A request to analyze 80 interview transcripts, find themes, compare roles, and draft an executive brief keeps producing incomplete output. What should the associate do?

Break the work into stages with a defined output and review point for each stage

Submit the same request repeatedly until every step appears

Ask Claude to skip the analysis and guess the likely themes

Remove the source transcripts to reduce the amount of context

Claude's first draft follows the requested structure but misses two mandatory policy points. Which next prompt is most useful?

Identify the two omissions, restate the relevant policy requirements, and ask for a revised draft

Ask Claude whether its first answer was perfect

Start a new chat with the identical prompt and no feedback

Change the tone from formal to friendly without mentioning the omissions

An associate needs help brainstorming possible workshop themes before choosing one. Which instruction best fits that task type?

Generate several distinct themes, state the audience need each serves, and avoid selecting a winner yet

Produce only one final theme and hide all alternatives

Write a legal approval memo for a theme that has not been selected

Return a numeric answer with no explanation

Claude summarizes a newly issued regulation and cites a precise subsection. What should happen before the summary is sent to a compliance team?

Verify the subsection and the summary against the official regulation

Trust the citation because it looks specific

Ask Claude to assign itself a confidence score and use that as proof

Improve the formatting and send it without checking the source

A project update is factually accurate but omits the budget risk and the decision deadline. Which evaluation finding is most precise?

The output has a completeness problem

The output has a spelling problem

The model choice is necessarily wrong

The source documents must be fabricated

Two Claude drafts contain the same facts. One is for engineers and one is for executives. What is the best way to compare them?

Use audience-specific criteria for detail, terminology, decisions, and next actions

Choose the longer draft automatically

Choose the draft with more adjectives

Use identical criteria that ignore the intended reader

Claude produces a table of vendor claims gathered from several sources. Which validation plan is strongest?

Trace each material claim to its source, check dates, and flag conflicting evidence

Check only whether the table looks consistent

Accept all claims that appear in more than one sentence

Remove source links so the table is easier to read

A team needs thousands of short, routine reply drafts, and speed and cost matter more than deep reasoning. What selection principle best fits?

Use a faster, lower-cost model that meets the tested quality threshold

Always use the most capable and expensive model

Choose a model only from its name

Ignore quality checks because the task is repetitive

A recurring research project needs stable instructions and a maintained set of reference documents. Which product feature is the best starting point?

A Claude Project configured with instructions and knowledge sources

A fresh unrelated chat for every task with no shared context

An image artifact with no source material

A public prompt that includes confidential documents

← All practice tests

PRACTICE ENGINE · CLAUDE CERTIFIED ASSOCIATE

Claude Certified Associate Practice Test.

28 free practice questions with answers and explanations.

No signup required. Choose a topic and review each answer.

Start practicing →

About these practice questions

Written by Every Exam Prep Editorial TeamSource and review policy

Verified against the official content outline September 9, 2026

These are original study questions written from published exam objectives—not recalled, copied, or confidential live-exam items. Always confirm current coverage with the official sources linked on this page.

Exam format and study resources

The Claude Certified Associate is administered by Anthropic, with 60 scored questions, a 2 hours time limit and a 720/1000 result.

This free Claude Certified Associate practice test has 28 original questions written to Anthropic's official content outline, last checked against it on September 9, 2026. Every question shows a worked explanation, and nothing here requires a signup.

As of 2026, the Claude Certified Associate exam fee is $99.

About this free Claude Certified Associate practice set

This independent set contains 28 original questions, with four questions for each of the seven domains in the Claude Certified Associate – Foundations blueprint. It is designed for practice and does not reproduce live certification items.

The official CCAO-F exam has 60 multiple-choice and multiple-response items, a 120-minute limit, and a scaled passing score of 720 on a 100–1,000 scale. The largest blueprint area is Output Evaluation and Validation at 21%, followed by Workflow Integration and Solution Design at 16% and Governance, Risk, and Responsible Use at 15%.

Use the topic filter to practice one domain, then switch to a mixed set. For each missed question, identify the decision you overlooked: task framing, validation, product or model selection, workflow design, configuration, governance, or troubleshooting. Check Anthropic's current certification page before registering because eligibility and program details can change.

Practice topic

Difficulty

QUESTION 1 / 28Prompting and Task ExecutionEasy0/0

A marketing associate asks Claude to draft a launch email but receives a generic response. Which revision is the best first step?

