CCA Domain Weighting: How to Split Study Time Across 5 Domains

The CCA-F exam weights its five domains 27% / 18% / 20% / 20% / 15%, so the simplest study plan is to split your hours the same way: in a 40-hour budget that is roughly 9 hours on Domain 1, 6 on Domain 2, 6.5 each on Domains 3 and 4, 4 on Domain 5, plus 8 hours reserved for mock exams and mistake review. Then move hours toward whichever domain your practice scores show is weakest.

Below is that arithmetic for both a 40-hour and an 80-hour budget, why the biggest domain deserves more than its share, and why the smallest one still cannot be skipped. Weights are as of mid-2026; confirm on the official Anthropic / Pearson VUE page before you register.

What are the CCA domain weights?

Domain	Weight	Approx. questions (of 60)
D1 Agentic Architecture & Orchestration	27%	~16
D2 Tool Design & MCP Integration	18%	~11
D3 Claude Code Configuration & Workflows	20%	~12
D4 Prompt Engineering & Structured Output	20%	~12
D5 Context Management & Reliability	15%	~9

The passing score is a scaled 720 on a 100–1000 scale. Anthropic does not publish the raw-to-scaled conversion, so treat the question counts as planning numbers, not a promise. What the table does tell you is that a perfect D1 and a blank D5 is a much better outcome than the reverse, and that four of the five domains are within a few questions of each other.

How do I convert the weights into study hours?

Take your total budget, set aside about 20% for mock exams and reviewing mistakes, and split the rest proportionally. Two worked budgets:

Domain	Weight	40-hour budget (32h study)	80-hour budget (64h study)
D1 Agentic Architecture	27%	9 h	17 h
D2 Tool Design & MCP	18%	6 h	12 h
D3 Claude Code Config	20%	6.5 h	13 h
D4 Prompt Engineering	20%	6.5 h	13 h
D5 Context Management	15%	4 h	9 h
Mock exams and mistake review	—	8 h	16 h
Total		40 h	80 h

Forty hours is a realistic floor for someone who already builds with Claude tools weekly; eighty is closer to what a strong generalist engineer with little agent experience should plan for. Neither number is a promise, and the 30-day study plan shows how the 40-hour version fits into a month.

Inside a domain, spend hours in this order: read the domain’s study articles once, do the domain practice set, re-read only the articles behind your mistakes. Reading everything twice before touching a question is the most common way to burn the budget.

Why does Domain 1 deserve more than 27% of my time?

Three reasons, and only one of them is the weight.

First, D1 is the largest slice: about 16 questions. Second, its core ideas are prerequisites for the others. Sub-agent context isolation (Sub-Agents See Nothing) shows up again in D3’s Explore sub-agent and D5’s delegation pattern. Programmatic vs prompt enforcement (The Model Will Override Your Instructions 4-15% of the Time) is the reasoning behind D4’s schema-enforced output and D2’s error design. Third, D1 questions are long: multi-agent scenarios with several constraints, where the “least bad” answer depends on noticing one word like “irreversible” or “must never”.

If you have only one weekend, spend it on Domain 1 and the mock exam. If you have a month, front-load D1 in week one so its vocabulary is available when you study the rest.

Why shouldn’t I skip Domain 5?

Fifteen percent looks small until you do the maths. Nine questions is more than the margin most candidates pass or fail by, and D5 questions are among the least guessable because they describe failure modes that only appear in long sessions: facts summarized into vagueness (The “$127.50 Refund” Became “Customer Requested a Refund”), the U-shaped attention curve, tool output that fills the context in four calls, and errors reported as “no results”.

D5 is also where the exam hides its reliability questions: escalation triggers, why model confidence is a bad routing signal (Confidence Scores Are Unreliable), and why aggregate accuracy hides failures. These are short articles with memorable numbers; four to nine hours covers them. Skipping them means guessing on questions that are cheap to learn.

When should I move hours away from the weight split?

After your first mock exam, not before. The mock exam samples 60 questions by official domain weight and reports per-domain results, which is the diagnostic the split needs. A practical rule:

• A domain under about 60%: add 25–50% more hours, taken from domains above 85%.
• A domain between 60% and 80%: keep the weight-based hours, but shift them from reading to practice-and-review.
• A domain above 85%: hold it with a short practice set every few days; do not keep reading.

Watch for the D3 trap. Daily Claude Code users often score well on D3 immediately and assume the rest will follow; it usually does not, because D1 and D5 test design decisions you may never have had to make. Move D3 hours to D1 and D5 rather than to more D3.

What is the exam really testing with this weighting?

The blueprint tells you the exam values orchestration and design judgment over syntax. Roughly two-thirds of the questions (D1, D4, D5) are about choosing which mechanism guarantees a behaviour: prompt, hook, schema, tool design or context structure. Plan your hours around learning those trade-offs, and the domain labels take care of themselves.

Next step

Take the free 60-question mock exam first to get per-domain scores, then apply the hour split above with those adjustments. Start reading with Domain 1 and keep Domain 5 on the schedule even if it is last.