Your notes are the best raw material for a practice quiz you will ever have. They cover exactly what your course covered, at the level of detail your teacher cared about. The problem is that notes are written to be read, and reading them back is the least useful thing you can do with them.
This guide is about the conversion step: taking a page of notes and turning it into questions you can actually be tested on. We will do it by hand first, because knowing what a good conversion looks like is what lets you judge an AI-generated one. Then we will run a real page of notes through our text to quiz generator and show you what came out, including the two questions it got wrong.
Why a quiz beats another read-through
The short version of a large literature: retrieving information strengthens memory more than re-exposing yourself to it. In Roediger and Karpicke's 2006 study (Psychological Science), students who read a passage and then took a recall test remembered more a week later than students who simply read it again, even though the re-readers had scored higher on a test given just five minutes after studying. Dunlosky and colleagues' 2013 review of ten common study techniques rated practice testing as one of only two "high utility" strategies; rereading and highlighting were rated low.
If you want the full background on retrieval, we covered it in our guide to active recall. This article assumes you are convinced and just want to know how to build the quiz.
Step 1: Make your notes quiz-able
Not every note can become a question. Before you write anything, skim the page and sort what is there into three buckets.
Quiz-able facts and relationships. Definitions, causes and effects, comparisons, sequences, named studies and what they found, numbers that matter. "Deeper, semantic processing leads to better recall than shallow processing" is quiz-able. So is "the recency effect disappears after a 30-second distractor task."
Context and signposting. "Unit 3," "see slide 14," "this will be on the exam." Useful to you, useless as questions. Skip it.
Fragments. "Baddeley - 3 parts - see diagram." You understood this when you wrote it. A quiz generator, human or AI, cannot turn it into a fair question because the actual content is not on the page. This is the most common reason AI quizzes from notes come out vague.
Fixing fragments takes five minutes and is worth doing even if you never make a quiz: expand each one into a full sentence that states the relationship. "Working memory has three original components: the central executive, the phonological loop and the visuospatial sketchpad." Now it is testable.
A useful rule: if a line in your notes does not contain a verb, it probably cannot be quizzed yet.
Step 2: Decide what each question should test
For every quiz-able line, ask what knowing it would look like on an exam. There are roughly three levels.
- Recall. Can you name or define it? "What are the components of working memory?"
- Explanation. Can you say why or how? "Why does a distractor task remove the recency effect but not the primacy effect?"
- Application. Can you use it on a case you have not seen? "A student revises in the library and sits the exam in a sports hall. Which principle predicts a small drop in recall?"
A good quiz from one page of notes mixes all three, weighted toward explanation. Pure recall questions are easy to write and easy to feel good about, but they are the ones most likely to give you a false sense of mastery.
Step 3: Write the questions (the manual way)
If you are writing by hand, use a mix of formats:
- Short answer for recall. Cover the answer, write it, check.
- Multiple choice for distinctions between similar ideas, such as proactive versus retroactive interference, where the wrong options can be the confusable neighbours. We wrote a separate guide on how to write good multiple-choice questions if you want to get this right.
- "Explain in two sentences" for mechanisms.
Write questions on one side of the page and answers on the other, or in a separate document. The single most important rule: the answer must not be visible while you attempt the question. A quiz you can see the answers to is just rereading with extra steps.
By hand, expect 20 to 30 minutes per dense page. Deciding what to ask is itself useful processing, but it is slow, which is why most students skip it.
Step 4: Or generate the quiz with AI, then check it
The faster route is to paste your notes into a generator and spend your time checking and taking the quiz instead of writing it. Here is exactly what that looks like with real output.
The input
These are psychology notes on memory, about 250 words, pasted as-is into the text to quiz generator with 10 questions at medium difficulty. (Excerpt shown; the full paste also covered Atkinson-Shiffrin, forgetting and interference.)
Levels of processing (Craik & Lockhart, 1972): deeper (semantic)
processing leads to better recall than shallow (structural or
phonemic). Craik & Tulving 1975: words judged for meaning ("does it
fit the sentence?") were recalled better than words judged for case.
Serial position effect: primacy effect (first items recalled well -
more rehearsal, into LTM) and recency effect (last items - still in
STM). A delay of about 30 seconds with a distractor task removes the
recency effect but not the primacy effect.
Encoding specificity (Tulving): recall is better when retrieval cues
match the encoding context. Godden & Baddeley 1975: divers recalled
words better in the same environment (underwater vs on land) where
they learned them.
What came out
These questions were generated by the tool from the notes above, copied without edits.
How does the serial position effect differentiate between the primacy and recency effects?