0/0session

0 answers saved

Browse all questions & answers

1. A marketing associate asks Claude to draft a launch email but receives a generic response. Which revision is the best first step?
- A. Add the audience, goal, product facts, tone, constraints, and requested output format
- B. Repeat the same one-sentence request in capital letters
- C. Ask for ten times more words without adding context
- D. Remove all constraints so Claude can decide the campaign strategy
Show answer & explanation
Answer: A
A structured prompt gives the model the information needed to produce a usable draft. More volume or emphasis does not repair missing context and success criteria.
2. A request to analyze 80 interview transcripts, find themes, compare roles, and draft an executive brief keeps producing incomplete output. What should the associate do?
- A. Break the work into stages with a defined output and review point for each stage
- B. Submit the same request repeatedly until every step appears
- C. Ask Claude to skip the analysis and guess the likely themes
- D. Remove the source transcripts to reduce the amount of context
Show answer & explanation
Answer: A
Task decomposition turns a complex request into inspectable stages. It also creates opportunities to validate the analysis before the final brief is drafted.
3. Claude's first draft follows the requested structure but misses two mandatory policy points. Which next prompt is most useful?
- A. Identify the two omissions, restate the relevant policy requirements, and ask for a revised draft
- B. Ask Claude whether its first answer was perfect
- C. Start a new chat with the identical prompt and no feedback
- D. Change the tone from formal to friendly without mentioning the omissions
Show answer & explanation
Answer: A
Diagnostic feedback should name the observed gap and the requirement that the revision must satisfy. A style change does not address missing substance.
4. An associate needs help brainstorming possible workshop themes before choosing one. Which instruction best fits that task type?
- A. Generate several distinct themes, state the audience need each serves, and avoid selecting a winner yet
- B. Produce only one final theme and hide all alternatives
- C. Write a legal approval memo for a theme that has not been selected
- D. Return a numeric answer with no explanation
Show answer & explanation
Answer: A
A brainstorming prompt should invite diverse options and clarify the dimensions that make the options useful. Prematurely forcing a single answer narrows exploration.
5. Claude summarizes a newly issued regulation and cites a precise subsection. What should happen before the summary is sent to a compliance team?
- A. Trust the citation because it looks specific
- B. Improve the formatting and send it without checking the source
- C. Ask Claude to assign itself a confidence score and use that as proof
- D. Verify the subsection and the summary against the official regulation
Show answer & explanation
Answer: D
Specific-looking citations can still be wrong. Validation against the authoritative source addresses accuracy; confidence language and formatting do not.
6. A project update is factually accurate but omits the budget risk and the decision deadline. Which evaluation finding is most precise?
- A. The output has a completeness problem
- B. The output has a spelling problem
- C. The model choice is necessarily wrong
- D. The source documents must be fabricated
Show answer & explanation
Answer: A
Accuracy alone is not enough. An output can contain no false statements and still fail because it omits information required for the audience's decision.
7. Two Claude drafts contain the same facts. One is for engineers and one is for executives. What is the best way to compare them?
- A. Use identical criteria that ignore the intended reader
- B. Choose the longer draft automatically
- C. Use audience-specific criteria for detail, terminology, decisions, and next actions
- D. Choose the draft with more adjectives
Show answer & explanation
Answer: C
Evaluation should reflect the intended audience and purpose. The same information may need different detail, vocabulary, and action framing for different readers.
8. Claude produces a table of vendor claims gathered from several sources. Which validation plan is strongest?
- A. Trace each material claim to its source, check dates, and flag conflicting evidence
- B. Check only whether the table looks consistent
- C. Accept all claims that appear in more than one sentence
- D. Remove source links so the table is easier to read
Show answer & explanation
Answer: A
A useful validation plan tests provenance, freshness, and disagreement. Visual consistency cannot establish whether the claims are supported.
9. A team needs thousands of short, routine reply drafts, and speed and cost matter more than deep reasoning. What selection principle best fits?
- A. Use a faster, lower-cost model that meets the tested quality threshold
- B. Always use the most capable and expensive model
- C. Choose a model only from its name
- D. Ignore quality checks because the task is repetitive
Show answer & explanation
Answer: A
Model selection should match the task's quality, speed, and cost requirements. The team should still evaluate a sample before scaling the workflow.
10. A recurring research project needs stable instructions and a maintained set of reference documents. Which product feature is the best starting point?
- A. A Claude Project configured with instructions and knowledge sources
- B. A fresh unrelated chat for every task with no shared context
- C. An image artifact with no source material
- D. A public prompt that includes confidential documents
Show answer & explanation
Answer: A
A Project is suited to recurring work that benefits from shared instructions and maintained knowledge. The configuration still needs review as requirements and sources change.
11. A long conversation now contains outdated assumptions that keep influencing new answers. Which action is most appropriate?
- A. Summarize the valid context and restart with a clean, explicit brief
- B. Continue indefinitely and hope the old context is ignored
- C. Add more unrelated conversation history
- D. Remove the current task requirements
Show answer & explanation
Answer: A
Restarting with a verified summary can reduce interference from stale context while preserving the information the next step actually needs.
12. An analyst must choose between a quick chat response and a structured artifact for a stakeholder deliverable. What should drive the choice?
- A. How the audience will review, reuse, and revise the output
- B. Which option has the most colorful interface
- C. A rule that every task must use the same format
- D. Whether the first draft can avoid all human review
Show answer & explanation
Answer: A
The output format should fit the workflow and audience. Reusability, review, and revision needs are more relevant than visual preference alone.
13. A team wants to add Claude to a customer-escalation process. What should it define before automating any step?
- A. Inputs, decision points, risks, owners, handoffs, and success measures
- B. Only the desired final wording
- C. A requirement that every step be delegated to AI
- D. A launch date before reviewing the current process
Show answer & explanation
Answer: A
Workflow design begins with the actual process and its controls. Mapping decisions and handoffs reveals where AI can assist and where people must retain responsibility.
14. Claude drafts meeting summaries well, but action owners are sometimes assigned incorrectly. What is the best workflow change?
- A. Ask Claude to promise that assignments are accurate
- B. Require a human participant to confirm owners and deadlines before distribution
- C. Send every summary automatically because most are correct
