The Days Solo Prep Actually Fails
My own average mood rating across 340 logged sessions is 3.60 out of 5. Not low, not high - a shade above "fine." That number is the honest baseline of CAT prep, and it's also exactly the kind of session solo discipline is worst at protecting. Nobody skips a session that feels like a 5. People skip the 3s - the days where nothing is wrong exactly, but nothing is pulling you toward the desk either. Solo prep has no mechanism for those days beyond whatever willpower you have left, which is precisely the resource that's lowest when the day's mood would only ever be a 3 anyway.
A study group - even an informal one of two or three people - adds a second mechanism that doesn't depend on how motivated you feel: you show up because you said you would, and because someone will notice the gap in your log if you don't. That's not a replacement for wanting to study. It's a backstop for the days wanting isn't going to happen.
What Actually Works
Parallel accountability, not co-studying
The format that holds up isn't solving problems together in real time - it's each person doing their own focused work, on their own schedule, but visible to each other. Sharing what you actually logged - subject, duration, maybe a note - takes under a minute and creates real social pressure to have something worth sharing, without the coordination overhead of matching schedules. I log VARC, DILR and QUANT sessions the same way whether or not anyone else sees them, but knowing a friend can see the log changes which sessions get skipped.
Comparing trends, not single mocks
A friend group that compares mock trends across weeks stays motivating. One that compares each individual mock score turns demoralising fast, and for a concrete reason: mock-to-mock variance is genuinely large. In my own logged data, section scores that were identical landed at meaningfully different percentiles depending on which mock they came from - see the percentile vs score numbers for exactly how much a single mock can mislead. Comparing raw scores after every mock means comparing a signal that isn't reliable yet. Comparing four-week trends filters most of that noise out.
Small groups, not large ones
Accountability weakens fast as group size grows past three or four people. In a large group, any one person's absence is easy not to notice, which defeats the entire point of the arrangement. Two or three people who genuinely check each other's logs beat a twenty-person group chat that goes quiet after the first week - and most of them do go quiet, because nobody in a group that size feels individually seen.
Discussion time, not primary practice time
Solving problems together in real time is a weak substitute for focused individual practice. CAT rewards a mental state you reach on your own, under time pressure, without someone else nudging you toward the next step of a DILR puzzle. Use shared time for accountability check-ins and discussing what went wrong in a mock, not as the format where the actual reasoning practice happens.
Why "Just Text Each Other" Usually Doesn't Survive
The failure mode isn't that people don't want accountability - it's that an informal group has no structure to fall back on once the initial enthusiasm fades. Week one, everyone texts their totals. Week three, half the group has gone quiet, and nobody wants to be the one who points it out, because pointing it out feels like an accusation rather than a shared habit. The groups that survive past the first month are the ones that made the mechanics boring and automatic from day one - the same log, the same format, the same time of day - rather than relying on everyone independently remembering to type out how their day went. A shared, pre-existing log that updates itself removes the one step that keeps failing: someone has to remember to report.
This is also why comparing subjects, not just totals, matters for keeping a group honest with itself. My own logged subjects are VARC, DILR, QUANT and a reading-habit subject I call NEWSPAPER. A friend who only sees my total minutes for the week has no idea I spent three days avoiding DILR entirely - the breakdown is what makes avoidance visible, to me and to them, in a way a single number never would.
Setting Up a Group That Doesn't Fizzle
- Fix a check-in cadence. Daily works well for the first month while the habit is still fragile; it can taper to every two or three days once consistency is established on its own.
- Share logs, not feelings. "Studied a lot today" isn't accountability. "45 min VARC, 3 RC passages, 70% accuracy" is - it's falsifiable, and it's the same level of specificity I log for myself regardless of who's watching.
- Agree upfront on what counts. Decide what a "session" means before the first week, so nobody is quietly softening the definition on their light days.
- Celebrate consistency, not scores. The group's job is reinforcing the daily habit, not ranking members by mock percentile - which, per the point above, is too noisy to rank anyone by after a single mock anyway.
Visibility as a Leaderboard, Used Well
A leaderboard of study consistency - streaks and logged sessions, not scores - works because it turns an otherwise invisible question, "did you actually show up today," into something social and mildly competitive, at low stakes. That's a different thing from comparing percentiles, which the section above already covers the risk of. Consistency is a fairer, less noisy thing to put on a leaderboard than a single mock's result, because a bad slot on one paper doesn't change how many days you showed up.
That visibility matters most right before a deadline, not in the comfortable middle of prep. I'm currently working toward a weekly target of 1,260 minutes with a hard date at the start of September to be hitting it consistently. That kind of ramp is exactly where a 3-out-of-5 day is most likely and most costly to skip, and it's exactly the situation a visible leaderboard is built for - not to shame a slow week, but to make it socially awkward to let the gap sit unaddressed for three days instead of one.
How Karma Yogi's Friends Feature Fits This
Karma Yogi's Friends page is built around exactly the parallel-accountability model above: a leaderboard ranked by logged sessions and streaks rather than raw mock scores, plus visibility into friends' consistency without turning it into score comparison. It's the same mechanism a good study group builds by hand, just running without depending on someone remembering to post in a chat that inevitably goes quiet. Add a study friend and start your streak together, free.
End of essay
- Anish Guruvelli