Habits & Routines

Build a Daily Study Habit That Survives Busy Weeks

Keep showing up and keep remembering: a 10-minute minimum session for bad days, a weekly plan, and review dates set by your test date, with the evidence and its limits.

By Editorial team

Published 10 min read

Ask this guide

Answers come straight from this guide’s text. Nothing is generated, and nothing you type leaves your browser.

0 of 5 doneSaved on this device only.

Say you study every evening for nine days. A work trip and a deadline swallow the tenth, the streak counter shows zero, and the plan quietly dies. The usual advice, "same time every day, start small", keeps you showing up. It does not say what to do with ten minutes on a bad day, or when old material should come back.

This guide covers both halves: a minimum viable session, a weekly plan, and review dates set from your test date. Each piece cites named studies, with their limits. The minimum session, the weekly template and the way they combine are our design, labelled as such and untested.

A daily habit and spaced review answer different questions

A daily habit answers "do I show up?" Lally and colleagues (2010) had 96 volunteers do one chosen behaviour daily in the same context for 12 weeks. Of 82 with enough data, a rise-then-level-off curve fitted 62, and missing one opportunity did not materially affect the process (details in our habit-formation guide). Stojanovic and colleagues (2020) saw a similar rise in 91 university students who tied a 10 to 30 minute study habit to a daily activity. That sample was university students, mostly young and 79% female, earning course credit, with no control group.

Spaced review answers "when does this come back?" Cepeda and colleagues (2008) found that the best gap between two study sessions depends on how long you need to remember. As a share of the test delay, the best gap was about 20 to 40% for a one-week delay and 5 to 10% for a one-year delay. By our arithmetic, that is one to three days when you need to remember for a week after the review, and three to five weeks for a year.

So the two ideas can disagree. Reviewing the same material daily suits a test a few days off and is too tight for anything you need next semester, while a daily session of only new material gets no spacing, because nothing returns. The system below keeps them apart: the daily session is a container, and a due list decides what goes in it.

What the evidence supports, and where it stops

Study Design Finding Limits
Cepeda et al. 2006, Psychological Bulletin Meta-analysis: 839 assessments, 317 experiments, 184 articles, all verbal recall Spaced beat massed (back-to-back) study in all but 12 of 271 comparisons; the best gap grew with the test delay 80% of gap comparisons had test delays under a day; 85% of data from young adults
Cepeda et al. 2008, Psychological Science 1,350+ adults in an online panel learned 32 obscure facts, reviewed once after gaps of up to 3.5 months, tested up to a year later The best gap rose with test delay but shrank as a share of it One review, not a schedule; trivia facts
Karpicke & Roediger 2008, Science College students learned 40 Swahili-English pairs; after one correct recall, pairs were dropped from study, from testing, from both, or kept in both A week later, conditions that kept testing recalled about 80%; those that dropped testing recalled 36% and 33%, however much they restudied One lab experiment with word pairs
Dunlosky et al. 2013, Psychological Science in the Public Interest Review of 10 techniques Practice testing and distributed practice rated high; interleaving moderate; rereading, highlighting and summarising low Authors' ratings, not a meta-analysis
Rohrer et al. 2020, Journal of Educational Psychology Preregistered trial: 54 seventh-grade maths classes, 787 students, 4 months 61% vs 38% on an unannounced test a month later (d = 0.83) One district, one subject
Brunmair & Richter 2019, Psychological Bulletin Meta-analysis of interleaving: 59 studies g = 0.42 overall, 0.34 for maths, but blocking beat mixing for words (g = -0.39) Effects depend on the material

Retrieval practice (testing yourself) and spacing have the widest support. The Karpicke and Roediger result matters for a bad-day session: once an item had been recalled once, restudying added nothing a week later, while more testing added a lot. Our learning with AI guide covers their 2006 result.

Note: Nothing above tests a 10-minute floor, a weekly template, or the review dates in the planner. Those are ours.

The minimum viable session, by minutes available

The rule: a bad-day session is retrieval from your due list, never rereading and never new material. That keeps the cue alive, uses the technique with the best evidence, and stops due reviews piling up. The 10 minutes and five questions are a judgement call, not a threshold from a study.

Time you have Session What you do What you skip
3 minutes Rescue Answer one due question from memory, check it, leave the rest overdue Everything else
10 minutes Minimum viable Answer up to 5 due questions closed-book (in week one, write 5 from your notes instead), check each against your source, mark done, log the score ("4/5") New material, tidying notes
25 minutes Standard 5 minutes on due questions, 15 on new material or practice, 5 writing 3 to 5 new questions
50 minutes Long Standard, plus a 15-minute mixed-practice block

Rescue and minimum sessions both count as study days in your 14-day total; our density-target guide suggests 9 or 10 of 14 for an effortful daily habit (a heuristic, not a research threshold). After three rescue days in a fortnight, cut the week's new topics instead of lowering the bar again.

A weekly plan with a fixed cue and rotating content

Stojanovic's team argued that study habits are more fragile than a fixed behaviour such as going for a run, because the content changes and not every step is cued each time. That is their argument, not a tested result. Our response: make the habit the start ritual, which never changes, and let the due list supply the content.

Write the cue as an if-then plan (how to write one): If it is 19:30 and I am at the kitchen table, then I open the due list. Same place, same first action.

