Your recruiter sent a CCAT link for a data science role, and now you want one straight answer: what score do you actually need to hit? The honest version is that no company publishes a fixed number, but data and analytics roles consistently sit near the top of the CCAT scale, and there is a defensible target range you can aim at. This page gives you that range, explains why data roles are held to a higher bar than most, breaks down which question types carry the most weight for you specifically, and shows you how to close the gap before test day.
First, a quick disambiguation so no wrong facts sneak in: this is the Criteria Cognitive Aptitude Test, the workplace hiring test from Criteria Corp, not the Canadian Cognitive Abilities Test used in school admissions for children. Same acronym, completely different test.
Quick takeaways
- Most data science roles target roughly the 75th to 90th percentile, a raw score around 31 to 35 out of 50.
- The CCAT average is 24 correct (50th percentile), so a “fine for everyone” score is usually not enough for a data role.
- No company publishes an official cutoff. Ranges are informed estimates from role difficulty and reported cutoffs, not guarantees.
- Math and logic plus spatial and abstract reasoning tend to carry the most weight for data roles, because they mirror the actual work.
- There is no penalty for wrong answers, so you should never leave a question blank.
- Which points you are losing, and why, matters more than the target number itself.
The short answer: the target range for data scientists
If you want a single number to aim at, aim for the top 15% of test-takers, which is a raw score of about 31 to 35 correct out of 50, roughly the 80th to 90th percentile. Clear the low end of that band and you are competitive at most companies hiring data scientists. Reach the high end and you clear the bar almost everywhere, including firms that set an aggressive cutoff.
Why that range and not lower? Because the CCAT is a general-population test, and the commonly cited average score of 24 (widely reported as the 50th percentile) represents the whole applicant pool across every kind of job. Analytical roles filter above that. A data scientist’s day is pattern recognition, quantitative reasoning, and quickly learning unfamiliar systems, which is close to exactly what the CCAT measures, so employers hiring for those roles tend to weight the score heavily and set the cutoff high.
Be clear-eyed about one thing: no employer publishes “you need a 33.” These ranges come from the general difficulty tier of analytical roles and from cutoffs candidates have reported, not from an official Criteria Corp table for data scientists. Treat 31 to 35 as a well-supported target, not a printed rule.
How data science compares to other roles
Data science sits at the top end of the CCAT expectation curve, alongside consulting and just above general software engineering. Here is how the common target bands line up so you can see where you stand relative to adjacent roles.
| Role | Typical target percentile | Typical raw score band |
|---|---|---|
| Administrative / support | ~30th to 50th | 20 to 24 |
| Sales | ~40th to 60th | 22 to 27 |
| Software engineer (general) | ~65th to 85th | 28 to 33 |
| Data scientist / analytics | ~75th to 90th | 31 to 35 |
| Consulting | ~75th to 90th | 31 to 35 |
| Executive / leadership | ~80th to 92nd | 32 to 36 |
Two things worth noticing. First, data science asks for a couple more correct answers than a general software role, which is a meaningful gap given how tightly scores cluster near the average. Second, the difference between “average” and “data-role target” is only about seven to eleven more correct answers, but on a 50-question, 15-minute test, those points are hard-won. That is the whole challenge in one sentence.
For the tech-role picture in full, see CCAT scores for tech jobs, and for the closely related engineering benchmark, CCAT scores for software engineers.
Which question types matter most for data roles
The CCAT reports a single composite score, so technically every question counts the same toward your raw number. But for a data-role candidate, the strategic reality is that the quantitative and abstract sections are both where you are expected to be strong and where you can gain or bleed the most points. The test breaks down roughly like this:
| Question domain | Approx. share of the 50 | Why it matters for data roles |
|---|---|---|
| Math and logic (word problems, number series) | ~34% | Direct proxy for the quantitative reasoning data work demands |
| Verbal reasoning (analogies, antonyms) | ~34% | Tests how fast you parse and reason with information |
| Spatial and abstract reasoning (patterns, outliers) | ~22% | Pure pattern recognition, the core cognitive skill of analytics |
For a data scientist, math and logic plus spatial and abstract reasoning together are close to two-thirds of the test, and they are the sections that best mirror the actual job. If you are already comfortable there, the CCAT rewards you. If your quantitative speed is rusty, or you have not seen abstract pattern questions since school, that is precisely where a data-role candidate loses the points that separate a 27 from a 33.
One honest caution: do not neglect verbal. It is a third of the test, and a data candidate who assumes they will coast on the math half and ignores verbal often leaves easy points on the table there.
A worked example: pacing to a data-role score
Here is what hitting the target actually looks like against the clock. You have 50 questions and 15 minutes, which is about 18 seconds per question. Less than 1% of people finish all 50, so the goal is never “answer everything.” The goal is “convert as many as possible into correct answers before time runs out.”
Say you are aiming for a raw score of 33 (roughly the 85th percentile). A realistic path: you work steadily and get through 42 questions in the 15 minutes, banking correct answers on the ones you know cold and making educated guesses on the two or three that are eating your clock. Of those 42 attempts, you convert 33 into correct answers. That is your 33. Notice you did not need all 50: you needed accuracy on the ones you touched plus enough pace to reach question 42.
Now the failure mode. Same person panics, tries to rush all 50, and lands 30 correct out of 48 attempts because accuracy collapsed under speed. Raw score 30, about the 78th percentile: still decent, but you just dropped from the strong-data band to the edge of it by rushing. On the CCAT, and especially for a data role where you need the higher end, controlled pace beats frantic speed every time. And because there is no penalty for wrong answers, the one thing you must never do is leave a question blank when time is nearly up: guess on everything unanswered in the final seconds.



