You have just been sent a link for the Criteria Cognitive Aptitude Test (CCAT) for a tech role, and the first thing you want is a number: what score clears the bar?
Here is the honest answer, and it is the one most prep pages avoid. Almost no employer publishes its CCAT cutoff. Criteria Corp sells the CCAT to employers and lets each one set its own score profile per role, so there is no public master list of required scores. Any page that hands you a confident table saying Google needs a 35 and Amazon needs a 30 is inventing it. What Criteria does publish is the scale you are being measured against, and that turns out to be far more useful than a made-up cutoff.
This guide gives you the published numbers, the small set of tech cutoffs that are actually known, an honest label on everything that is candidate rumour, and a prep plan that works no matter which company sent you the link. Everything downstream depends on four facts Criteria publishes about its own test.
- The CCAT is 50 questions in 15 minutes. That works out to 18 seconds per question.
- The average test taker gets 24 questions right. Not 40, not 35. Twenty-four.
- Fewer than 1% of candidates answer all 50 questions. Running out of time is the normal outcome, not a personal failure.
- The published standardisation sample of 985 working adults produced a mean of 24.2, a median of 24, and a standard deviation of 8.58.
Two more format facts matter for strategy. There is no calculator, and there is no penalty for a wrong answer. Each question has five answer options, so a blind guess is worth 0.20 of a point on average, while a blank is worth exactly zero. Leaving anything unanswered is the only guaranteed way to give points away.
Now put a target next to that. Because Criteria publishes both the mean and the standard deviation, you can convert a raw score into an approximate percentile yourself, treating the distribution as roughly normal:
| Raw score | Approximate percentile |
|---|---|
| 24 | 50th, the average |
| 28 | 67th |
| 30 | 75th |
| 32 | 82nd |
| 35 | 90th |
| 38 | 95th |
| 40 | 97th |
| 42 | 98th |
Read that table twice, because it reframes the whole exercise. A 32 is not a mediocre result, it is roughly the 82nd percentile. A 40 is genuinely rare. And any article telling you a role “needs 45 to 50” is describing a score almost nobody produces, given that under one percent of candidates even reach question 50.
The tech cutoffs that are actually known
Crossover
This is the one set of real numbers in this guide, and it is first-hand: I worked at Crossover. The internal bar is 35 for most entry-level roles, and 40 and above for the high-paying roles at $100K and up.
Crossover does not publish those figures anywhere public. They sit inside the hiring portal rather than on a marketing page, which is why you will not find a URL to cite for them, and why so many prep sites quote a different number every time. Stating it plainly is more useful than dressing it up as a “study.”
Vista Equity Partners
Vista is the most frequently reported CCAT user in tech, and Vista’s use of cognitive aptitude testing across its portfolio companies has been widely written about. What is not available is a published, role-by-role cutoff.
Candidates commonly report targets in the mid to high thirties for technical roles. Treat that for what it is: an aggregate of individual candidate reports, not a Vista-published requirement. If you are interviewing there, aiming at 35 or better puts you around the 90th percentile, which is a defensible plan regardless of where the real line sits.
Everyone else
For any other employer, the correct answer is that the cutoff is not published. Not “roughly 32.” Not published at all. Employers are under no obligation to disclose which assessment they use or where they set the line, and most of them do not.
The fastest way to find out is still the least glamorous one: ask your recruiter. Most will tell you whether there is a threshold, and some will tell you the number.
What about FAANG?
Meta, Apple, Amazon, Netflix and Google are not known to use the CCAT for engineering hiring. They run their own coding challenges, system design interviews and behavioural loops. If you land on a page listing Google or Amazon inside a CCAT score table, treat that as a strong signal that the rest of the page is invented too.
The CCAT shows up far more often at private equity backed software companies, enterprise and B2B SaaS employers, remote-first companies hiring at volume, and staffing or consulting firms that need to screen a large applicant pool quickly and cheaply.
How to set your own target without a published cutoff
Since almost nothing is published, work backwards from the percentile table instead of hunting for a company-specific number that does not exist.
A general tech role at an unknown employer. Target 30 to 32. That is the 75th to 82nd percentile, comfortably clear of the pool, and it is a realistic jump from an average starting point.
A Crossover entry-level role. Target 35, because that is the actual bar.
A Crossover role paying $100K or more. Target 40 or above, same reason.
A highly competitive technical role where nothing is published. Target 35 and up. At roughly the 90th percentile you are ahead of nine candidates out of ten, which is about as much insurance as a single test score can buy.
Notice what is missing from that list: separate numbers for data analyst versus data scientist versus engineering manager. Nobody publishes role-level CCAT cutoffs at that granularity, and inventing them would only be a better-looking version of the same guess. The bands above are honest, and they are enough to prepare against.
Which areas carry the most weight for a tech candidate
Criteria groups CCAT questions into three areas: verbal reasoning, math and logic, and spatial reasoning. Criteria does not publish an exact question count per area, so any claim that “there are exactly 17 math questions” is a third-party estimate rather than a fact from the test publisher.
What you can do is prepare in proportion to how much each area tends to cost people time.
