CAT Study Plan for 3 Months vs 6 Months, Week by Week

Six months buys you a second pass over the syllabus; three months does not, and that single difference is what every other difference between these plans comes from. CAT 2026 is on 29 November, so a 13-week plan starts 31 August and a 26-week plan started 31 May. The arithmetic that decides which you are running is unforgiving: matching a 26-week plan at 16 hours a week means 32 hours a week for 13 weeks, which is a full-time schedule. You cannot compress the total. You can only compress what is in it. My own log is the cautionary number here - across 20 mocks my DILR mean moved 2.2 marks while QUANT moved 8.8, and DILR is precisely the section that needs the slow build a compressed plan removes first.

What follows is both timelines written out week by week, the four things a 13-week plan actually sacrifices, and the cut order for when a plan meets a bad fortnight - which it will.

The fixed facts both plans are built against

Everything below derives from the paper, not from preference. CAT is three independently timed 40-minute sections and you cannot borrow time between them, which is why "finish the syllabus" is a weaker goal than "be able to select and abandon under a 40-minute clock".

The full pattern, including the Non-MCQ rule that cost me marks before I read it properly, is in the pattern and syllabus post.

The two timelines side by side

Same destination, different amount of room. What changes is not the content list - it is how many times you get to go round it.

The 6-month plan, week by week

Twenty-six weeks, starting 31 May for a 29 November exam. Hours assume 15-18 a week, which is the bracket I would defend for someone with a job. Everything here overlaps deliberately - the phases are not sequential blocks with clean edges.

The 3-month plan, week by week

Thirteen weeks, starting 31 August. This is the plan most people asking the question are actually on - as I write this on 10 September, the exam is 80 days away, which is week two.

What three months actually sacrifices

This is the part most comparison posts skip, and it is the reason to read one. Four things go, in this order:

1. The second pass, entirely. In a 26-week plan you cover the syllabus, then cover it again with an error log open. That second pass is where a topic stops being something you have seen and becomes something you can do under a clock. A 13-week plan has one pass and then mocks. Everything after week 7 is repair work on whatever the first pass left broken.
2. Roughly half the analysed mocks. The useful count is not mocks taken but mocks analysed with room to act on the analysis. Twenty-six weeks supports something like 22-26; thirteen weeks supports 12-14. My own log has 20 attempts in 90 days with a median gap of six days - and seven of those nineteen gaps were two days or less, which produced no usable signal at all because nothing could be revised in between. The mock-count post has the full gap analysis.
3. DILR, disproportionately. This is the specific cost I can put a number on. Across my 20 mocks, QUANT moved from a first-five mean of 26.8 to a last-five mean of 35.6, and DILR moved from 18.4 to 20.6. DILR improves through exposure to set types over months, not through drilling a topic list, and exposure is exactly what a compressed calendar cannot manufacture. If you are on 13 weeks, protect the daily DILR set above almost everything else, because it is the one that will not respond to a late surge.
4. Spacing. The same total hours spread over 26 weeks are worth more than over 13, and this is one of the better-evidenced claims in the whole learning literature. In a dataset of 854,064 online game players, a 24-hour gap between practice sessions was worth roughly 50% extra practice at the same volume. A compressed plan is massed practice by construction. The revision-schedule post covers what you can still do about it.

Which plan you are actually on

Not a matter of preference. Count backwards from 29 November and be honest about the weekly hours you have already demonstrated, not the ones you intend.

The cut order, for when the plan meets a bad fortnight

Every plan meets one. A plan with no stated cut order gets cut wherever the pressure lands, which in practice means the full mock goes first - the one item that produces the data telling you whether the rest is working. My own order, in the sequence things get dropped:

1. Extra Quant practice beyond the weak-topic list
2. Sectional mocks
3. New topics not yet started
4. Never: the weekend full mock and its analysis

Write yours down before you need it. The decision is much worse when made at 11pm on a Thursday in a bad week.

The thing neither plan can fix

Both timelines above are about coverage and cadence, and neither addresses the harder problem: your raw score can stay flat for weeks while you are genuinely improving, because paper difficulty varies and raw score is a noisy instrument. In my own log the mean overall score across mocks 1-5, 6-10 and 11-15 was 61.8, 59.8 and 62.0 - essentially flat for ten weeks - while the mean percentile over the same stretch went from 73.4 to 80.5. If you plan against raw score you will rewrite your plan in week 8 for no reason. Plan against percentile and section trend. The percentile post has 45 real score-to-percentile pairs showing how wide that spread gets.

Where this is weak

• Verify the exam date yourself. Every week count here derives from 29 November 2026. Check iimcat.ac.in, and if the date moves, shift every block by the same number of weeks rather than re-planning.
• No study has ever tested a CAT study plan. The phase lengths above are my construction from the paper's structure and my own log, not a finding. Anyone presenting a week-by-week CAT plan as evidence-based is overstating it, including where this post sounds confident.
• The mock-count ranges are estimates. Twenty-two to twenty-six over 26 weeks and 12-14 over 13 weeks are what a weekly-plus cadence with real analysis supports. They are not measured against outcomes, because nobody has that data.
• My DILR number is one person, badly confounded. DILR moved 2.2 marks in my log, and I practised it in blocks rather than across set types. That is consistent with "DILR needs long exposure" and also with "I practised it wrong", and my log cannot separate the two.
• The spacing evidence is from a browser game. The 854,064-player study measured rapid perception and motor responses over hours, not exam preparation over months. The direction transfers; the 50% figure does not.
• I have not sat CAT. Everything here is measured against mock percentiles computed on the cohort that sat that specific paper, which is a much smaller and more self-selected group than the real exam's.
• Hours brackets assume you can protect them. A 32-hour week is arithmetically what the compressed plan needs to match the long one. For most people with a job it is not available, which is the real reason the short plan sacrifices content rather than intensity.

Sources

• Stafford, T., & Dewar, M. (2014). Tracing the trajectory of skill learning with a very large sample of online game players. Psychological Science, 25(2), 511-518. Read in full via the White Rose postprint. The spacing figure is from that text.
• Mock log, 20 attempts, 10 May to 8 August 2026, with per-section scores and inter-mock gaps - first-party, published in full in the 20-mock post.
• Session log: 340 sessions, 20,100 minutes, 135 active days - first-party, from my own Karma Yogi account.
• CAT paper structure: three 40-minute sections, +3 per correct answer, -1 on incorrect MCQs and no penalty on Non-MCQs, per the pattern reference.

Whichever plan you are on, the number that tells you it is working is the section trend across mocks, not the weekly hour total. Karma Yogi keeps both by date, so a flat raw score with a rising percentile looks like what it is instead of like failure.