The facts
| Item | Detail |
|---|---|
| Questions | 65 in total: 50 scored + 15 unscored trial items, mixed in and not identified |
| Question types | Multiple choice (1 correct of 4) and multiple response (2 or more correct of 5 or more) |
| Score scale | 100–1,000 scaled |
| Pass mark | 720 |
| Scoring model | Compensatory: you need 720 overall, not a pass in each domain |
| Wrong answers | No penalty; unanswered questions are marked wrong |
| Result | Pass or fail with the scaled score, in your AWS Certification account within 5 business days |
Why scaled, not raw?
AWS runs many versions (forms) of the exam at the same time, and some forms are slightly harder than others. AWS uses statistical equating to convert your raw result on the 50 scored items to the 100–1,000 scale. That way a 720 means the same level of ability whichever form you took. In practice:
- The number of correct answers needed to pass varies slightly by form. There is no fixed “35 of 50”.
- AWS does not publish the raw-to-scaled conversion. Anyone quoting an exact cut score is guessing.
- A harder form needs fewer correct answers for the same scaled score, so a hard exam is not a reason to panic.
The unscored 15
The 15 unscored questions are new items that AWS is trying out for future forms. They look exactly like scored items, so you cannot find them and should not try. Two practical consequences:
- Give every question full effort. You cannot tell which ones count.
- If a question seems strangely obscure or badly worded, it may be a trial item. Make your best choice, flag it if you want, and do not let it knock your confidence.
How multiple response items are marked
Multiple response items tell you how many options to choose (“Choose two.” or “Choose three.”). AWS does not publish partial-credit rules for this question type, so treat every multiple response item as all-or-nothing: it counts only if you select every correct option and nothing else. Courselo mocks score them this way. That is why one weak pick on a Choose two item costs as much as a completely wrong answer.
Multiple response items are a minority of the questions. They take longer to answer but carry the same weight as any other item, so read them carefully but spend no more than 3 minutes on each.
Compensatory scoring and domain weights
Scored content is weighted by domain. Out of 50 scored questions, that means roughly:
| Domain | Weight | About this many scored questions |
|---|---|---|
| D1 Design Secure Architectures | 30% | 15 |
| D2 Design Resilient Architectures | 26% | 13 |
| D3 Design High-Performing Architectures | 24% | 12 |
| D4 Design Cost-Optimized Architectures | 20% | 10 |
Because scoring is compensatory, a strong Domain 1 can make up for a weak Domain 4. But weakness costs more marks in a heavy domain. Dropping from 90% to 60% in D1 costs about 4–5 scored questions. The same drop in D4 costs about 3.
Reading your official score report
The report shows:
- Pass or fail and your scaled score.
- For each domain, whether you meet competencies or need improvement. There is no per-domain number.
A report with one domain marked needs improvement and an overall score of 780 tells you exactly where to work before a professional-level exam. If you failed, the domains marked needs improvement become your first study block for the retake. The retake wait is 14 days, with no limit on attempts and a full fee each time.
Courselo mock scores on the AWS scale
Every question in a Courselo mock is scored (there are no hidden trial items), and multiple response is all-or-nothing. We convert your percentage correct to an estimated scaled score with a slightly conservative curve:
| Percent correct | Estimated scaled score | Full mock (65 questions) | Diagnostic (24 questions) |
|---|---|---|---|
| 55% | 610 | 36 correct ≈ 613 | 13 correct ≈ 602 |
| 70% | 720 (pass) | 46 correct ≈ 726 | 17 correct ≈ 726 |
| 78% | 780 (safe pass) | 51 correct ≈ 785 | 19 correct ≈ 792 |
| 85% | 850 (top marks) | 56 correct ≈ 862 | 21 correct ≈ 875 |
| 90% | 900 | 59 correct ≈ 908 | 22 correct ≈ 917 |
Between these points the curve is linear. For a full mock, 45 correct (about 714) is a fail and 46 (about 726) is a pass. That one question shows how thin the margin is at 720, and why Courselo asks you to reach 780 before you book.
What the targets mean
| Target | Scaled score | What it means in practice |
|---|---|---|
| Pass | 720 | You pass on a typical form, with no margin for a hard one |
| Safe pass | 780 | Book the exam when both full mocks reach this |
| Top marks | 850 | About 85% correct and no domain below about 75%; you can tell apart the two plausible answers |
| Elite | 900 | You pick the right option even on the hardest multiple response items |
Scoring myths
How to use this
- For your first full mock, aim for 780+. If you score below 720, go back to your weakest domain’s lessons and banks before you take mock 2.
- Count your multiple response misses separately. If they cost you 3 or more points on a mock, work through the multiple response method in the Decoding scenario questions guide.
- Aim for 850+ on mock 2, with no domain below 75%. That profile is the one that usually converts into a top-marks result on the real exam.