Implementation Guide
Skills Matrix.
A skills matrix is a visible grid showing who can perform which station or skill, and to what level of qualification, for one team or area. Track demonstrated capability at the process, not training attendance: a signed attendance sheet says someone sat in a room; a matrix level says who can run the station alone, at takt, to standard. The matrix is also a plan: the gaps it makes visible are next quarter's cross-training schedule.
What a skills matrix is
A skills matrix is a grid with the stations and critical skills of one area on one axis, the people of the team on the other, and a qualification level in every cell. In one glance it answers the two questions every supervisor faces daily: who can run each station right now, and where the team is one absence away from trouble.
The matrix has two jobs. Today, it is an operating tool: the morning staffing decision, the vacation approval, the answer to two people calling in sick. Over the quarter, it is a development plan: every gap against the team's coverage rules becomes a cross-training entry with a trainer and a date. Most plants have training records. Far fewer can answer, at 6:10 on a Monday with two people out, who is qualified to run final test alone.
One definition carries the whole method: a skill on the matrix is the ability to perform a documented standard, assessed by observation at the process. That anchor is what makes a level mean the same thing on both shifts, and it is what separates a skills matrix from a list of courses people once attended.
Where the skills matrix sits in the roadmap
On the TeamGuru deployment roadmap the skills matrix is its own practice in the Scale stage: skills matrix and capability development. It sits there because systematic capability building is what lets a plant spread its management system without diluting it. Do not wait for the Scale stage to start, though. A basic matrix earns its place as soon as standard work exists, because standards define the skills worth assessing. And the daily system consumes it every morning: the tier 1 coverage decision in lean daily management is a read of the matrix.
- Before Daily Management
- You are here Skills Matrix & Capability
- In parallel Obeya & Management Reviews
- After Standardize & Scale
Why attendance tracking fails
The default in most plants is a training matrix in an HR system: courses completed, certificates issued, signatures collected. It exists because audits ask for it, and it answers the question auditors ask: was this person exposed to the training. It cannot answer the question production asks: can this person do the work. The two questions diverge in three predictable ways.
- Attendance is not performance. A signed sheet says someone sat in a room for four hours. It says nothing about whether they can run the station at takt, alone, across variants, on a bad day. The gap between those two states is weeks of supervised practice, and the attendance record cannot see any of it.
- Skills decay and the record does not. Someone trained three years ago who has not touched the station since still shows as trained. The paper stays green while the capability quietly expires.
- It hides coverage risk. A report saying 100 percent of the team is trained looks safe, while in reality one person is the only one who can actually run a critical station independently. The average hides exactly the single point of failure a supervisor most needs to see.
Tracking demonstrated capability costs more than filing certificates: someone has to observe real work, record evidence, and revisit levels when they expire. That cost is the difference between wall decoration and a tool that decides staffing. Keep the training records for the auditors. Run the line on the matrix.
The four qualification levels
Four levels are enough. More granularity feels precise and assesses nothing extra. What gives the scale its meaning is the evidence column: a level is awarded when the evidence exists, at the process, and not before.
| Level | What it means | At the station | Evidence required |
|---|---|---|---|
| 1 | In training | Works at the station only under direct supervision, learning the standard step by step. Not counted in coverage. | Training plan opened with a named trainer. Safety induction for the station complete. The standard walked through at the process, with the trainer present for every cycle. |
| 2 | Performs to standard with support | Runs the station correctly, but not yet at takt in all conditions or not across all variants. Needs a level 3 or 4 within reach. | One observed run covering every step of the standard with no deviations, self-checks included. Cycle time within roughly 20 percent of takt. Knows the station's escalation rule. |
| 3 | Performs independently at takt | Runs the station alone for a full shift, at takt, across all product variants. Handles routine deviations and escalates the rest. This is the level coverage planning counts. | Observed runs at takt across variants. No quality escapes attributable to the station over an agreed window, 30 days is typical. Correct escalation demonstrated live or walked through. |
| 4 | Trains and audits others | Everything in level 3, plus qualified to train, assess and audit others at this station. First voice when the standard is revised. | Train-the-trainer complete. Has brought at least one operator to level 2 or higher. Audits performed to schedule, with findings that hold up. |
Some companies draw the same scale as quarter-filled circles on the wall. The notation does not matter. The evidence does, and so does the rule that only level 3 and above count as coverage.
