The 7-point checklist (works for Reflex, Marrow, or anything else)
1 · QBank quality over video polish
Score improvement correlates with questions attempted and reviewed, not hours watched. Compare: total questions, image-based question share, and how explanations teach the wrong options.
2 · Built-in spaced repetition
Does the app resurface your incorrect and flagged questions on a schedule? If revision is manual, you are the algorithm — and you'll skip it in the ward-posting weeks.
3 · Recency of updates
NEET-PG recall patterns shift yearly. Check the date on recent-exam question banks and how fast image banks and guideline changes (e.g., new hypertension cutoffs) get patched in.
4 · Test-mode fidelity
Grand tests should mirror the real interface, timing, and negative marking (+4/−1). Percentile predictions only mean something with a large, serious test-taking cohort.
5 · Analytics that change behavior
Subject-wise accuracy is table stakes. Look for time-per-question analysis and weak-topic heatmaps — the data that tells you what to do tomorrow.
6 · Offline & low-bandwidth mode
Internship means erratic connectivity. Downloadable videos and an offline QBank are worth more than any exclusive lecture series you can't open in the hostel.
7 · Switching cost honesty
Your notes, bookmarks, and error log live inside the app. Whichever you pick, plan to commit — a mid-prep switch typically costs 2–4 weeks of re-orientation. Trial both for a week before subscribing, not after.
Worked example: the 30-day retention gap
Suppose you learn 40 pharmacology facts today. With zero reviews, the decay model above leaves ~5 recallable at day 30 (≈13%). With three spaced reviews (days 2, 7, 21 — about 25 minutes total), retention lands near 74%: ~30 facts. Same lectures, same app, 6× the yield for 25 minutes. That's why the checklist weighs scheduling features so heavily — the best app is the one that makes those three reviews impossible to skip.