Adaptive practice
Practice that works back from your target
Most practice gives everyone the same questions in the same order. PrepScroll starts from the score you need, works out which question types will close the gap fastest, and re-plans after every answer you give.
- Re-planned
- After every answer
- Each session
- Five sources of practice
- Difficulty
- 70–80% chance of success
- Every question
- Says why it was picked
It starts with what you tell us
When you sign up, a minute of questions gives the plan its starting point. Only your test and target are required.
- Your target score, plus per-skill minimums if your visa or university sets them.
- Your test date, which decides how the plan balances building skills against exam practice.
- Any past scores you remember, so the first estimate isn't a guess.
- Up to three question types you find hardest, which get a boost.
And which of these sound like you:
- “I pause a lot or lose my words when I speak”
- “People sometimes don't understand my pronunciation”
- “I make grammar mistakes when I write or speak”
- “I use the same simple words again and again”
- “I misspell words, especially in a hurry”
- “My essays and long answers feel disorganised”
- “I run out of time in reading”
- “I miss names, numbers and details when listening”
- “I freeze when the clock is running”
How your next question is chosen
The same five steps run every time you answer.
- 1
Start from what you tell us. Your target, any per-skill minimums, past scores and the statements you pick give a first estimate of every skill.
- 2
Measure as you go. Every scored answer updates the estimate for its skill, its question type and the sub-skills underneath, such as fluency, pronunciation, grammar and spelling. Recent answers count more.
- 3
Rank question types by payoff. A type rises when it's far below your target, carries a lot of weight in its skill and improves quickly with practice.
- 4
Pick questions at the right level. Within a type, you get questions you have about a 70–80% chance of doing well on.
- 5
Re-plan after each answer. The questions already on screen stay put; everything after them is picked again with the new evidence.
The ranking rule
Every question type gets a payoff score. The highest scores come first in your plan.
Gap to your targetHow much it countsHow fast it improvesMinutes it takes
Then ×1.4 for a skill below your minimum, ×1.25 for a type you said feels hard, and ×1.1 for a skill you chose to focus on. Types you haven't tried get a small bonus so nothing is left unmeasured.
How fast practice pays off
Some question types improve quickly with practice because they reward a learnable technique; long essays move slowly. A typical type is ×1.00.
- PTE Write From Dictation×1.45
- PTE Repeat Sentence×1.35
- PTE Read Aloud×1.25
- PTE Reading & Writing Fill in the Blanks×1.20
- PTE Describe Image×1.20
- PTE Summarize Written Text×1.20
- PTE Highlight Incorrect Words×1.20
- PTE Reading Fill in the Blanks×1.15
- PTE Summarize Spoken Text×1.15
- IELTS Listening Form, Note, Table and Flow-chart Completion×1.15
- PTE Re-tell Lecture×1.10
- IELTS Reading True / False / Not Given×1.10
- CELPIP Reading Part 1: Reading Correspondence×1.10
- CELPIP Speaking Task 3: Describing a Scene×1.10
- CELPIP Speaking Task 1: Giving Advice×1.10
- IELTS Writing Task 1×1.05
- CELPIP Writing Task 1: Writing an Email×1.05
- CELPIP Writing Task 2: Responding to Survey Questions×0.95
- PTE Write Essay×0.90
- IELTS Writing Task 2×0.85
Try it: watch the plan re-rank
An example student aiming for 79 in PTE. Answer the question in front of you, well or badly, and see what comes next.
Your next question
Write From Dictation
Picked for you: Write From Dictation: about 19 below your target of 79
Answer the question and the plan re-ranks, just as it does after every real answer.
Ranked by how fast they close the gap estimate / target
- 1Write From Dictation60 / 79
- 2Describe Image55 / 79
- 3Read Aloud68 / 79
- 4Reading & Writing Fill in the Blanks64 / 79
- 5Write Essay72 / 79
A simplified model: the ranking uses the real rule and question-type weights; the starting estimates are an example, and each answer here moves the estimate by a fixed step. Real plans also mix in review, trainers, stretch and new types.
What a session is made of
Every 20 questions come from five sources. The mix shifts as your test date gets closer.
Your next 20 cards
The mix changes as your test date gets closer. Pick a stage to see it.
- Targeted8 of 20Question types that close the gap to your target fastest.
- Review4 of 20Mistake Vault cards and low scores from two or more days ago, so fixes stick.
- Trainers5 of 20Short drills for the sub-skill holding you back: sounds, grammar, spelling.
- Stretch2 of 20A harder question in a type you're good at, so your ceiling keeps rising.
- Explore1 of 20A type you haven't tried yet, so nothing on test day is new.
The right difficulty
Within a question type, you get the questions you have about a 70–80% chance of getting right, based on your answers and how other students did on each one.
Every card tells you why
A line on each card says why it's in your session, so the plan never feels random.
- Picked for you: Write From Dictation: about 6 below your target of 79
- Picked for you: Writing is under your 7 minimum, and Writing Task 2 counts for a lot of it
- Picked for you: Back from your Mistake Vault
Trainers built from your own answers
Some of your practice is made just for you, from the mistakes in your own answers. It stays on your device.
Your sound practice
Lines holding the words you paused on or blurred in your recent speaking answers, to shadow until they're clear.
Your spelling sprint
Every word you misspelled in a dictation answer, mixed with the right spellings, against the clock.
Fix your own mistakes
Grammar slips from your latest written answer, turned into a spot-the-error quiz.
Grammar check
Two grammar areas a day, rotating, so the rules you need most keep coming round.
Common questions
What does adaptive practice mean in PrepScroll?
Your practice isn't a fixed list. Each time you answer, PrepScroll updates its estimate of every skill and question type, then re-plans the questions ahead of you so the next ones target the gap between where you are and the score you need.
What do you ask when I sign up?
Your test and target score (with per-skill minimums if your visa or university needs them), your test date, any past scores you remember, a short list of 'sounds like me' statements, and up to three question types you find hardest. Only the test and target are required.
How does it decide which question type comes next?
It ranks every question type by the gap between your estimate and your target, how much that type counts in its skill, and how fast practice tends to raise it, divided by how long one question takes. Skills under your minimum and types you said feel hard get a boost.
Does my test date change the plan?
Yes. With six or more weeks to go, more of each session is short trainers that build the skills underneath. In the last two weeks it's mostly real exam questions, with no new question types.
Does it guarantee a higher score?
No. No practice tool can promise a score. What it does is spend your practice time where an improvement counts most toward your target, and show you why each question was picked.
Is my data shared?
Your plan is built on your device from your own answers. Your profile and study settings sync to your account so the plan follows you between devices. You can turn off sharing answers to question boards in Settings.
One engine, every test
PTE Academic, IELTS and CELPIP all use the same engine, tuned to each test's scoring scale and question weights.
- PTE Academic guideScored from 10 to 90, overall and for each skill. Many universities and visa routes ask for 65 to 79.
- IELTS guideEach part gets a band from 1 to 9 in half bands, and your overall band is their average. Many universities ask for 6.5 or 7.
- CELPIP guideEach skill gets a level from 1 to 12, which lines up with the Canadian Language Benchmarks. Many immigration routes ask for 7 or 9.
Stop guessing what to practice next
Set your target and test date, and every question after that is picked to close the gap. Free to start, nothing to install.