk task. For LOTO, ask the worker to identify all energy sources, prove isolation, explain stored energy, and state restart conditions. For manual handling, use the logic from manual handling risk assessment and ask the worker to identify load, posture, frequency, route, assistance and mechanical aids.

Step 5: How do you handle expired or weak evidence?

Expired or weak evidence should create a temporary restriction, not a quiet exception. By day 18, define what happens when a certificate expires, a worker has not performed the task for 12 months, an observation identifies a gap, or an incident suggests that the previous sign-off is no longer reliable. The matrix should make weak evidence visible within 24 hours.

Many companies avoid this because restriction creates staffing pressure. That pressure is exactly why the rule must be written before the shortage appears. If production can override expired competence because the shift is short, the matrix becomes a decoration. As Andreza argues in Sorte ou Capacidade, or Luck or Capability, counting on luck does not hold over the medium and long term.

Use simple status labels: current, restricted, suspended and reassessment required. A restricted worker may perform the task with named supervision. A suspended worker may not perform it until verification is complete. Reassessment required means the person may know the job but the evidence is too old, too narrow or contradicted by field findings.

Step 6: Assign owners and review dates

By day 21, each task row needs an owner, a verifier and a review date. The owner maintains the task standard, the verifier confirms worker capability, and the supervisor controls day-to-day assignment. Without these 3 roles, the matrix becomes an EHS spreadsheet that operations consult only after an audit finding.

Ownership should follow risk, not hierarchy alone. EHS can own the method, but operations must own who is assigned to the work. Maintenance must own technical skill for maintenance tasks. HR can support records, but HR should not decide whether a person is competent to isolate hazardous energy or accept a scaffold.

OSHA's safety management guidance emphasizes hazard prevention and control, which supports a practical rule for the matrix: competence ownership belongs where the hazard is controlled. Put the name of the owner in the row and give that owner a monthly exception report.

Step 7: Run a 10-person field sample

Between days 22 and 26, sample 10 people across shifts, employment types and risk tasks. Do not audit only the best crew or the easiest day shift. Select at least 2 supervisors, 2 contractors if contractors perform critical work, 2 newer workers, and 4 experienced workers whose competence has been assumed for years.

This is the proprietary test that separates a real matrix from a record project. If the sampled workers cannot explain the task limits, stop criteria, abnormal conditions and escalation route, the matrix is not ready. The problem may be training, but it may also be procedure usability, supervisor pressure, missing tools or unclear authority.

Use the sample to repair the system before launch. If 6 of 10 workers pass the document check but only 3 can demonstrate the task under field conditions, the issue is not individual weakness. The issue is that the organization has been measuring attendance while assuming capability.

Step 8: Publish the first version and control changes

By day 30, publish version 1 of the competence matrix with a controlled change routine. The first version does not need to contain every task in the company, but it must contain the critical tasks chosen in Step 1, the authorization status for each named person, the evidence category, the verifier, the owner and the next review date.

Do not launch it as a static file. Add a change trigger when a task changes, a new procedure is issued, a serious incident occurs, a worker transfers roles, a contractor scope expands, or a supervisor identifies a repeated field gap. The matrix should move with work, because stale competence data is worse than no data when leaders trust it blindly.

The 30-day target is useful because it forces a pilot rather than a perfect document. Start with the highest-risk 20 to 40 tasks, test them in the field, then expand. The article on building a safety trainer competency plan can support the next layer once the task matrix shows where trainers need deeper capability.

Competence matrix versus training matrix

The difference is practical. A training matrix asks who attended, when the record expires and which course is missing. A competence matrix asks whether the person can perform a defined task safely under known conditions, with evidence that a qualified verifier has checked.

Use both, but do not confuse them. The training matrix supports the competence matrix, while the competence matrix protects the work assignment decision. When the two disagree, the field evidence should force a management decision.

Conclusion

A 30-day competence matrix works when it starts with safety-critical tasks, tests real field capability, and makes weak evidence visible before the next assignment. The purpose is not to create another EHS file, because the real decision is whether a specific person should perform a specific task today.

If your organization needs to move from training attendance to verified capability, Andreza Araujo can support the diagnostic, field verification model and leadership routine. Start a conversation with Andreza Araujo.