The CCAR-P exam format: length, timing and how scoring works

How the exam is structured

The sitting is a single fixed-length form. No adaptive branching, no separate sections to finish in order, and no penalty for moving around. You can flag an item, carry on, and come back to it while the clock still has time on it.

The pacing number is the one worth internalising. Under two minutes an item sounds comfortable until you meet a scenario that spends four paragraphs describing a retrieval pipeline before it asks you anything. Those items exist. They cost three or four minutes each, and they are paid for by the recall items that take twenty seconds. Average two minutes on the easy half and you will not finish.

What a scaled score of 720 actually means

This is the most misread number in certification prep. 720 out of 1000 is not 72%. A scaled score is a raw count run through a conversion that accounts for how difficult the particular form you sat happened to be. Two candidates who both got 44 items right can walk away with different scaled scores if one of them drew a harder set.

So you cannot reverse-engineer "I need 45 correct" and aim at it. What you can do is treat the scale as a way of ranking your own attempts over time. A practice mock scored on the same range tells you whether you are trending toward the cut or away from it, and while you are still studying that is the only question that matters.

How the domains are weighted

Every question belongs to exactly one domain, and the weights decide how many items each domain contributes to a form. This is the most useful table in your preparation, because it converts directly into a revision timetable.

Two things fall out of this. Integration, solution design and evaluation are 52% of the exam between them, so a weakness in any one of the three is close to fatal. And D7 is four items. Four. If you are spending an evening on developer productivity while shaky on retrieval strategy, you are optimising the wrong end of the table. The full blueprint breakdown covers what sits inside each one.

The three question formats

Nearly everything is a scenario followed by one of three shapes, and knowing which one you are looking at changes how you should answer it.

1. 1

   Single best answer

   Four or five options, exactly one credited. The distractors are usually not wrong in the abstract. They are wrong for the constraint stated in the stem, so read the constraint first and the options second.

2. 2

   Multiple response

   Pick two or three from a longer list, and the stem tells you how many. There is no partial credit. A set with two right answers and one wrong scores the same as a blank, which makes over-selection the most expensive habit on the exam.

3. 3

   Matching

   Pair each scenario with the right technique, pattern or control. Also scored all-or-nothing, so one crossed wire sinks the item. Work the pairs you are certain of first and let elimination handle the rest.

What the score report tells you afterwards

You get a scaled score, a pass or fail, and the useful part: a performance breakdown against all 38 objectives instead of just the seven domains. A domain-level report tells you that you were weak on integration, which spans everything from authentication gaps to chunking strategy. An objective-level report tells you it was chunking.

That granularity is why this site tags every question to an objective and drills at that level too. If you failed and are planning a retake, the objective rows on your report are the study plan.

Using a practice mock to predict your real score

A practice score only predicts anything if the practice conditions match. Three rules make the difference between a number that means something and a number that flatters you.

1. 01Sit it timed, in one go. A 63-item form pulled apart across three evenings measures your knowledge, not your exam. Fatigue in the last twenty items is real and measurable.
2. 02No feedback until you submit. The moment you see a green tick mid-form, you have learned the answer to the item you were about to reason about. Guided and quiz modes are for learning; they are not mocks, and this site excludes them from the projection for exactly that reason.
3. 03Repeat with fresh forms. One mock is a data point and a lucky draw flatters. Two or three sittings that all land above the cut is evidence. One that squeaks over it is not.

Aim comfortably above 720, not at it. A pool you have seen before is always easier than a live form you have not. If your mocks land at 730 you are not ready, you are on the line with all the variance still ahead of you.

Common questions about the format