CCA Exam Passing Score: What 720 Out of 1000 Really Means

The CCA Foundations (CCA-F) exam passing score is a scaled score of 720 on a scale of 100 to 1000. That is not the same as answering 72% of the questions correctly; it is a position on a scale that Anthropic maps raw results onto so that different exam forms are comparable. Confirm the current passing score on the official Anthropic Academy or Pearson VUE page before you register.

Scaled scoring confuses almost everyone the first time, so this post explains how it works, why it exists, and how to interpret practice results in light of it.

Why is the passing score scaled instead of a percentage?

Certification programs almost never publish a raw percentage cut because they run more than one version of the exam. Two candidates may sit different sets of 60 questions drawn from the same pool, and one set may be marginally harder. If the pass mark were “43 correct out of 60,” the candidate with the harder form would be penalised for luck.

Scaling fixes that. Each form is statistically equated so that a given scaled score represents the same level of competence regardless of which questions you saw. The scale runs from 100 to 1000, and 720 is the point the program has decided represents “minimally competent.” The number of raw correct answers that lands on 720 can vary a little between forms, which is exactly why Anthropic does not publish it.

So how many questions do I need to get right?

Nobody outside the program can give you an exact number, and anyone who does is guessing. What you can say safely:

• 720 is 72% of the top of the scale, or about 69% of the way from the 100 floor to the 1000 ceiling. Either way, it signals “clearly above two-thirds.”
• On a 60-question exam, that suggests you should be aiming to answer around 42 to 45 questions correctly with margin, not scraping the floor.
• Multiple-response questions may be scored all-or-nothing or with partial credit; Anthropic has not published which. Assume all-or-nothing when you practise so you are not surprised.

The practical implication: do not treat 72% as a target. Treat 80% on realistic practice as your readiness threshold, because the real exam will feel harder than any mock and nerves cost a few points.

How does domain weighting affect the score?

The five domains carry different weights, and the question count you see reflects those weights:

Domain	Weight
D1 Agentic Architecture & Orchestration	27%
D2 Tool Design & MCP Integration	18%
D3 Claude Code Configuration & Workflows	20%
D4 Prompt Engineering & Structured Output	20%
D5 Context Management & Reliability	15%

There is one overall pass mark, not five. That means a very strong Domain 1 can carry a shaky Domain 5. It also means the reverse: bombing Domain 1 (about 16 questions on a 60-question form) leaves you very little room elsewhere. Anthropic has not published minimum per-domain scores, so plan on a single hurdle, but read the domain weighting strategy post before deciding where to spend your hours.

How does our 72% mock threshold map to 720?

Our mock exam uses 60 questions sampled by official domain weight, a 90-minute clock, and a 72% pass threshold. The 72% is deliberately chosen to echo the 720 figure, but it is a raw percentage of correct answers, not a scaled score. Roughly:

• Below 60% on the mock: you have gaps in at least two domains. Go back to the study guide before booking.
• 60 to 72%: you understand the material but your judgment is inconsistent under time pressure. Work through your mistakes list and drill weak domains in the practice hub.
• 72 to 80%: borderline. You might pass, but a harder form or a bad day could push you under 720.
• 80% or higher on two mocks in a row: book the exam.

Remember our mock uses 90 minutes rather than the real 120, so it is tighter on pacing than the real thing. If you clear 80% with time to spare here, the extra 30 minutes on exam day is margin.

What is the exam really testing at 720?

The cut score is set to identify people who can make sound design decisions with Claude under realistic constraints, not people who memorised the docs. Most questions describe a scenario and ask which option is the least bad given the stated constraints. A candidate at 720 can reliably reject options that ignore a constraint (cost, latency, isolation, reliability), and usually picks the option that respects all of them.

That framing matters for scoring strategy: eliminating two obviously wrong choices before deciding between the remaining two is often the difference between 68% and 76%.

Next step

Take the free CCA-F mock exam and check your score against the bands above. Then open your per-domain breakdown; if any domain is under 70%, that is your first stop in the study guide, starting with Domain 1 if it is on the list.