The Best Way to Memorize for an Exam: 14-Day Protocol

Why does rereading feel effective but fail on exam day?

Rereading a 100-page PDF feels productive because it builds fluency: the material looks familiar, so your brain mistakes recognition for recall. On exam day, when nothing is in front of you, that familiarity is worthless. You need retrieval, not recognition.

The fix is testing yourself. In a widely cited study, Roediger and Karpicke found that students who practiced recalling a passage retained far more after a week than students who reread it the same number of times, even though the rereaders felt more confident (Psychological Science, 2006).

So the protocol in the rest of this article is built around retrieval first: close the PDF, write the formula or definition from memory, then check. Every section assumes you spend your time producing answers, not rereading them. If you want to see a session captured end to end, the tutorials show it.

How should you memorize formulas differently from definitions?

Each material type fails under retrieval in a different way, so each needs a different encoding method. A formula you can read but not reproduce, a definition you recognize but cannot produce, a framework you half-remember, a list you drop items from. Match the drill to the failure mode.

Formulas: retrieve by solving, not by staring

Staring at a formula sheet builds recognition, and the exam demands production. Write the formula from memory first, then immediately apply it to a problem with real numbers. If you get the recall wrong, the application wrong, or both, that error is the whole point of the drill. One solved problem per formula per session beats five minutes of re-reading.

Definitions: cloze cards and elaboration

Recognition is the trap: the term looks familiar on the page, so rereading feels productive. Turn each definition into a cloze card with the key term deleted, and force production. Then add one elaboration: a case from your cold-call classroom, a contrast with a neighboring term, or a sentence in your own words. Elaboration is what separates a definition you can recite from one you can use.

Processes and frameworks: explain-back passes

Frameworks like Porter's Five Forces or the steps of a fiduciary duty analysis break at the missing step, not the middle. Close the source, explain the whole process out loud or in writing, then open the source and mark what you dropped. The dropped step becomes your next retrieval target. Two explain-back passes per session is enough.

Lists and verbatim material: chunking and mnemonics

Raw lists of five-plus items exceed what working memory holds, so chunk them into groups of three or four and give each chunk a hook: an acronym, a sentence, or a spatial layout on your iPad. Retrieval order matters: practice recalling the chunks out of sequence, because exams rarely hand you the list in the order you memorized it. The same principle applies to any practice material you generate from lectures, whether you build cards by hand or with a tool; the encoding step, not the tool, does the memorizing.

What is the optimal spacing schedule for the last 14 days before an exam?

Fourteen days is enough for three full retrieval passes plus a taper. The evidence base is the Cepeda et al. meta-analysis of 254 studies, which found that distributed practice beat massed practice across every retention interval tested, with optimal gaps growing as the test gets further away (Cepeda et al., 2006).

Days 14-10: full coverage. One pass over every topic, no skipping, building your retrieval cues as you go. Days 9-6: first retrieval passes, testing yourself on everything once, marking what fails. Days 5-2: weak-spot drilling, dropping topics you retrieved cleanly and hammering the ones you missed. Day 1: taper. One light pass over your marked weak spots, then sleep.

The long middle stretch is where the schedule earns its keep. On day 9 you retrieve the whole course cold, and whatever you fail is not a verdict, it is a map: those misses are the only things that need real time on days 5-2. Most students invert this, rereading everything evenly until the night before, which is exactly the massed practice the meta-analysis shows underperforms. If you generate practice material from your own lecture PDFs and transcripts, the day 9 pass runs faster because the retrieval cues are already in the format you will drill against, which is the workflow Praesto is built for. The schedule does not care which tool builds the cues; it cares that every day after day 9 is spent retrieving, not rereading.

How do you build retrieval practice into a daily study session?

Run every session on the same 60-90 minute structure so you never waste the first twenty minutes deciding what to do.

Minutes 0-10: cold retrieval. Close every source and write down everything you remember from yesterday's material. No peeking, no topic map open in a second window. The discomfort is the point; what you fail to pull up is exactly what you review next.

Minutes 10-50: new encoding. Work through today's material once, actively: annotate the PDF, redraw the process diagram from memory, say the definition out loud before checking it. One pass done properly beats three passes skimming.

Minutes 50-70: practice questions. Answer questions on both today's and yesterday's material, mixed. Interleaving old and new forces discrimination, which is what cold calls demand.

