Writing learning objectives that hold up

The test

An objective is good if you can write an exam question for it that has a defensibly right answer. That is the whole test, and it is stricter than it sounds.

"Understand our refund policy" fails it. What question would you write? Every candidate question is either trivially easy or secretly about something more specific. "Decide whether a given refund request needs manager approval" passes, because the question writes itself and the answer is checkable.

Verbs that do work

Objectives built on observable verbs survive the Assessor. Objectives built on internal states do not, because there is no way to test whether someone understands, knows, appreciates or is familiar with anything.

• Decide, choose, approve — for judgement calls with a right answer
• Identify, locate, name — for things that must be found or recalled under pressure
• Explain, justify — for reasoning you want a person to be able to reproduce
• Complete, follow, execute — for processes with a defined sequence
• Avoid: understand, know, be aware of, appreciate, be familiar with

One objective, one idea

Compound objectives are the most common problem at the outline gate. "Understand the refund policy and escalate correctly" is two objectives wearing one coat, and it produces a lesson that half-covers both.

The tell is the word "and". Split it, and you usually discover one half was well-formed and the other was vague — which you could not see while they were joined.

Coverage is not optional

Every objective needs at least one exam item pointing at it. When the Assessor cannot write a question for one, that is not a gap in the exam — it is a signal that the objective was never testable, and the Verifier raises it as a finding rather than letting it slide.

If you see an objective sitting at zero coverage in the dashboard, do not go looking for a question to bolt on. Go back and rewrite the objective.

Size them to the material

An objective should be answerable from what you actually supplied. If the sources say nothing about escalation paths, an objective about escalating correctly cannot be grounded, and the Author will either leave it thin or reach for something the Verifier will then flag.

The Architect sizes the outline to the corpus for this reason. When it proposes fewer objectives than you expected, the usual cause is not caution — it is that the material ran out.

What to do at the gate

Read the objectives, not the module titles. The titles are packaging; the objectives are the contract for what the exam will test and what the lessons must support.

For each one, ask the test question: could I write an exam item for this? If the answer is no for any of them, redirect the outline. It costs one credit and a few minutes now, against a full regeneration later.