Practice Hub: PM Guide¶
Kwilo's exam-prep practice surface (JEE/NEET), explained for someone who has never seen the code. No AI guesses the ranking; it is one transparent rule.
Naming
This surface shipped as "Study Plan" and is now called Practice Hub: the JEE/NEET, past-paper-ranked practice engine that is live today. The name Study Plan now refers to a separate, broader bet, generating a personalised plan and practice from NCERT books or a learner's own uploaded material. See Study Plan (Vision).
Want the full visual explainer?
The richer, designed walk-through with a worked example is embedded below, and also opens full-screen: open the explainer in a new tab.
The promise¶
A learner subscribes and we say: "We tell you exactly what to study next, ranked by what the exam actually asks, weighted by where you are weak." That's only honest if the order comes from real data, computed the same way every time.
The whole idea in one line¶
For every chapter, ask two questions and multiply the answers:
Priority = how much the exam cares × how weak you are
- How much the exam cares (Importance): counted from the real past-paper question bank. Same for every learner.
- How weak you are (Gap): from your own practice history. Personal to you.
Rank every chapter by that number. The top eligible chapter becomes "study this today."
Why multiply, not add¶
Multiplying means a chapter only ranks high when both are true: the exam cares AND you're weak.
| Exam cares a lot | Exam barely asks | |
|---|---|---|
| You are weak | ✅ Study this: high payoff | Low priority, few marks |
| You are strong | Don't re-drill, you own it | Ignore, waste of time |
The plan hunts the top-left box: tested and weak. A learner's most-tested chapter can still rank low if they already know it; their weakest chapter can rank low if the exam never asks it.
The engine¶
The flow: gate (Pro + b2c_user) → resolve the learner's goal → compute Importance (pluggable per exam) × Gap (universal) → sort → pick today's chapter (respecting prerequisites) → assemble the plan. If the question bank isn't loaded, it returns an honest empty state, never a fake plan.
The three learner states¶
- New / free user: can practise the top-ranked chapter of each subject for free; every lower-ranked chapter is locked, and Pro unlocks the rest. The conversion hook.
- Pro, no history yet (cold start): accuracy shows as a dash, order driven by exam importance alone, phase diagnose.
- Active Pro user: full personalized ranking, real accuracy, readiness climbing. Every practice session re-ranks.
A "no history" learner who practices one chapter becomes "active" on the next load. The plan is never stale.
What to remember¶
One rule: study what's both heavily-tested and where you're weak. One honesty rule: never invent a score we haven't earned: show a dash or degrade to empty, rather than fake a plan. Every rank is explainable to the learner who asks "why this chapter?"
Related¶
- Practice Hub Engine (Design & Business Logic): the multi-stream architecture, goal resolution, scoring math, and API behind this rule
- Study Plan (Vision): the next bet, AI-generated plans and practice from NCERT or a learner's own material
- B2C flow: where the Practice Hub sits in the paid journey
- Practice Hub gate (feature doc): the freemium lock around it
- Full visual explainer