Minutes 70-90: brain dump. One blank page, everything from the full session, then check it against your notes and mark the gaps as tomorrow's opening retrieval targets.

A tool like Praesto can generate the practice sets from your lecture transcripts so the practice-question block starts with questions instead of you writing them. Its privacy policy covers how your recordings and transcripts are handled, which matters if your school records classes. The honest limitation: no app can do the cold retrieval for you, and that 10-minute block is where most of the retention comes from.

How do you turn lecture PDFs and recordings into practice material without losing information?

The information loss you noticed with a general chatbot is a context window problem: when a 100-page deck exceeds what the model holds at once, it summarizes and drops detail. Chunk by topic instead. Split the PDF into sections, process each separately, then merge the outputs. Stripping slide formatting to plain text before feeding it in also helps, because models handle dense structured slides poorly.

Then convert, don't just summarize. Summaries create a fluency illusion: they feel familiar, but rereading them is one of the weakest study techniques in the Dunlosky et al. review of learning methods. Generate quizzes, flashcards, and a topic map from the source material itself, and answer from memory before checking.

On tools: NotebookLM generates quizzes. Otter.ai and Jamworks handle recordings and transcripts; Playbacks does similar. Praesto targets this pipeline for a full course load, turning sessions into practice sets, though it assumes Apple devices. Whatever you use, spot-check the output against the original deck; every tool, including ours, drops edge cases. If output looks thin, use the contact page rather than trusting it blind.

How do you stay present in a cold-call classroom and still retain the material?

In a cold-call class, being called on is itself retrieval practice: you are pulling arguments from memory under pressure, which is exactly what the exam asks of you later. So the worst move is transcribing. Capture lightly instead: a few pencil notes on the iPad marking who said what and which cases the professor pushed on, nothing more.

Encode after class, the same day. Turn the transcript or your sparse notes into a practice set: flashcards for definitions, prompts for the positions argued, spaced-repetition flashcards for anything verbatim. Waiting a few hours before encoding makes the retrieval attempt stronger.

Praesto handles the capture side: it records the session and generates review material so you can stay present. Its limits are real. It cannot reconstruct a discussion it failed to record, and its output still needs your editing for nuance, since discussion-heavy classes turn on reasoning you have to check yourself.

What should you do the night before and morning of the exam?

Taper. The night before, do one light pass of retrieval only: flashcards, your formula sheet, a blank-page recall of each topic. No new material, no re-reading the 100-page PDF. Sleep is when consolidation happens, so protect it (Rasch and Born on sleep and memory). Protect at least seven hours.

Morning of: a short brain dump. Blank page, write every formula and definition you can from memory, then check against your sheet and patch the gaps. Ten minutes, done. Skip the group panic-review and the last-minute new deck; retrieval you already failed at 2am is not going to fix itself in the hallway.

FAQ

Is it better to memorize with flashcards or practice questions?

Neither format wins on its own; retrieval does. Flashcards suit definitions and verbatim material, while practice questions suit formulas and frameworks because they force application, not just recall. Match the format to the material's failure mode, and judge each by what you can produce cold, not by how familiar the material feels.

How many hours a day should I study 2 weeks before an exam?

Two focused 60-90 minute sessions per day is enough, structured around cold retrieval, new encoding, mixed practice questions, and a brain dump. More hours spent rereading buys less than fewer hours spent retrieving. Every session should start with blank-page recall of yesterday's material, and days 5-2 should target only your marked weak spots.

Can I memorize a lot of material in one day?

You can encode it, but you will not retain it. Massed practice underperforms distributed practice across every retention interval tested in the Cepeda et al. meta-analysis of 254 studies. One day is enough for a full pass that builds retrieval cues and marks what fails, but retention requires at least one later retrieval pass, ideally two, before the exam.

Do spaced repetition apps work for law and medical school material?

Yes, with a caveat about what you put in them. Definitions, drug lists, and elements of a rule fit flashcards well. Case reasoning and clinical processes need explain-back passes and applied questions, not card flips, because they fail at a missing step under pressure. The schedule works; the encoding step, not the app, determines whether material sticks.

What is the fastest way to memorize a 100-page PDF?

Chunk it by topic, process each section separately, and convert each chunk into retrieval material: cloze cards for definitions, worked problems for formulas, and explain-back prompts for processes. Then retrieve from memory before checking. Reading the PDF straight through is faster but produces recognition, which is worthless when the exam closes the book.