Quick answer: Don't wait until you've finished studying to start answering questions. Two separate things are true. First, attempting questions before you've learned the material improves how well you learn it afterwards — and the research shows this holds even when you get the answers wrong, provided you then see the correct answer. Second, and independently, an early sample of questions tells you where your gaps are, which a syllabus cannot. Neither claim requires you to sit a full-length mock exam cold. A handful of questions is enough to start.
Most people preparing for a certification exam follow the same order: read everything, then test yourself. Testing is treated as the final check — the thing you've earned the right to do.
It's an intuitive order. It's also backwards, and there's a reasonably well-established body of research explaining why.
Why "read first, test later" feels right
Answering questions about material you haven't studied feels pointless. You'll get most of them wrong. Getting things wrong feels like wasted time, and worse, like evidence you started too early.
That instinct is the problem. In the research on this, the wrong answers are part of the mechanism, not a sign of a mistake.
There's also a measurement problem with studying first. Reading a chapter produces a feeling of understanding that's a poor guide to whether you could actually retrieve any of it under exam conditions. Recognition is easy; retrieval is hard. Only one of them is what the exam tests.
What the research actually shows — and what it doesn't
The relevant finding is called the pretesting effect, and it's distinct from the better-known testing effect (which is about reviewing by self-testing after you've studied).
In the study that established it, Richland, Kornell and Kao (2009) had people attempt questions about material they hadn't seen, then study it. The comparison group spent that same time studying. The pretested group remembered more afterwards.
The important detail is how carefully that result was isolated. The researchers deliberately used items participants were almost certain to get wrong — around 5% of guesses were correct, and those were removed from the analysis. So the benefit appeared on exactly the items people had failed to answer. It isn't the case that pretesting helps because you happen to know some of it already.
Later work has added the conditions that matter:
- Feedback matters. An unsuccessful attempt followed by the correct answer beats spending that same time studying. The cited work varies when that feedback arrives rather than removing it, so treat "review the answer" as a condition of the effect, not an optional extra.
- The benefit persists beyond 24 hours, which matters for an exam you're preparing for over weeks.
- It holds across adulthood — but people consistently underestimate it. Learners judge pretesting to be less useful than it measurably is, which is probably why so few do it.
Now the limits, because they matter and are easy to gloss over.
These are laboratory studies. They use word pairs and short passages, over intervals of minutes to days. Nobody has run a controlled study on "sit a full certification mock cold, then study for three months." Anyone who tells you the research proves a specific exam-prep protocol is overstating it, and we're not going to do that.
What the research supports is narrower and still useful: starting retrieval early helps you learn, and failed attempts followed by good explanations are productive rather than wasted. That's enough to change when you start. It isn't enough to prescribe how many questions, or to claim a particular routine is optimal.
The second reason, which needs no research at all
Set the learning effect aside entirely. There's a separate, simpler argument.
A syllabus tells you what's on the exam. It cannot tell you what you don't know. Only your answers can do that.
An exam content outline gives you a list of domains and weights, all of which look equally like things you should study. An early sample of questions turns that flat list into a ranked one — these areas are fine, these are shaky, this one you've never really understood. Everything you do afterwards is better targeted.
This is ordinary diagnostic reasoning, not a research finding, and we're keeping the two arguments separate on purpose. The first is supported by published studies with stated limits. The second is just what happens when you measure something before acting on it.
What a good first session looks like
Short. Reviewed. Not a mock exam.
- Small is fine. You don't need a full-length simulation to learn something useful. A handful of questions across different domains will already show you where the ground is uneven.
- Read every explanation, including the ones you got right. This is the part that carries the effect. A question you answered correctly by elimination is a gap wearing a disguise.
- Don't time yourself yet. Timing tests endurance and pacing. Those matter later. Right now you're gathering information.
- Write down what surprised you. The gap between what you expected to know and what you did is the most useful output of the session.
At PM Pixel this is why we start people with a five-question set rather than a full exam — it's a low-friction way to apply the principle, and it's designed so you see a full explanation after each question. To be clear about what that is and isn't: five is our practical starting dose, not a number validated by the research. The evidence supports starting retrieval early and reviewing explanations. It says nothing about five being an optimal number.
If your first result is bad
It probably will be. That's the expected outcome of testing yourself on material you haven't studied — it's the design, not a verdict.
The useful move is to treat the result as information rather than a judgment. It tells you which domains to open first. It tells you nothing about whether you'll pass, and reading it that way this early is a mistake in both directions: an encouraging early score can produce complacency just as easily as a discouraging one produces panic.
We're deliberately not going to tell you what any particular score means at this stage. That's a genuinely different question, asked much later in preparation, and we've addressed it separately for the PMP and for CPMAI. Early on, the number isn't the point. The pattern is.
FAQ
Isn't it a waste of time to answer questions I can't possibly know?
No — that's the finding that makes pretesting interesting. In the research, the benefit showed up specifically on items people got wrong. The condition is that you then see the correct answer and the reasoning. Attempting without reviewing is much less useful.
Should I take a full-length practice exam before I start studying?
We wouldn't recommend it. The research supports starting retrieval early; it doesn't support any particular format, and a cold full-length simulation is a large time cost for information a much shorter session would give you. Full-length exams are valuable later, when you're building endurance and pacing.
Does this replace studying?
No. Pretesting improves how well subsequent study works. It's a change to the order, not a substitute for the work.
How many questions should I start with?
Enough to cover several domains and few enough that you'll actually review every explanation. We start people with five because it's low friction, not because five is a research-validated number.
What if I get a low score early on?
Expect it. Early results indicate where to study, not whether you're ready. Readiness is a separate question you should ask much later, using different evidence.
Sources
- Richland, Kornell & Kao (2009) — The Pretesting Effect: Do Unsuccessful Retrieval Attempts Enhance Learning? (retrieved 2026-08-17) — the founding study; benefit measured on items not successfully retrieved
- The Pretesting Effect: Exploring the Impact of Feedback and Final Test Timing — Journal of Cognition (2025) (retrieved 2026-08-17) — the role of corrective feedback and timing
- Testing before learning: Exploring the robustness of the pretesting effect — Memory & Cognition (2025) (retrieved 2026-08-17) — replication and boundary conditions
- The pretesting effect is robust throughout adulthood, but metacognitive beliefs about pretesting and challenge differ (2025) (retrieved 2026-08-17) — learners underestimate the benefit
Free, no-catch practice for real exam readiness.
Start with five questions. See what you already know.