Math and logic
This is where the clock does the most damage, because one stubborn word problem can eat 60 seconds and quietly cost you three easier questions elsewhere. Number sequences are the highest-yield thing to drill, because the underlying rules repeat.
What comes next in this sequence: 2, 6, 18, 54, ?
A) 108 B) 162 C) 156 D) 144 E) 180
The differences are not constant (6 minus 2 is 4, but 18 minus 6 is 12), so it is not an addition rule. Check the ratios instead: 6 over 2 is 3, 18 over 6 is 3, 54 over 18 is 3. Each term is the previous one times three, so the answer is 54 times 3, which is 162, answer B. Ten seconds, no calculator required. Our number sequence pattern guide drills the rest of the rule set.
Verbal reasoning
Engineers routinely write this area off and lose points to it. Analogies go quickly if you name the relationship in words before you look at the options.
COMPILER is to SOURCE CODE as:
A) Translator is to Language
B) Engine is to Fuel
C) Printer is to Paper
D) Keyboard is to Input
E) Monitor is to Display
The relationship is “converts one form into another form,” which lands cleanly on A. Naming the relationship first is the whole trick, and it stops you being pulled toward an option that is merely topically related.
Spatial reasoning
Criteria describes this area as rotating or flipping images in your head, recognising patterns, and identifying the odd one out. The fastest solvers check one attribute at a time, such as rotation, then shading, then element count, rather than trying to take in the entire pattern at once.
Where the honest gaps are
Several things you will see asserted with total confidence elsewhere are simply not published. It is worth knowing which, so you can spot a page that is guessing.
- Criteria does not publish a scratch paper policy for the CCAT. Whether pen and paper are allowed is set by the employer and the proctoring setup, so confirm it before test day rather than assuming either way.
- Criteria does not publish a per-area question count.
- Criteria does not publish employer cutoffs, by role, by industry, or by company size.
- Criteria is refreshingly candid about its own free samples. They “provide shortened, simulated versions” that “do not reflect the exact experience or full difficulty level,” and the practice test “is not scored and no results will be available.”
That last point matters more than it looks. A five-question teaser is useful for format familiarity and useless as a score predictor. If you want a number you can actually act on, you need a full-length run: 50 questions, 15 minutes, real clock, no pausing.
A prep plan that works without knowing the cutoff
Weeks 1 and 2, find the leak. Take one full-length free CCAT practice test under strict timing. Do not stop early and do not pause. Then look at where your misses cluster by area. Almost everyone has one area doing most of the damage, and it is rarely the one they expected.
Weeks 3 and 4, drill the leak and the clock. Spend roughly half your time on your weakest area and the rest on timed mixed sets. This is also where you build the pacing habit of around 18 seconds per question, which our 15-minute time management strategy breaks down minute by minute.
Final week, rehearse the real thing. Full timed runs only, plus a pass through our comprehensive prep guide to tidy up loose ends. Practise the two habits that reliably add points: skip anything that has not resolved in about 15 seconds, and in the final 30 seconds fill every remaining blank with a single letter. With five options and no wrong-answer penalty, that is free expected value.
The three mistakes that cost tech candidates the most
- Overthinking spatial questions. Run the attribute check, take the answer, move on. Deep analysis is what the 15-minute clock is designed to punish.
- Skipping “easy” math to hunt for the hard questions. Every question is worth the same. There are no bonus points for difficulty.
- Chasing perfection. Thirty-five correct answers beats twenty-five perfect ones. The test rewards coverage, and the scoring model has no way to reward elegance.
FAQ
Q: Do all tech companies use the same CCAT score requirement?
A: No, and almost none of them publish one. Cutoffs are set per employer and per role. Crossover’s bar is 35 for most entry-level roles and 40 or above for roles paying $100K and up, which I know from working there. For any other employer, ask your recruiter.
Q: What is a good CCAT score for a tech job?
A: Against Criteria’s published average of 24 correct out of 50, a score of 30 to 32 puts you around the 75th to 82nd percentile and is a solid general target. A 35 puts you near the 90th percentile.
Q: Is it bad that I did not finish all 50 questions?
A: No. Criteria states that fewer than 1% of candidates answer all 50. Not finishing is the expected outcome, and the scoring model already accounts for it.
Q: Can I negotiate a CCAT requirement?
A: Generally not at the screening stage, since that step is often automated. A strong referral or portfolio changes how much weight the rest of your application carries, but it rarely moves the threshold itself.
Q: How important is the CCAT compared to technical interviews?
A: It usually acts as a gate. Clearing it gets you into the technical loop, but it will not compensate for poor technical performance later on.
Q: What is the difference between the CCAT and a coding challenge?
A: The CCAT measures general cognitive ability and how fast you pick up unfamiliar problems. A coding challenge measures the skills you already have. Plenty of employers use both, precisely because they answer different questions.
Start with the free practice test to see where you actually sit against that 24-question average, then point your prep at whichever area is costing you the most. And if your interview loop stacks other assessments on top of the CCAT, PrepClubs is a full paid platform covering 20+ other pre-employment tests, with a pass guarantee.
