Below are four items modelled on the screens above, written for this page. For each one the answer matters less than the thinking, because the thinking is what the next question will test again in another form.
1. A boundary item. A sign-up form accepts ages from 18 to 65, inclusive.
Which set of test values is most efficient? (a) 20, 40, 60 (b) 17, 18, 65, 66 (c) 0, 50, 100 (d) 18, 65
Answer: (b). Faults cluster at the edges, where a developer typed "greater than" and meant "greater than or equal to." Testing each edge and the value just outside it catches that slip at both ends.
Option (d) checks the edges but never proves the form turns anyone away. Options (a) and (c) spend their tests in the safe middle.
2. A partition item. A shipping calculator charges one rate for parcels up to 2 kg, a second rate above 2 kg and up to 10 kg, and refuses anything heavier. What is the smallest number of test cases that covers every class of input once?
Answer: four. One weight in each valid band (say 1 kg and 5 kg), one over the limit (12 kg), and one invalid entry such as zero or a negative number.
People who answer three forget the invalid class below the range. That forgotten class is often where the bug hides, since nobody on the team thought to type minus one.
3. A priority item. Three bugs are open the day before release. The company logo sits two pixels off on the home page.
The forgot-password link sends no email to users with a plus sign in their address. The checkout total is wrong when a discount code and a gift card are used together. Rank them.
Answer: checkout, then password, then logo. The checkout bug costs money and trust on every affected order. The password bug hits a small group but locks them out completely.
The logo is visible and harms nobody. Items like this also check whether you keep severity, how bad the damage is, apart from priority, how soon it must be fixed. A typo on the payment button can be low severity and high priority at once.
4. A logic item. Every build that passes the smoke test goes to the regression suite. Build 412 did not go to the regression suite.
What must be true? (a) Build 412 failed the smoke test (b) Build 412 either failed the smoke test or was never smoke tested (c) Build 412 has a bug (d) Nothing can be concluded
Answer: (b). The rule only says what happens to builds that pass. A build that skipped regression either failed or was never tested at all.
Option (a) is the tempting trap, and (c) goes further than the facts allow. Good testers make this move all day. They refuse to claim more than the evidence shows.
Practising QA aptitude questions when the vendor offers no practice test
Several vendors say plainly that candidates cannot preview their tests. You can still rehearse the skills. For the reasoning half, general logic and number practice transfers well, and our piece on attention to detail tests covers the spot-the-difference items.
For the knowledge half, open any everyday form, like a hotel booking page, and write the boundary and partition cases for each field. Then write three bug reports for things you find, and time yourself. Twenty minutes of that each evening, over seven days, builds the reflex these screens reward.
Why a percentile of 70 beats a raw score of 30 out of 50
Most vendors hand back a percentile, not a percentage correct. Score at the 70th percentile and you beat 70 out of each hundred people you were compared with, usually other applicants for similar roles.
Raw scores on reasoning screens tend to look low, because the tests are built so that few people finish in time. On the CCAT, answering all 50 questions is rare, so a raw score in the 20s or 30s can still rank well.
What counts as a strong QA aptitude score
No vendor publishes a pass mark. Every employer chooses where to draw its line, often per role, and seldom shares it. A rough rule many candidates hear is that the 70th percentile keeps you in most processes, but a crowded graduate scheme may set the bar higher and a hard-to-fill role lower.
So a good score on a QA aptitude test means clearing that employer's bar, and you usually learn where the line sits by getting a call back. If one section came in low, read it narrowly. A weak number score means you were slower with numbers, under that clock, on that day.
It is no sign you would make a poor tester, and it shows you where to put your practice before the next application. Our guide to logical reasoning and aptitude test scores goes further on reading a result.