C:\ARCHIVE\M_001_MARGIN\case_study.txt
Margin
Every note, pinned to its second.
A study companion for students who learn from long recorded lectures. It works offline, keeps notes on the lecture’s timeline, and turns revision into minutes.
01TL;DR
Students don’t lose focus. They lose the lecture.
Students learn from 2-hour recorded lectures. 177 of 607 relevant posts are about studying offline, and 20 say a “downloaded” lecture still wouldn’t play.
Collected 10,657 public app reviews and forum posts, coded 935 against a study-moment codebook, and tested 7 hypotheses before designing anything.
An offline-first player where every note is pinned to its second, resume starts where you understood, and revision replays only your pins.
one has to search for the course and find the specific lecture
02Research
10,657 posts, narrowed to the 607 that matter.
Keyword matching alone was only 65% precise, so I coded every candidate by hand against 11 study-moment codes and kept each source link in a spreadsheet.
03Findings
Study sessions break on access, not attention.
Posts per code, n = 607. A post can carry several codes.
04Hypotheses
Seven hypotheses, tested against the data.
| ID | Hypothesis | Verdict | Evidence |
|---|---|---|---|
| H1 | Students lose their place in long lectures | Confirmed, sharpened | 97 posts. Finding fails twice: the lecture (18) and the second (17). Resume restarts from zero (9). |
| H2 | Notes live apart from the video and lose context | Confirmed | 127 posts. 15 ask to see notes beside the video; 7 say blocked screenshots broke their notes. |
| H3 | Revision is the most painful moment | Changed | 59 posts, mostly positive: revision is why students value recordings. |
| H4 | Recommendations pull students away mid-study | Rejected | 32 posts (5%), and 6 say online lectures are less distracting than class. |
| H5 | Low data and slow networks shape how they study | Confirmed, strongest | 177 posts (29%). 20 say downloads still need the internet or disappear. |
| H6 | Students build workarounds that reveal needs | Confirmed | 40 workarounds, and 114 public GitHub tools built to bookmark video moments. |
| H7 | Students would trust AI summaries with a source link | Not supported | No student asks for it, so AI summaries stay out of v1. |
05The surprise
Distraction wasn’t the problem.
I expected the feed to be the enemy. The data says otherwise.
130 reviews describe playback that freezes, lags or goes black.
The tool breaks focus, not the feed.
download the lectures and then turn off ur internet to increase focus
Design decision: no focus mode. The watch screen simply has nothing else on it.
06Landscape
No product connects the whole study loop.
| Product | Resume | Notes on a moment | Search inside | Revision | Offline |
|---|---|---|---|---|---|
| YouTube | — | No | Partly | No | Partly |
| PW (Physics Wallah) | — | No | — | Partly | Yes |
| Unacademy | — | No | — | Partly | — |
| Coursera | — | Partly | Partly | Partly | Yes |
| Khan Academy | Yes | — | — | Partly | Yes |
| Google Keep | — | No | Yes | No | Yes |
| Notion | — | No | Yes | No | — |
— = not listed in the product’s official help pages or store listing (checked 28 Oct 2026).
The gap: players handle watching and note apps handle notes. Coursera comes closest, with notes saved against a frame, but community posts report that feature missing on Android. Nothing lets a student mark where they understood, or revise by replaying their own moments.
07Insights → principles
Four insights, four design principles.
“Downloaded” doesn’t mean offline
“I downloaded some lectures but can’t play them without internet” (Adda247 review, 2023)
Offline is the defaultNotes belong next to the lecture
“I want to see the class notes side by side with the lecture” (PW review, 2023)
Every note knows its secondFinding fails twice
“the video starts playing from the beginning rather than resuming” (Udemy review, 2021)
The student’s pins are the mapRevision is the payoff
“Watch lectures at 1.5X speed (2X if review).” (Hacker News, 2019)
Revise the moments, not the lectureBacklog clearer
Works through missed recorded classes in long late-night sittings, pausing to check the teacher’s PDF. Today’s apps restart the video, resume from zero, and lose downloads.
Revision sprinter
In the days before an exam, rewatches only the parts that matter, at high speed. Today they can’t find the part again, skipping lags, and screenshots are blocked.
Behaviour modes instead of personas: the evidence describes study moments, not demographics.
08Concepts
Three concepts, one direction, two cuts.
| Concept | Score | Why |
|---|---|---|
| A. Smarter player with pins Chosen | 14 / 15 | Pins live on the player itself, which fixes access and reliability, where the evidence is (177 + 130 posts). |
| B. Notes-first notebook | 10 / 15 | Depends most on player APIs, and embedded videos can’t be clipped or frame-captured. |
| C. Exam planner | 8 / 15 | Cold start with no pins, and no help at the 11 pm first watch. Its revision mode became an entry point in A. |
- Cut: AI summaries. No student asks for them; trust needs proof first.
- Cut: a focus mode. Distraction isn’t the problem, so the watch screen simply stays empty.
09Final design
Every note pinned to its second.
Find the moment, then revise the moments
Each decision traces to evidence
- A pin is a second plus text, not a screenshot. 7 reviews say screenshots are blocked in their app.
- Search ranks the student’s own pins first. Every result shows lecture, time and source. 18 posts can’t find the lecture.
- Revise replays pins as 20-second clips, up to 2×. 59 revision posts; students review at “2X if review”.
- Downloads show size and free storage up front. A 4-hour class can cost “approx 3 gb”.
- Yellow means pin, and only pin. Pin marks carry an ink outline: yellow alone is 1.4:1 on paper, the outline 15.8:1.
10Iteration
From first wireframe to final screen.
| # | v1 problem | Fix in the final design |
|---|---|---|
| 1 | The pin hid on the video overlay, out of thumb reach | A persistent Pin button in the thumb zone |
| 2 | Notes hid behind a swipe-up sheet | Pins listed under the video, always visible |
| 3 | An unlabelled flag meant “understood” | A labelled “Understood here” marker |
11Evaluation
Two review rounds, every failing step fixed.
| Severity | Issue in v1 | Fix |
|---|---|---|
| 4 | Offline shown only as a crossed-cloud icon | A banner says what still works; every lecture shows its status |
| 3 | Pin out of thumb reach, hidden with the controls | A persistent Pin button, bottom right |
| 3 | Notes hidden behind a swipe-up sheet | Notes listed under the video |
| 3 | Silent auto-resume and an unlabelled flag | Two labelled resume choices |
| 3 | Search results without lecture or time | Every result shows lecture, time and source |
12Measuring success
How I’d measure it, with Google’s HEART framework.
| Goal | Signal | |
|---|---|---|
| Happiness | Studying feels less frustrating | Post-exam in-app rating, 1 to 5 |
| Engagement | Pins become part of studying | Pins per hour of lecture watched |
| Adoption | New users get value fast | Share of new users who pin within their first lecture |
| Retention | Used for revision, not just once | Share of users who open Revise in the week before exams |
| Task success | Students find the part they need | Median time from search to playing the right moment |
Total watch time must not rise. Margin should make studying shorter, not longer.
13Outcome
What the research proved, and what Margin does.
Study apps fail on access, not attention.
Offline and playback reliability are the two biggest codes, at 177 and 130 posts, so Margin treats offline as the default state.
Notes belong on the timeline.
A pin, a second plus text, turns a 2-hour lecture into a map the student made, searchable across every lecture.
Revision is the payoff.
Revise replays only the student’s pins: 12 moments in about 4 minutes, instead of hours of rewatching.
Next folder / R 002
Ruko →
A warning people believe, at the second it matters. Scam safety for parents aged 45+.
