# Placing a person on the levels

The method lives or dies on honest assessment — and the levels are built to make honesty achievable. They're defined by *observable behaviour*: not "how good is this person?" but "what happens when they do the work?"

The tells, level by level:

**Beginner** — the work gets done, but with help or documentation *along the way*, not just at the edges. Fresh from a course or a book.

**Confident** — most tasks delivered independently, reaching for docs only on the less common parts. The work gets done, by them, routinely.

**Expert** — almost everything handled without help; lookups only for obscure cases. The strongest single tell: **they can teach it.** Someone who can walk a colleague from beginner to confident in a skill is an expert in it.

(And the standing footnote: **even experts look things up.** Skills evolve; needing materials at any level is normal and expected. For small no-level skills, none of this applies — they're a plain yes/no.)

## The scale is fixed; the sources vary

Because the tells are behavioural, *anyone who has seen the work* can read them — which is what lets the same scale accept assessments from many sources:

- **Manager assessment** — the default in most professional environments; managers see delivery month after month, which is exactly what the tells describe.
- **Self-assessment** — the natural default for personal use, and a valuable input everywhere; nobody knows better how often help was needed along the way.
- **Peers and 360 feedback** — colleagues who work alongside someone often have the clearest view of all, especially for skills a manager rarely observes directly.
- **Credentials and certifications** — solid evidence for beginner (the basics are demonstrably known) and useful signals beyond; they certify knowledge, so pair them with delivery for the higher levels.
- **Teaching-back** — the built-in test that needs no interpretation: teach it, and expert is demonstrated, not claimed.

Which sources carry the decision is a cultural and organisational choice — the method doesn't mandate a mix. What it insists on is only this: every source answers the *same* behavioural question against the *same* scale. That's what keeps a level meaning one thing no matter who assessed it — and it's why disagreement between sources is useful rather than awkward: it's a small, concrete conversation about one skill, settled by looking at the work.

![](https://docs.skillforests.com/uploads/images/gallery/2026-07/method-05-01.png)