Identify the exact test you are facing
Before prep starts, confirm the vendor and test variant. Your invitation email almost always states it. If it does not, ask the recruiter in a two-line email. Treat this as reconnaissance, not admin.
Mixing up a CCAT with a Wonderlic costs you three minutes on a 15-minute test. That is roughly nine unanswered questions. Mixing up an SHL Verify Interactive with an SHL Verify G+ means preparing for an adaptive test instead of a static one. The strategy flips.
Once you have the vendor, pull up the most recent publicly available information on test length, section mix, scoring approach, and retake policy. Spend 30 minutes here. The payoff shows up within your first practice session.
Learn the format before practicing
Format is structure, not content. Question count, time limit, section labels, whether wrong answers are penalized, whether you can flag and return, whether questions get harder as you go. All of that is public.
Compute the per-question time budget out loud. A 50-question, 12-minute Wonderlic gives you 14.4 seconds per question. A 50-question, 15-minute CCAT gives you 18 seconds. A 30-question, 36-minute Watson-Glaser gives you 72 seconds. These numbers dictate your entire approach, and they change how you read a question.
Candidates who practice without this internalized pacing end up training the wrong muscle. They read carefully, they solve every problem, they feel good, and then they freeze when the real clock appears. You want the clock built into your nervous system before you start drilling.
Start untimed, end timed
Your first 20 to 30 practice questions should be untimed. The goal is comprehension. You are mapping what the test actually looks like, learning the vocabulary of the answer choices, and building an internal library of question shapes.
Only after accuracy holds above 75 percent should you layer in time pressure. Rushing into timed mode tanks confidence and teaches your brain that guessing is the default, which is the opposite of what you want early.
Once timed mode kicks in, practice in short sets of 10 to 15 questions rather than full-length mocks. Short sets let you review immediately and build focused reps on specific question families.
Build a mistake journal
A mistake journal is a simple running list. For every wrong answer, write the question type, the pattern you missed, and a one-sentence correction. That is it. No fancy system.
After 100 practice questions, five recurring patterns will account for 60 to 70 percent of your misses. Those five patterns are your prep priority for the next week. Everything else gets background drilling.
The mistake journal is the single highest-leverage habit in cognitive-test prep. It converts raw practice volume into targeted improvement. Candidates who review their journal for 10 minutes before each session gain roughly two to four percentile points per hour of study compared to those who just grind questions.
Use full-length simulations only at the end
Full mocks are a calibration tool. They stress-test your stamina, your pacing, and your ability to recover from a bad stretch. They are terrible learning tools because review happens 30 minutes after you forgot what you were thinking.
One full mock four days out and one full mock 48 hours out is enough. Any more and you are burning energy you need for the real thing.
After each mock, spend 45 minutes reviewing only the questions you got wrong or flagged. Compare your mock pacing to your practice pacing. If you attempted ten fewer questions than usual, diagnose whether it was fatigue, one sticky question, or a shift in difficulty.
Sleep, eat, and move on test morning
Cognitive tests punish fatigue more than any other hiring instrument. Sleep deprivation compresses working memory and slows processing speed. Both feed directly into your score.
Aim for eight hours the night before. Eat a protein-forward breakfast two hours before the test. Hydrate, but not so much that you need a bathroom break mid-test. Avoid excess caffeine. If coffee is normal for you, drink your normal amount. If it is not, do not start today.
Do not study the morning of the test. Five to ten easy questions to warm up is fine. Anything longer burns cognitive fuel you need at minute six of the real thing.
On test day, commit to moving on
The single largest score killer on timed cognitive tests is over-investing in a hard question. If you spend three minutes on a problem you eventually guess, you have lost six easier questions at the end of the section. That is a trade you never want to make.
Build a hard rule: if a question is not yielding after 25 seconds on a short-form test or 60 seconds on a long-form test, flag it, guess, and move on. The guess costs you nothing because almost no cognitive test penalizes wrong answers. The unanswered easy question at the end of the section costs you a guaranteed point.
Practice this rule under timed conditions until flagging feels mechanical rather than emotional. On game day, you will encounter at least two questions designed to trap you. The playbook is simple: do not let them.
