Claude Certified Associate – Foundations Study Guide

This guide is for anyone preparing for Claude Certified Associate – Foundations, Anthropic's certification for professionals who use Claude as a productivity tool and build Claude Projects in their day-to-day roles — operations, marketing, project management, education, communications and general knowledge work — rather than for developers building on the API. It is organised around the official Exam Guide (version 1.0, effective July 2026): what the exam covers, how much each part is worth, and how to spend your preparation time accordingly.

One thing to settle first. Certification is currently available only to organizations in the Claude Partner Network. If you are studying ahead of your organization joining, the material below still maps directly to the skills the exam tests — but you will not be able to register until it does.

You will face 60 items in 120 minutes — two minutes an item on average. They are multiple choice and multiple response, and every multiple-response item tells you how many answers to choose.

The exam is delivered by Pearson VUE, online or at a test center, and it is scored against a fixed standard: a scaled score of 720 on a 100–1,000 scale passes, regardless of how anyone else does. There are no prerequisites, and no coding or API experience is expected.

Seven domains, ordered here by weight so the largest comes first. Each description summarises the tasks the Exam Guide lists for that domain — exam items are written against those tasks.

Output Evaluation and Validation (21%)

This domain covers judging whether an output is accurate and complete; spotting hallucinations, inconsistencies and bias; applying fact-checking; deciding when a human has to review it; editing and comparing outputs for their audience; and choosing the right output format — an artifact, inline text or structured data.

Workflow Integration and Solution Design (16%)

This domain covers using Claude to analyse requirements and use cases, to research and plan, to support designing and iterating a solution, to fit into existing workflows or redesign them, and to explain what Claude can and cannot do to stakeholders.

Governance, Risk, and Responsible Use (15%)

This domain covers telling appropriate uses from inappropriate ones, handling data sensitivity, regulation and privacy, following your organisation's AI policy, and understanding the ethical side of using AI at work.

Prompting and Task Execution (14%)

This domain covers writing effective prompts for business and technical work, decomposing a complex request, iterating a prompt to improve the output, and adapting the approach to analysis, research, drafting or brainstorming.

Product and Model Selection (12%)

This domain covers choosing between Projects, research mode, chat and artifacts; telling Haiku, Sonnet and Opus apart and matching the model to cost, speed and quality; and managing context limits — when to restart, summarise or persist.

Configuration and Knowledge Management (12%)

This domain covers configuring a Claude Project with instructions and knowledge sources, managing uploaded knowledge and connectors such as Google Drive and Gmail, writing system-level instructions, and keeping all of it current.

Troubleshooting and Optimization (10%)

This domain covers diagnosing an underperforming prompt or a poor output, adjusting based on feedback and results, and making a workflow more efficient.

Spend your time in proportion to the weights, and start with the domain that is worth the most.

1. Output Evaluation and Validation (21%). Take outputs you have already produced and audit them: verify every figure and citation against its source, and note where you would have trusted something you should not have. The skill tested is judgement about when an answer is safe to use.
2. Workflow Integration and Solution Design (16%). Pick a real process in your own work and redesign it around Claude, then explain the change — and its limits — to someone who would have to approve it.
3. Governance, Risk, and Responsible Use (15%). Read your organization's AI policy, then work through cases where the right answer is to anonymise data, escalate, or not use Claude at all.
4. Prompting and Task Execution (14%). Practise decomposing a large request and iterating a prompt, and notice how the approach should differ for analysis, research, drafting and brainstorming.
5. Product and Model Selection (12%) and Configuration and Knowledge Management (12%). Build a Project with instructions and knowledge sources, then decide which model fits which task on cost, speed and quality.
6. Troubleshooting and Optimization (10%). Take a prompt that is performing badly and fix it deliberately, one change at a time.

• Read the number before the options. Multiple-response items tell you how many answers to select. Choosing two when the item asks for three is simply wrong.
• Do not chase domain scores. Your report shows a percentage per domain, but only the total scaled score decides pass or fail. A weak domain is survivable if the heavy ones are strong.
• Expect judgement, not trivia. The largest domain is about deciding whether an output can be trusted. When two options both produce an answer, the better one is usually the one that checks it.
• Treat the retake rules as a reason to wait until you are ready. The fee applies to every attempt and the wait grows from 14 to 30 to 90 days, so a rushed first sitting is expensive.
• Bring the right ID. A valid, unexpired, government-issued photo ID whose name matches your registration exactly.

The single most important resource is the official Exam Guide itself: it lists every task the exam is written against and includes sample questions that show the item style. The separate practice exam that existed on the previous platform was retired when delivery moved to Pearson, so the guide's samples are now the only official examples.

Beyond that, the guide's own preparation list points to Anthropic's documentation and help articles for Projects, Artifacts, Memory, Skills and Code Execution, and to hands-on work: configure a Project with instructions and knowledge sources, then evaluate what it produces for accuracy and bias. It also says plainly that there is no single required course, and that no resource guarantees a pass.

For volume practice, our free Claude Certified Associate practice test has 150 questions spread across all seven domains by their official weights.