How to build a matrix the line actually uses
A first matrix for one team takes about two weeks: a day to define the skills, a week or two of observation to baseline honestly, and an hour to make it visible. The sequence matters more than the tooling.
List skills from standard work, not job titles
Each row of the matrix is a standard someone can be assessed against: a station, a setup, a critical check. If no standard exists for it, there is nothing objective to assess, so write the standard first. Job titles like machine operator hide exactly the differences the matrix exists to show.
Scope it to one team
One area's stations and people on one grid, plus a few shared skills the team really uses, forklift, first response, quality gate. Keep it between five and twelve skills. A plant-wide matrix maintained centrally is a report; a team matrix on the wall is a tool.
Baseline honestly, by observation
A qualified assessor watches each person work against the standard and scores what they see, not what the training file says. Expect the baseline to come out lower than everyone assumed. That is not an insult, it is the finding, and it is the reason the exercise pays for itself.
Award levels at the process, with evidence
The supervisor and a qualified trainer assess together, at the station, against the written criteria for each level. The evidence gets recorded: what was observed, when, by whom. A level nobody can trace to an observation is an opinion with a number on it.
Make it visible in the area
The matrix hangs next to the team board, or lives on the same screen, where the people it describes can see it. Keep it out of the annual pay conversation, at least in the first year, or every level becomes a negotiation instead of an assessment.
The honest baseline is the step teams most want to skip, and the one that decides whether the matrix is trusted. If the first version of the grid flatters everyone, the first absence that goes wrong exposes it, and the tool is dead in a month. Assess what you can observe, and let the wall show a harder picture than the training files did.
Worked example: one cell, six operators
The numbers are illustrative but internally consistent, using the same 450-person components manufacturer that runs through the transformation roadmap: 456 units per day, two shifts, a takt of 118 seconds. The assembly cell runs five stations with six operators per shift. This is the day-shift matrix after an honest baseline by observation.
| Station | Op A | Op B | Op C | Op D | Op E | Op F | Level 3+ |
|---|---|---|---|---|---|---|---|
| Sub-assembly | 4 | 3 | 2 | 3 | 1 | 2 | 3 |
| Welding | 3 | 4 | 1 | - | 2 | - | 2 |
| Main assembly | 3 | 3 | 3 | 2 | 2 | 1 | 3 |
| Fitting and torque | 2 | 3 | 4 | - | - | 2 | 2 |
| Final test coverage risk | - | 2 | 3 | - | - | 1 | 1 |
The cell's coverage rule is a minimum of two operators at level 3 or higher for every critical station on every shift. Four stations pass. Final test fails: Op C is the only person who can run it independently, and Op A, the cell's most versatile operator, cannot run it at all. Every time C is absent, final test runs at level 2 with support, which pulls the team leader off coaching for the day and puts the last check before shipping in its weakest configuration.
The matrix turns that risk into two entries in next quarter's cross-training plan. Op B moves from level 2 to level 3 on final test: two protected practice blocks per week with C alongside, assessment at the process at the end of the quarter. Op C moves from level 3 to level 4: train-the-trainer, so the cell owns its own trainer instead of borrowing one. Both entries carry a name and a date, and both are reviewed monthly like any other action. That is the whole method: the matrix finds the risk, the plan closes it, the review checks that it happened.
Using the matrix to plan coverage and cross-training
A matrix without decision rules is a mood board. Agree the coverage rules once, write them next to the grid, and let every review compare the grid against them. A workable starting set:
- Minimum two operators at level 3 or higher for every critical station, on every shift. One is a single point of failure with a name.