- D. Delete owners and deadlines from all future summaries
Show answer & explanation
Answer: B
The workflow should place verification at the point where an error would create downstream harm. Human confirmation is a focused control rather than abandoning the useful drafting step.
15. A stakeholder asks what value Claude adds to a research workflow. Which explanation is strongest?
- A. It can accelerate defined synthesis tasks, while source validation and final decisions remain assigned to named reviewers
- B. It guarantees that every research claim is correct
- C. It removes the need to define requirements
- D. It should replace every specialist involved in the workflow
Show answer & explanation
Answer: A
A credible value statement names the supported task, expected benefit, and operational limits. Absolute guarantees misrepresent the technology and the workflow.
16. A pilot reduces drafting time but increases correction time. What should the team do next?
- A. Measure total cycle time and error types, then redesign the weak step before scaling
- B. Scale immediately because drafting alone became faster
- C. Ignore correction work because it happens later
- D. Stop measuring once the pilot is launched
Show answer & explanation
Answer: A
Workflow value must be measured end to end. A local speed gain is not an improvement if it shifts more work or risk to a later stage.
17. A Project keeps producing answers based on an old employee handbook. What is the best first action?
- A. Replace or remove the outdated source and verify which current document should govern
- B. Add a prompt asking Claude to pretend the handbook is current
- C. Keep both versions without identifying which one controls
- D. Change the output tone while leaving the sources untouched
Show answer & explanation
Answer: A
Configuration quality depends on maintained knowledge. Prompt changes cannot reliably correct a stale source that remains in the working context.
18. Which Project instruction is most useful for a team that prepares weekly operational reports?
- A. Never ask for clarification under any condition
- B. Invent estimates whenever a metric is unavailable
- C. Use the approved metric definitions, flag missing data, separate facts from recommendations, and follow the report template
- D. Make every report impressive
Show answer & explanation
Answer: C
Effective system-level instructions define durable process rules, source expectations, uncertainty handling, and output structure.
19. A connector exposes several folders with similar file names and different access levels. What should the owner do before relying on it?
- A. Confirm permissions, identify authoritative locations, and test retrieval with representative tasks
- B. Assume every connected file is current and approved
- C. Broaden access to all folders for convenience
- D. Hide the connector from reviewers
Show answer & explanation
Answer: A
Knowledge management includes access control, authority, and retrieval quality. Connection alone does not establish that the right files will be used safely.
20. A useful Claude workflow conflicts with the organization's AI policy. What should the associate do?
- A. Pause the conflicting use and seek an approved path or escalation
- B. Proceed quietly because the output is helpful
- C. Ask Claude to decide whether the policy applies
- D. Move the work to a personal account
Show answer & explanation
Answer: A
Responsible use requires following organizational controls and escalating uncertainty. Moving work outside approved systems increases rather than resolves risk.
21. When should a team review its Claude Project configuration?
- A. Never, because Project instructions are permanent
- B. On a defined schedule and whenever policies, sources, users, or workflow goals change
- C. Only when the interface design changes
- D. Only after a serious error reaches a customer
Show answer & explanation
Answer: B
Configurations require maintenance. Scheduled and event-driven reviews help keep instructions, permissions, and knowledge aligned with current work.
22. A spreadsheet contains customer names and account numbers, and policy restricts sharing regulated personal data. What is the best response?
- A. Remove or anonymize the identifiers in line with policy before using the data
- B. Upload the file unchanged because the analysis is internal
- C. Add a prompt telling Claude not to notice the identifiers
- D. Rename the file so the sensitive fields are less obvious
Show answer & explanation
Answer: A
The control must address the sensitive data itself and follow organizational policy. A prompt or file-name change does not remove exposure.
23. A recruiter wants Claude to rank candidates using historical hiring decisions that may contain bias. What should happen first?
- A. Assess the data, criteria, legal and policy risks, and require appropriate human oversight
- B. Automate the ranking because historical decisions are objective
- C. Hide the ranking method from candidates and reviewers
- D. Use the model output as the sole hiring decision
Show answer & explanation
Answer: A
A high-impact use case requires scrutiny of data, criteria, fairness, policy, and accountability before deployment.
24. Claude drafts medical information for a public handout. Which workflow is most responsible?
- A. Ask readers to verify every claim themselves after publication
- B. Remove all citations so the handout appears simpler
- C. Publish immediately if the writing sounds confident
- D. Use approved sources and require qualified human review before publication
Show answer & explanation
Answer: D
High-stakes information needs authoritative sources and qualified review. Fluent wording is not evidence of clinical accuracy.
25. A prompt that worked last month now produces inconsistent reports. What is the best first troubleshooting step?
- A. Compare inputs, instructions, sources, expected format, and recent workflow changes to isolate the variable
- B. Rewrite the entire process without examining the failure
- C. Add several unrelated examples at once
- D. Assume the model is the only possible cause
Show answer & explanation
Answer: A
Diagnosis begins by making the failure observable and narrowing likely causes. Changing many variables at once makes it harder to learn what fixed the problem.
26. Claude follows the requested tone but repeatedly omits one section of a report. Which change is most targeted?
- A. Remove the report template
- B. Replace all reference documents
- C. Add an explicit section checklist and require a completeness check before finalizing
- D. Increase creativity without changing the structure
Show answer & explanation
Answer: C
The observed failure is structural completeness, so the next test should make the required sections explicit and verifiable.
27. A workflow's output is accurate, but reviewers spend time reformatting it for the same template every week. What is the best optimization?
- A. Specify the reusable template and validate that the generated structure matches it
- B. Reduce fact-checking to save time
- C. Ask for longer prose
- D. Add more reviewers without changing the output
Show answer & explanation
Answer: A
Optimization should remove repeated low-value work while preserving quality controls. A stable output schema addresses the measured bottleneck.
28. After a successful prompt revision, what should the team do to make the improvement persist?
- A. Update the shared instructions or template, document the reason, and test it on representative cases
- B. Keep the fix only in one person's chat history
- C. Delete the prior evaluation criteria
- D. Assume the fix works for every future task without testing
Show answer & explanation
Answer: A
A durable improvement belongs in the maintained workflow and should be checked across representative inputs before broad use.