- A) Primacy is influenced by rehearsal, recency is influenced by distraction (correct)
- B) Both effects are equally strong regardless of position
- C) Primacy is based on visual information, recency on verbal
- D) Recency is stronger due to semantic processing
Tool's explanation: "The primacy effect occurs because first items are rehearsed more and transferred to long-term memory, while the recency effect is due to the last items still being in short-term memory, as explained in the source."
In the context of encoding specificity, when is recall most effective?
- A) When retrieval cues differ from encoding context
- B) When the same environment is used for learning and recall (correct)
- C) When semantic processing is employed
- D) When rehearsal is maximized
Tool's explanation: "Recall is more effective when the retrieval cues match the encoding context, as evidenced by Godden and Baddeley's study where divers recalled words better in the same environment where they learned them."
These are decent. The encoding specificity question uses "semantic processing" as a distractor, which is a real idea from the same page and a plausible wrong answer for someone who has blurred the two theories together. That is what a good distractor does.
What it got wrong
Two of the ten had a problem, and you should expect this with any generator.
The question on proactive interference marked "It disrupts the retrieval of old information" as correct. That is the definition of retroactive interference. The tool's own explanation got it right ("old information disrupts the learning and retrieval of new information"), which is the tell: when the explanation and the marked answer disagree, trust neither and check your notes.
A question on the recency effect keyed "After a short delay without distraction" as correct, when the notes say the delay has to be filled with a distractor task. One word flipped the meaning.
Both errors sit on distinctions students commonly confuse, so catching them was useful revision in itself. The general lesson: an AI quiz is a first draft. Whenever your answer disagrees with the key, go back to the source before accepting either one. For more on using AI without letting it do the thinking for you, see using AI without cheating yourself.
How many questions per page of notes
A practical ratio: about 5 to 8 questions per page of dense notes, or roughly one question per 30 to 50 words of real content. Fewer than that and whole ideas go untested. Many more and you start testing trivia, like dates in parentheses, instead of ideas.
For a full lecture's worth of notes, 10 to 20 questions is the right range for a single sitting. The Notiq generator offers 10, 15 or 20 at a time. If your notes are longer than one lecture, split them by topic and make a separate quiz per topic, so you can see which topic your mistakes cluster in.
For planning how many questions to do across a whole exam, we have a separate guide on how many practice questions to do before an exam.
When to take the quiz
Timing matters more than most people expect.
- Not immediately after writing the notes. Everything is still in short-term memory, so you will score close to 100% and learn nothing about what you will retain.
- The next day is the first useful attempt. Enough forgetting has happened that retrieval takes effort, and effortful retrieval is what strengthens memory.
- Again two or three days later, with the questions shuffled or regenerated. If you take the identical quiz repeatedly you start remembering the answer's position rather than the idea.
- Again the week before the exam, mixed with quizzes from other topics.
This spacing is the same logic behind spaced repetition for flashcards. If your notes are mostly definitions and terms rather than mechanisms, you may get more from cards than from a quiz; the flashcards from text tool takes the same paste.
What to do with the results
The score is the least interesting part. What matters:
- Every wrong answer becomes a note fix. Go back to the page and rewrite that line so it states the idea clearly. Often the reason you got it wrong is that your note was a fragment.
- Every lucky guess counts as wrong. If you were not sure, mark it.
- Retake only the missed questions the next day, then the full quiz a few days after that.
The takeaway
Notes become useful when they turn into questions. Clean up fragments so every line states a relationship, aim for 5 to 8 questions per dense page, wait at least a day before the first attempt, and treat every wrong answer as a pointer back to your notes. Generating the quiz with AI saves the time for actually taking it. Just check the answer key, because the tool will occasionally be wrong in exactly the places you are most likely to be wrong too.
If you want to try it on your own notes, paste them, or attach a photo, PDF, DOCX or PPTX, into the free Notiq text to quiz tool. The first five questions show without an account.
Frequently asked questions
Can I make a quiz from handwritten notes?
Yes. Take a clear, well-lit photo of each page and attach it to the text to quiz tool; the text is extracted before the quiz is generated. Messy handwriting and arrows between boxes lose information, so if a page is mostly diagrams, add a few typed sentences explaining the relationships.
How long should my notes be to get a good quiz?
At least a few full paragraphs. Around 150 to 400 words of real content gives a solid 10-question quiz. Very short notes produce repetitive questions, and the tool accepts up to about 20,000 characters, roughly 3,000 words, per quiz.
Should I take the same quiz more than once?
Yes, but not back to back. Retake it after a gap of a day or more, and shuffle the order or regenerate it so you are answering from the idea, not the position of the answer. Focus repeat attempts on the questions you missed.
Are AI-generated quizzes accurate?
Mostly, but not always. In the example above, 2 of 10 questions had a miskeyed answer. Treat the key as a draft: when your answer and the key disagree, check your notes before accepting either one.
