Standard work
ILUO training matrix
ILUO training matrix is licensed CC BY 4.0. Attribution: Katafacts (katafacts.com).
Customise with AISkill file ↓1 · What it is
What it is
An ILUO training matrix tracks real capability, not just who's been shown something once: I (in training), L (learned), U (able to work unsupervised), O (able to train others). Describe the process area and the team, and a real skill list gets drafted — but who's actually at which level is never guessed. You record that yourself, and the matrix honestly flags any skill where nobody has reached U or O yet.
2 · When to use it
When to use it — and when not to
Use it when
- A process depends on more than one person being able to do it — you want to know honestly where that's actually true and where it isn't.
- You want to catch a single-point-of-failure skill before it becomes a problem, not after someone's out sick.
- You've just written a work instruction and want to track who's actually been trained on it.
Not when
- Only one person will ever do this task — a training matrix for a team of one adds process without adding redundancy.
- You don't have real information about who can do what — a matrix filled with guesses is worse than no matrix, since it looks like coverage that isn't real.
- The skill list is still unsettled — write the work instruction first, then track training against it.
3 · How to fill it in
How to fill it in
- Team
- Who's on the team? Skill levels are recorded directly, not AI-assigned.
- Skill matrix
- Skills drafted from your process area — record each person's real level (I/L/U/O) per skill. A skill with nobody at U or O is flagged, not hidden.
- Narrative
- Drafted from the skill list actually built.
4 · What good looks like
What good looks like
The example below trains the same renewal-risk support process the work instruction documents, with the same team. The one skill nobody's past in-training on lines up exactly with the step the work instruction itself flagged as still missing a key point — the same real gap, visible from two different artifacts.
Same example, as a downloadable xlsx workbook.
Download .xlsxILUO training matrix
Renewal-risk ticket handling — training matrix
Support — renewal-risk queue
Priya Nair · 2026-09-25
Team
- Priya Nair
- Marcus Lee
- Dana Whitfield
Skill matrix
4 skills · 3 people trained · I = in training, L = learned, U = unsupervised, O = trains others
| Skill | Priya Nair | Marcus Lee | Dana Whitfield |
|---|---|---|---|
| Confirm the ticket is tagged renewal-risk | U | L | I |
| Route to the on-call renewal-risk responder | O | U | — |
| Send the initial-response macro within the SLA | U | — | I |
| Log the resolution path for the weekly metrics reviewNo one unsupervised | — | I | — |
Narrative
Three of four skills have at least one person who can work them unsupervised — Priya can train others on both ticket confirmation and routing. Logging the resolution path is the one skill with nobody past in-training yet, which lines up with that same step still being flagged with no key point in the work instruction — worth closing both gaps together.
5 · Common mistakes
Common mistakes
Filling in levels based on a guess or a title, not actual demonstrated ability.
The matrix only has value if it reflects reality — an optimistic matrix hides exactly the single-point-of-failure risk it exists to catch.
Treating 'L' (learned) as good enough coverage.
Learned isn't the same as able to work unsupervised — a skill where everyone is still at L has no real redundancy yet, even though it looks staffed.
Building the matrix once and never updating it as people leave or skills change.
A stale matrix gives false confidence — treat it as a living record, updated whenever training status actually changes.
6 · What it connects to
What it connects to
upstream
Work instruction
A skill in this matrix is only trainable against a real work instruction — build that first if it doesn't exist yet.
downstream
7 · Where AI helps
Where AI helps
Judgement — stays yours
- Deciding what level someone has actually reached on a given skill
- Deciding which untrained skill to prioritise closing first
Analysis — AI helps
- Drafting a specific, trainable skill list from a rough process description
- Drafting the narrative from the skill list actually built
Drudgery — automated
- Counting skills and people trained
- Identifying which skills have nobody at U or O
- Exporting to xlsx in the house matrix format
9 · Rate this kata