Most weekly reviews collapse into “I studied a lot” or “I was lazy.” Useful reflection is more specific: which sessions produced retrieval, which only produced re-reading, where understanding fractured, and what to change next week. The 15 prompts below run a structured Sunday-night audit on your study week — and they pair well with the free study features that shipped across the major assistants in the past year.
TL;DR
- Reflection only works if you reflect on data (logged hours, topics, quiz results), not memory. Track during the week; audit on Sunday.
- The single highest-value habit is separating retrieval (recall, problem-solving, writing without notes) from recognition (re-reading, highlighting, watching). Recognition feels productive and teaches almost nothing.
- Paste your week’s log into any of the 15 prompts below. For tutoring-style follow-up, ChatGPT Study Mode (free on all plans since July 2025) and Khanmigo answer Socratically instead of handing you the answer.
- Time-box reflection to 20–30 minutes and always end with one concrete behavior change.
Who this is for and when to use it
University students, self-directed learners, exam preppers, MOOC marathoners — anyone trying to convert effort into measurable progress. Best moments: the Sunday-night weekly review, an end-of-module checkpoint, a post-exam debrief, or a tutoring follow-up.
Skip these prompts if you have not tracked time or content this week. Reflection without data becomes wishful thinking, and an AI will happily generate plausible-sounding insights from nothing.
Which AI to run these in (June 2026)
All 15 templates are model-agnostic, but the surrounding features differ. Current free study options and tiers:
| Tool | Best for reflection | Free tier (June 2026) | Notable feature |
|---|---|---|---|
| ChatGPT (GPT-5.5) | Socratic follow-up, quiz generation | Free $0; Study Mode on all plans | Study Mode guides step-by-step instead of answering; uses memory if enabled |
| Claude (Sonnet 4.6 / Opus 4.7) | Long logs, honest critique | Free $0 (limited Sonnet 4.6) | 1M-token context — paste a full month of logs without trimming |
| Google Gemini 3.1 Pro | Workspace users, big paste-ins | Free tier with large context | Reads pasted chapters; Google AI Pro is $19.99/mo (renamed from Gemini Advanced early 2026) |
| NotebookLM | Turning sources into review material | Free, roughly a few audio overviews per day | Study Guides, Briefing Docs, and podcast-style Audio Overviews from your own notes |
For ruthless honesty on a long study log, Claude’s 1M-token context (standard on Sonnet 4.6 and Opus 4.7) lets you paste a whole month without trimming. For step-by-step coaching rather than a graded report, run the reflection inside ChatGPT Study Mode, which uses custom pedagogy instructions built with researchers from 40+ institutions to keep responses Socratic.
Prompt structure formula
Every reflection prompt below carries six elements. When a result is too generic, the missing element is usually the constraint or the output format:
- Role — research tutor, peer reviewer, exam coach, debate partner.
- Context — your level, subject, deadline, course or program.
- Goal — one concrete deliverable: 12 quiz items, a 1-page lit matrix, a 4-week revision plan.
- Constraints — word count, depth, allowed sources, what to skip, what to never claim.
- Output format — numbered list, table, JSON, or graded blocks (E / M / H) so you can paste into Notion, Anki, or Word.
- Examples / signal — 1–2 reference paragraphs or anti-examples (“not the way Wikipedia explains it”).
15 copy-ready prompt templates
Replace each [bracketed] placeholder with your own input before sending.
1. Feynman-gap reflection
Surfaces where understanding actually broke.
You are a study coach. Below are the topics I studied this week: [topics]. For each, I will try to explain it in 3 sentences. Identify the sentences where my explanation is vague, missing a mechanism, or relies on a borrowed phrase rather than understanding. Suggest 1 follow-up question per gap.
[paste my explanations]Variables to swap: topics, explanations
Optimization: If the model is too gentle, add: “Be ruthless. Mark any sentence that a knowledgeable reader could not paraphrase back to you.”
2. Retrieval-vs-recognition audit
Below is my study log for the week. For each session, classify the activity as primarily retrieval (active recall, problem solving, writing without notes), recognition (re-reading, highlighting, watching), or hybrid. Total hours per category. Flag the imbalance if recognition dominates.
[paste log]3. Time-on-task honesty
I budgeted [planned hours] for [topic]. Actual time: [actual hours]. Below are time entries: [paste]. Identify (a) hours that were truly focused, (b) hours that were context-switching or shallow. Suggest one schedule change for next week.4. Confidence-vs-accuracy gap
Below are 10 self-quiz items I attempted. For each: my confidence (1-5) before answering and whether I was correct. Identify where confidence overshoots accuracy (overconfidence zone) and suggest specific drills.
