How Pretty Scrum’s Adaptive PSM I Practice Works
Most free PSM I practice tools shuffle the same 80 questions and call it a quiz. Pretty Scrum doesn’t. The next question you see is picked based on what you’ve already gotten wrong, how recently you saw it, and which Scrum Guide topic still has the weakest signal in your history. This page explains exactly how that works so you can decide whether the approach fits how you study.
If you’d rather just try it: start a 5-minute study session. Otherwise, keep reading.
Video: A Smarter Way to Prepare for PSM I
The problem with random question pools
Random shuffles are easy to build, but they teach you to memorize specific questions instead of understanding the underlying Scrum framework. Three concrete failure modes:
- You repeat what you already know. Random selection means a question you got right four times in a row has the same chance of appearing as one you’ve never seen. That’s wasted minutes.
- You memorize answer positions. When “the Product Owner orders the Product Backlog” is always option B, your brain learns the position, not the rule. The real PSM I randomizes options.
- Topical gaps stay invisible. If you’ve never seen a Sprint Retrospective question, a random pool won’t fix that. You walk into the exam blind on a topic you didn’t even know you’d skipped.
Adaptive selection fixes all three.
How the algorithm picks the next question
Every question in the bank is tagged with a Scrum Guide topic (one of the 19 chapters — see the full Scrum Guide chapter index). For every user session, Pretty Scrum tracks four signals:
- Correctness history per topic. If you’ve answered 3 out of 4 Sprint Planning questions correctly, that topic is weighted lower than Sprint Review where you got 1 of 3.
- Recency. A question you saw 30 seconds ago is suppressed; one you saw two days ago is fair game.
- Topic coverage. If you haven’t seen any questions from a given chapter yet, that chapter is prioritized so the bank surfaces every topic at least once before deepening on weaknesses.
- Difficulty balance. Questions are tagged easy / medium / hard. After a wrong answer the next question on that topic drops a notch in difficulty so you build understanding before being tested at the harder level.
The selection function combines those four into a weight per question. Higher weight = higher chance of being picked next. The result: you spend roughly twice as much time on your weak topics as on your strong ones, without anyone (you or me) having to declare which topics those are.
Why this beats “just take more practice exams”
The standard advice for PSM I prep is “take the Scrum Open Assessment until you score 100% three times in a row, then the mlapshin quiz.” That works, but it’s slow because a fixed 80-question quiz mixes 60 questions you already know with 20 you don’t. Adaptive practice strips out most of the 60.
In testing, my own prep took 9 days. The first 3 days were reading the Scrum Guide; the next 4 were adaptive sessions of 10–30 minutes; the last 2 were full 80-question timed mock exams. The adaptive sessions surfaced exactly the topics I was weakest on (Sprint cancellation rules, Definition of Done ownership, the Scrum Master’s authority limits) and let me drill them in 5-minute blocks on my phone. Random quizzes would have shown me a Sprint Planning question I’d already aced four times instead.
I wrote up the full prep timeline in How to pass the PSM I exam.
What the algorithm doesn’t do
A few things Pretty Scrum is intentionally not:
- It’s not predictive. Pretty Scrum doesn’t forecast your real-exam score. It surfaces gaps; closing them is on you. The only honest exam-readiness check is two timed 80-question mock exams above 85% — see the PSM I mock exam.
- It’s not a teacher. Each question has an explanation, but explanations are short by design. The teaching happens when you read the Scrum Guide itself. Pretty Scrum is a feedback loop on top of that reading.
- It doesn’t reproduce official Scrum.org questions. Every question is original, written by me, inspired by the 2020 Scrum Guide. That’s deliberate — reproducing official assessments would be both legally murky and pedagogically harmful (it teaches answer recall, not Scrum).
Privacy and tracking
The algorithm needs your answer history to weight questions. That history is stored:
- Anonymously by default. If you don’t sign in, your progress is kept locally in your browser and pseudonymously in the backend (Supabase, EU region) tied to a random session ID. No name, no email.
- With a LinkedIn account if you opt in. Signing in lets your progress sync across devices. The only data pulled from LinkedIn is name, email, and your LinkedIn ID. None of it is shared.
Full details in the privacy policy.
Roadmap for the algorithm
Two improvements are on the way:
- Spaced repetition for revisited questions. Currently a wrong answer drops the question back into the pool with elevated weight. The next iteration will use SM-2 (the Anki algorithm) so revisits happen at increasing intervals rather than the next session.
- Confidence-weighted scoring. When you tap an answer, you’ll be able to mark it “sure” or “guess.” Guesses you got right will still count as weak signal — you got the answer but not the understanding.
Both are open questions in terms of what helps learners most. If you’ve used Pretty Scrum and have a view, email me at maurice@prettyscrum.org.
Try it
The fastest way to understand the methodology is to feel it. Open a 5-minute study session, answer 10 questions, and notice how the next questions tilt toward the topics you stumbled on.
Author: Maurice Heinze (PSM I, certified by Scrum.org). Last updated: 2026-05-04.