CCAO-F questions and answers — Domain 1: Prompting and Task Execution

8 practice questions on Prompting and Task Execution, the domain weighted at 14% of the CCAO-F exam. Every answer explains why the distractors fail — that is the part that transfers to questions you have not seen before.

Question 1

A marketing coordinator asks Claude to "write a product announcement" and gets back generic copy that could describe any product. What is the most effective next step?

1. Regenerate the response two or three times and pick the best one
2. Supply the audience, the specific benefit, the required length and an example of a previous announcement
3. Switch to the most capable model available
4. Ask Claude to make the copy more exciting

Answer: B. Generic output is a symptom of a prompt with no specifics in it. Audience, purpose, format and a sample of house voice are the four things a colleague would have needed, and supplying them fixes the cause. Regenerating re-rolls the same weak prompt, a heavier model has nothing extra to work with, and "more exciting" adds adjectives rather than substance.

Question 2

Which request is most likely to produce a usable first draft?

1. "Summarise this document."
2. "Summarise the attached vendor contract in 200 words for our finance director, focusing on payment terms and termination clauses, and flag anything unusual."
3. "Give me a professional summary of the attached file."
4. "Read the attachment and tell me what you think."

Answer: B. It names the source, the length, the reader, the aspects that matter and what to surface. The other three leave every one of those decisions to be guessed, so the output has to be generic to be safe.

Question 3

An analyst asks Claude to analyse a 40-page research report and simultaneously produce a board summary, an all-staff email and a slide outline. All three come back thin. What is the best approach?

1. Ask for each of the three deliverables in a separate conversation with the report attached each time
2. Extract and confirm the key findings first, then generate each deliverable from the confirmed findings
3. Ask for the three deliverables again but request more detail in each
4. Split the report into four parts and repeat the request for each part

Answer: B. Sequencing puts a checkpoint where it matters. If the extraction is wrong, you catch it once, before it has propagated into three documents. Separate conversations lose that shared, verified base; asking for "more detail" does not address why attention was spread thin; splitting the report fragments findings that span sections.

Question 4

Which technique gives the biggest improvement in output consistency for a non-technical user?

1. Writing longer prompts
2. Providing one or two examples of the output you want
3. Asking Claude to be more accurate
4. Using formal language in the prompt

Answer: B. Examples communicate format, tone, depth and structure at once, and far more precisely than description. Length alone does not help; instructing accuracy is not a mechanism; formality of phrasing does not change output quality.

Question 5

A user wants Claude to analyse survey responses without inventing themes that are not present in the data. Which instruction helps most?

1. "Be accurate and do not make anything up."
2. "Use only the attached responses. Quote the specific responses supporting each theme, and say so if a theme is not clearly supported."
3. "Take your time with this analysis."
4. "Use your knowledge of survey analysis best practice."

Answer: B. It restricts the source material, requires evidence for each conclusion, and gives explicit permission to report weak support. That combination is what makes the output checkable. A general instruction to be accurate provides no mechanism, and inviting general knowledge is the opposite of what was needed.

Question 6

When is asking Claude to explain its reasoning before answering most worthwhile?

1. Every time, as a matter of routine
2. When the task involves analysis or a recommendation you will need to evaluate
3. When you want a shorter response
4. When reformatting text you have supplied

Answer: B. Reasoning is worth its cost on analytical work: it improves the answer and lets you see where the logic went wrong. On a reformatting task it adds length and delay for nothing, and it will not shorten anything.

Question 7

A first draft is close but too formal for an internal audience. What is the most efficient response?

1. Start a new conversation with a better prompt
2. Ask Claude to rewrite the same draft in a conversational register for an internal team, keeping the structure
3. Ask Claude to try again
4. Rewrite it yourself, since the tone is wrong

Answer: B. Targeted correction keeps everything that was already right and changes only what was wrong. Starting over discards the working structure and tends to reintroduce problems you had already fixed; "try again" does not say what to change.

Question 8

Which task type carries the highest accuracy risk and therefore needs the most careful prompting?

1. Reformatting a table you pasted in
2. Rewriting your own paragraph more concisely
3. Producing factual research about a market you have not supplied documents for
4. Translating a document you attached

Answer: C. Risk concentrates where Claude must supply facts from its own knowledge rather than transform material you gave it. Reformatting, rewriting and translating all operate on supplied text, so there is little for the model to invent.

← Previous: Cheat sheet  |  Next: Questions & answers — Domain 2 →