- Every operator at level 3 or higher on at least two stations, so rotation and absence coverage are real options, not favors.
- At least one level 4 per station somewhere on the site, so the team can grow its own capability instead of borrowing trainers.
- Every new hire reaches level 2 on a first station within an agreed window, typically four to eight weeks, and the matrix shows who is overdue.
Cross-training costs capacity, which is why it dies when it is left to quiet moments: there are no quiet moments. Schedule it like maintenance, in protected slots on the production plan, with the trainee paired to a level 4 and the practice counted as work, not as a favor. A plan that only runs when the day goes well is not a plan.
The cadence: the supervisor owns the matrix and updates it the day a level is awarded or expires. Once a month, the matrix goes to the site level of the tier meeting cascade: gaps against the coverage rules, progress of the cross-training plan, requalifications coming due. The assessments and qualification audits that keep the levels honest belong in leader standard work, so they happen on a routine instead of a memory.
Versatility, rotation and growth paths
The same grid that manages risk also manages monotony. Rotation across stations breaks up repetitive strain and keeps attention fresh on quality checks, but rotation is only safe when the matrix shows real level 3s at every position in the loop. Rotation also protects the levels themselves: a skill that is never practiced is a level 3 quietly becoming a level 2 that the wall still shows as green.
For the people on it, the matrix is the most concrete growth path the plant offers: the route from level 1 to level 4 on one station, and from two stations to five, is visible, assessable and public. Level 4 deserves to be treated as a recognized expert role, because it is where the plant's next trainers and team leaders come from. A simple trend metric helps: the average number of stations per operator held at level 3 or higher. Watch its direction quarter over quarter, and resist turning it into a target, because averages are exactly what the matrix exists to see past.
Common failure modes
Skills matrices rarely fail as an idea. They fail as a practice, in four recurring ways, and each one has a countermeasure that works better as a standing rule than as a one-time fix.
The HR-owned spreadsheet
The matrix lives in a shared drive, is updated once a year for the audit, and nobody at the line has ever seen it. Ownership belongs to the team and its supervisor: the matrix hangs in the area and changes the day a level changes, or it is a record, not a tool.
Levels inflated to pass audits
The week before a customer audit, everyone becomes level 3. Inflated levels turn coverage planning into fiction and surface as quality escapes on the station someone could supposedly run alone. The countermeasure is evidence per level, assessment at the process, and spot audits of qualifications like any other standard.
Levels that never expire
A level awarded in 2022 still counts, though the operator has not touched the station since. Skills decay; a matrix without requalification cannot see it. Give every skill an expiry, shorter for critical and rarely performed skills, and let the matrix flag anyone overdue.
A matrix without a training plan
The grid is current, honest and visible, and the same gaps sit in it quarter after quarter. A matrix that does not drive a schedule is a well-maintained picture of a problem. Every gap against the coverage rules becomes a dated cross-training entry with a named trainer, reviewed monthly.
What happens next
On the roadmap, the matrix is how the capability development practice stops being an intention. Every practice built in the earlier stages, boards, standard work, problem solving, audits, needs people qualified to run it, and scaling any of them to a second area or a second site is a capability question before it is a rollout question. Mature plants extend the same grid beyond production skills to management-system skills: running a tier meeting, coaching a problem-solving case, auditing a standard. The mechanics are identical: a defined standard, observed performance, evidence, expiry.
The hard part of sustaining a matrix is bookkeeping: levels, evidence, expiry dates and training actions across teams and shifts, changing weekly. The wall version is right for the team and wrong as the only copy, because paper cannot flag an expiring qualification or connect a gap to a scheduled action. This is where TeamGuru fits the practice: the skills management use case keeps the matrix live, with evidence behind each level, expiry dates that flag themselves, and every gap linked to a dated training action, so the morning coverage decision and the quarterly capability review read the same data the roadmap says to close.