[paste]5. Energy and focus pattern
Here is my hourly study log for the week with focus self-rating (1-5): [paste]. Identify my 2 best-focus time windows and 2 worst. Suggest a redesigned schedule that matches hardest tasks to highest-focus windows.6. Weekly takeaway: 3 lines
Compress my study week into 3 lines: one biggest insight, one biggest gap, one promise to myself for next week. Use my own voice, not corporate-speak. Inputs: [paste reflections, logs, errors].7. Mistake taxonomy
Below are all the errors I made in homework and quizzes this week. Group them into 3-5 categories (e.g., misread, computational, conceptual, careless). For each category: a 1-line root cause and a targeted drill.
[paste errors]8. Spaced-review forecast
Topics studied this week and their first-pass dates: [paste]. Build my spaced-review schedule for the next 4 weeks (days 1, 3, 7, 14, 30). Output as a calendar-friendly list.The 1 / 3 / 7 / 14 / 30 expanding interval reflects the spacing effect: reviewing just as memory is about to decay beats massed re-reading. Spaced repetition combined with retrieval practice is one of the most strongly evidenced study strategies in cognitive psychology, so it is worth scheduling rather than improvising.
9. “Did I understand or just remember?” test
For each of the topics [list], generate 1 question I should be able to answer if I truly understand the mechanism (not just recall the term). I will try to answer; flag the ones where I fall back on memorized phrasing.10. Teaching-test rehearsal
I will explain [topic] as if teaching a friend who has no background. Pretend to be that friend; ask 4 naive follow-up questions. Stop when the answer becomes circular or hand-wavy; mark that as a gap.11. Plan-vs-actual variance
Compare last week's plan to last week's actual: [paste plan, paste log]. Identify the top 3 variances. For each: likely cause (overestimation, distraction, scope creep, illness), one concrete fix.12. “What I would tell past-me” letter
Write a 200-word letter from my future self (Sunday next week) to my current self (today), based on this week's data: [paste]. Voice: kind but specific. Include one warning, one encouragement, one practical instruction.13. Next-week one-page plan
Based on this week's reflection inputs: [paste], build next week's study plan: top 3 priorities, time budget per topic, 2 retrieval sessions per priority, 1 buffer slot, 1 weekly review slot. Constraint: max [hours per week] hours.14. “If I had only 4 hours” focus
Given my current weak spots [list weak spots], if I had only 4 hours next week, what is the highest-leverage way to spend them? Output as 4 one-hour blocks with specific activity and expected outcome.15. Mid-term checkpoint summary
For my [N]-week course on [topic], here are my weekly summaries so far: [paste]. Write a 250-word mid-term checkpoint: what I have mastered, what is still shaky, what is missing entirely, what to adjust in the second half.Common mistakes
- Reflecting without data — write down hours and topics during the week, not from memory on Sunday.
- Confusing “felt productive” with “produced retrieval.” Recognition activities are the usual culprit.
- Vague conclusions (“study more X”) instead of specific drills.
- Ignoring confidence calibration — overconfidence costs more than underconfidence because it stops you reviewing.
- Repeating the same plan after a bad week without changing the variable that broke.
- Skipping mistake taxonomy — error patterns repeat unless named.
- Reflecting only on outcomes, not on process; both matter.
How to push results further
- Keep a single doc per week: plan at top, log in middle, reflection at bottom. Paste the whole thing in one go — Claude and Gemini handle the length.
- Time-box reflection to 20–30 minutes; longer becomes rumination.
- Always end with one concrete behavior change, not a generic vow.
- Rotate templates so you do not just answer the same questions every week.
- Pair the retrieval audit (template 2) with the confidence audit (template 4) once a month.
- Use template 12 (letter to self) when motivation is low; it tends to work better than abstract goal-setting.
- Feed the output of template 8 (spaced-review forecast) into NotebookLM to auto-generate review questions or an audio overview from your own notes.
- Track which reflection insights you actually act on; if it is zero, your reflection format is broken.
FAQ
- How long should a weekly reflection take?: 20–30 minutes. Longer tends to spiral into rumination; shorter usually means you are not engaging with the data.
- Which AI is best for this?: For honest critique of a long log, Claude (Sonnet 4.6) or Gemini 3.1 Pro handle the length. For step-by-step coaching that asks instead of tells, use ChatGPT Study Mode (free on all plans) or Khanmigo. There is no single winner — the prompt matters more than the model.
- Can AI tell me how to study?: It can structure the reflection and point at obvious imbalances. It cannot replace knowing your subject, your goals, or your energy.
- Should I share weekly reflections with a tutor?: If you have one, yes — the patterns across weeks are far more useful than a single session report.
- What if every week looks bad?: The problem is usually upstream of reflection: workload, sleep, or motivation. Reflection cannot fix structural issues; talk to a mentor.
- Is reflection useful even when results are good?: Yes. Successful weeks teach you replicable patterns; without reflection you cannot copy them.
Related
- Study plan prompts
- Exam revision plan prompts
- Flashcard prompts
- Lecture notes cleanup prompts
- AI Helps You Run a Weekly Study Reflection
- Learning & Research Prompts hub
- External: Retrieval and spaced practice — Evidence Based Education
Tags: #Prompt #Study #Study #Study plan