Explore more Technology & IT Certifications

Family overview →

More in this category

2026 statistics

Key facts: Claude Certified Associate exam

MCQ questions

720/1000

To pass

Time limit

$99

Exam fee

Get a free Claude Certified Associate study plan

A week-by-week plan plus new practice questions, straight to your inbox.

Official sources

Primary documents used to verify the exam details shown on this page.

Claude Certified Associate – Foundations Exam Guide, version 1.0Anthropiceverpath-course-content.s3-accelerate.amazonaws.comeffective July 1, 2026
Claude Certification FAQAnthropicanthropic-partners.skilljar.com
Claude Certified Associate – Foundations certification pageAnthropicanthropic-partners.skilljar.com

Last verified against the official exam content outline: September 9, 2026

Frequently asked questions

How many questions are on the Claude Certified Associate – Foundations exam?

The CCAO-F exam has 60 items. The item formats are multiple choice and multiple response, and each multiple-response item states how many responses to select.

How long is the Claude Certified Associate exam?

The official time limit is 120 minutes.

What score do I need to pass the Claude Certified Associate exam?

The passing score is 720 on a scaled range from 100 to 1,000. The result report includes pass or fail status, a scaled score, and percent-correct information by domain.

Is this the official Anthropic practice exam?

No. These are independent, original practice questions. Anthropic says the practice exam from its previous platform was retired; the official exam guide includes sample questions that demonstrate item format and style.

Who can register for Claude certification?

Anthropic currently limits certification access to people at Claude Partner Network organizations and requires a recognized partner-company email address. Check the current official eligibility rules before planning an exam date.