2.1 · Lesson 1 of 6
Accuracy and Completeness
What you need to know
Domain 2 is the heaviest CCAO-F domain (21%). Lesson 2.1 starts evaluation before polish: did the output answer what was asked, and did it leave out anything material? A fluent reply that quietly skips a caveat, an exception, or half of the request is still a failed response in claude.ai work.
Check the ask first
Put your original request next to the output and tick each part that was addressed. Do this before you edit tone or length. Coverage failures are cheaper to catch before you invest in rewriting.
Omission is a failure mode
Summaries often drop the exception, the deadline, or the caveat that changes what a decision-maker would do. When you check completeness, ask what a reader would regret not knowing. Completeness includes conditions and limits, not only the headline answer.
Do not grade by tone
Confident, fluent prose feels trustworthy. Evaluate whether claims are present, correct, and complete. When a source exists, confirm that key numbers, dates, and names are present and unchanged. Asking Claude what it assumed or left out can surface gaps, but it is not a substitute for your own check.
Worked example
A manager asks for a summary of a new leave policy. Claude returns a clean three-paragraph brief. Every sentence matches the policy text — but the effective date and the carve-out for contractors are missing. A reader would act on the wrong timeline and the wrong population.
The exam-shaped judgment: treat the omission as a defect, rebuild from a source checklist, and request a revision that restores the missing items — do not accept “shortening” as an excuse.
Exam traps
Every sentence is true, so the output is complete
Dropped exceptions, limits, and deadlines are failures even when the remaining prose is accurate.
Grade the answer by how confident it sounds
Tone is not accuracy. Tick the ask and compare against the source.
Ask Claude what it left out and treat that as proof
Asking Claude to list omissions is a useful lead. It does not replace your own checklist from the source.
Fix style before checking whether the ask was answered
A beautiful answer to half the question is still incomplete. Check coverage first.
Practice scenario
A policy summary is fluent and every sentence in it is true. It leaves out the deadline and the exception for small firms. How should an analyst judge it?
Build exercise
Audit a Summary for Accuracy and Omissions
2.1 · Beginner · ~25 min · claude.ai
What you will practise
- How to check whether the output answered what you asked.
- How to spot a material omission, such as a dropped caveat or deadline.
- Why a fluent summary can still be wrong.
- How to build a short checklist from the source.
Full steps and the skill check are on the Domain 2 exercises page. Keep your source checklist open while you work.
Open Domain 2 exercisesPath: /learn/claude-associate/exercises/2-output-evaluation-validation
Sources
- CCAO-F Exam Guide v1.0 — Domain 2 task statement 2.1 — Accuracy and Completeness objectives
- evggzzz/ccao-f-guide — task 2.1 Knowledge/Skills — answered-the-ask and material omission (rewritten)
- Amey-Thakur associate-foundations notes — evaluation flowchart — distilled into site voice
- docs/ccao-f-implementation-plan.md — Domain 2 lesson map