WEEKLY STUDY PLAN (our template, untested)
Cue: If ________ at ________, then I open the due list.
Every day: due questions first. Bad day: 10 minutes, due questions only.
Goal: ________   Test or goal date: ________   Mode: exam / lifelong

Mon  25 min     new topic A
Tue  25 min     topic A practice
Wed  25 min     new topic B
Thu  25 min     mixed practice A + B (only if you can do each alone)
Fri  10-25 min  catch up on topic B
Sat  20 min     unaided check: 10 questions, no notes, score strictly
Sun  10 min     plan: count the last 14 days, clear overdue,
                add new questions to the planner

Rules: bad-day sessions count. Never make up a missed day; reschedule
its reviews. Cut new topics before cutting the due list.

Mixed practice: when to interleave and when not to

Interleaving means switching between problem types in one session (a SQL join, then a window function, then a subquery) instead of finishing one type first. Rohrer's team lists caveats: teachers reported interleaved sets took longer, the benefit may shrink at short test delays, students probably had some blocked practice first, and everyone saw the correct solutions. So interleave only after you can solve each type alone, check your answers, and expect it to feel slower. Use it where choosing the method is part of the skill: maths, and by extension statistics and code. For vocabulary, the meta-analysis favoured blocking.

Review dates: exam mode and lifelong mode

The free Spaced Review Planner runs in your browser and uploads nothing. You add what you learned, preferably as a question, pick a schedule, and open "Due today and overdue" each day. Answer from memory first, check your source, then mark it done or move it to tomorrow. It can export every review to a calendar (.ics).

Goal Planner preset (days after learning) Dates if learned Mon 19 Oct What it rests on
Test within about 2 weeks Exam crunch: 1, 2, 4, 7 20, 21, 23, 26 Oct Loosely, Cepeda 2008: for a one-week delay, the best gaps were a fifth to two fifths of it
Test in 4 to 8 weeks Standard: 1, 3, 7, 14, 30 20, 22, 26 Oct; 2, 18 Nov Convention. Gaps should grow with the delay (Cepeda 2006, 2008), but this exact set is untested
Keep for months or years Long-term: 1, 7, 30, 90, 180 20, 26 Oct; 18 Nov; 17 Jan; 17 Apr A 30-day gap sits inside the 18 to 37 days implied by the one-year ratio; the 180-day step is beyond the longest gap tested

Whether widening gaps beat equal gaps is unsettled. Cepeda's 2006 meta-analysis found 22 comparisons leaning towards widening, but with large variation between studies, and three earlier sets of studies with gaps of a day or more split three ways. Treat a preset as a sensible schedule, not an optimum. A custom list also works: up to ten whole days from 1 to 365, ascending.

A hypothetical fortnight, with real numbers

Say you are a career-changer learning SQL for an exam on Monday 23 November, studying after dinner at the kitchen table with a 10-minute floor. On Monday 19 October you add "When does ROW_NUMBER() restart?" to the planner with the Standard preset. It comes due on 20 and 22 October, then 26 October, 2 November and 18 November, five days before the exam.

Week 1:  Mon full | Tue floor | Wed full | Thu rescue | Fri full  | Sat full | Sun miss
Week 2:  Mon full | Tue miss  | Wed full | Thu full   | Fri floor | Sat full | Sun full

Tuesday's late train leaves 10 minutes, so you do the floor. On Thursday a delay leaves 3: you answer one due question and leave the rest overdue. The fortnight has 9 full sessions, 3 floor or rescue sessions and 2 misses, which is 12 of 14 days and above a 10-of-14 target. Reviews due on the missed days wait in the overdue list and are cleared oldest-first in the next session, with no make-up session.

Where to start this week

  • Pick one cue that already happens daily and write the if-then line.
  • Define your 10-minute floor in one sentence, such as "5 due questions, closed book".
  • Add this week's topics to the planner as questions, and choose a preset by test date.
  • Book a 10-minute Sunday planning slot.
  • After two weeks, count sessions in the last 14 days; adjust the plan, not the bar.

More routines are in the habits topic hub.

Frequently asked questions

Is it better to study a little every day or a lot a few times a week?

The studies behind this guide do not compare the two directly. The habit studies measure how automatic a daily behaviour becomes, not how much was learned; the memory studies measure gaps between repeat exposures to the same material. They support spreading study across days instead of one sitting, and timing reviews by how long you need to remember. A daily session helps you show up, but spacing does not require one.

What counts as a minimum study session on a bad day?

Ours is 10 minutes: up to five due questions answered closed-book, checked, marked and logged, with no new material. With only three minutes, answer one due question and leave the rest overdue. This is our design, not a tested threshold. It rests on the finding that testing beat restudying a week later.

How many days can I miss before I lose my progress?

In Lally and colleagues' data, one missed day did not materially affect habit formation. In Stojanovic and colleagues' sample, repetitions preceded by a pause of a day or more did not significantly slow the rise in automaticity. Neither source covers long lapses. For memory, a review a day or two late probably costs less than one done too soon: in Cepeda 2008, scores climbed steeply as the gap grew and declined gradually after the peak (our reading of their curve). After a longer gap, do the oldest overdue questions first and leave the rest for tomorrow.

Do I need the planner, or can I use flashcards or a paper calendar?

The method is the due list: items, dates and answering before checking. A paper calendar or any app that shows what is due today works. The planner adds presets, rescheduling and calendar export.

Sources and further reading