PrepClubs attempt data, 727 attempts and 20,554 individual question sittings across CCAT, Wonderlic, PI Cognitive and Cubiks Logiks. Published in full at /research.
The 50-second rule, measured rather than guessed
Every prep site tells you to move on from hard questions. Almost none of them tell you when. We went looking for the number in our own data, because we hold the timestamp of every answer our candidates submit.
Across 20,554 individual question sittings on speeded tests, the median question takes 14 seconds. Three quarters are done inside 24 seconds. Nine in ten are done inside 39 seconds. The top one per cent take 103 seconds, which is nearly two minutes spent on a single item while the clock runs on everything behind it.
That distribution is the whole argument. A question that has already eaten 50 seconds is not a hard question you are about to solve. It is an outlier you have misread, and the evidence says it will keep taking your time.
We then split candidates by how often they let that happen. Those who never spent 50 seconds on anything reached 93.8 per cent of their test. Those where more than one question in twenty ran long reached 65.3 per cent. A third of the paper, unseen, because of a handful of questions.
| Percentile | Seconds on one question | What it means |
|---|---|---|
| Median | 14 | The normal working pace on a speeded test |
| 75th | 24 | Still healthy. You are thinking, not stuck. |
| 90th | 39 | The point where it starts costing a later question |
| 99th | 103 | Almost two minutes on one item. This is the trap. |
PrepClubs attempts on CCAT, Wonderlic, PI Cognitive and Cubiks Logiks. The pacing split is observational: candidates who linger may also be weaker overall, so treat the gap as an upper bound on what pacing alone buys you.
Never leave a blank, and what a blank actually costs
Guess on everything you cannot answer. You have read that advice before. What you have probably not seen is the size of the prize, so here it is from our own attempts.
Seven candidates in ten leave at least one question unanswered on a speeded test. The median attempt leaves eight. At four options a question, blind guessing on eight blanks returns about two marks. Against a mean score of 28.3, those two marks are a seven per cent lift for roughly four seconds of clicking.
Seven per cent sounds small until you remember where cutoffs sit. Most employers screen at a percentile band, and the middle of a score distribution is crowded. Two marks is frequently the difference between the shortlist and the rejection email, and it is the cheapest two marks available anywhere in your preparation.
The habit that captures it is mechanical, not clever. When the invigilator calls one minute, stop solving. Sweep every unanswered question and put something in each box. Do not read them. Do not agonise. The only answer guaranteed to score nothing is the one you never gave.
One exception worth knowing: a handful of instruments do penalise wrong answers, and a few of our own candidates behave as though every test does. Check the instruction screen. If it does not mention a penalty, there is not one.
| Test | Questions left blank | Marks a full sweep would add | Lift on the mean score |
|---|---|---|---|
| Cubiks Logiks | 24.4% | 3.0 | 10% |
| CCAT | 20.5% | 2.0 | 7% |
| Wonderlic | 20.3% | 2.0 | 8% |
| PI Cognitive | 18.8% | 1.9 | 7% |
| IBEW aptitude | 7.9% | 1.1 | 2% |
| Watson-Glaser | 0.4% | 0.1 | 0% |
PrepClubs attempts, 2026. Marks assume blind guessing at the options offered. Watson-Glaser barely appears because its clock is generous enough that almost nobody runs out of time, which is exactly why the guessing advice matters less there.
Putting the playbook together for a full week of prep
Day one is format and baseline. Day two and three are drilling your two weakest question types. Day four is rest or light review plus your mistake journal. Day five is mixed timed sets. Day six is a full mock. Day seven is light review and sleep.
This structure is not magic. It is the minimum viable system that makes sure no critical phase gets skipped. Customize the timings to your life. Keep the structure.
One closing note on expectations. Seven days will not make you a better reasoner, and you should be suspicious of anyone who tells you otherwise. What it reliably does is stop you leaking marks you have already earned, through questions you never reached and boxes you never filled. On our own attempt data that leakage is worth several marks to the average candidate, which is usually the whole gap between a near miss and a pass.