Usually returns sooner
New words can disappear quickly, so Rankri checks them earlier until they become strong.
How Rankri decides what should return next.
Rankri uses modern spaced-repetition research as its base, then connects it with subject-aware scheduling, five-star recall, optional learning screens, mock imports and an offline-first study experience.
Question-level historySubject-aware timingOffline studyCross-device sync
01
The scheduler begins with one question.
Rankri does not revise an entire Topic on one fixed date. Every Question keeps its own history. A strong Question can return after more days, while a weak Question comes back sooner.
STEP 1Learn or attemptSee the question and recall the answer.
STEP 2Rate memoryChoose one to five stars.
STEP 3Check historyRankri checks past reviews and subject type.
STEP 4Choose a dateThe next return is calculated.
STEP 5Update againEvery new review changes the future timing.
02
Subjects do not use identical timing.
A newly learned word often needs to return sooner than a familiar Maths formula. Rankri uses different subject targets in the background, while the student still uses the same simple five-star review.
Follows rule strength
The system checks whether the rule stays clear across more than one example.
Strong facts return after more days
A fact that stays clear can wait longer, while a confused fact comes back sooner.
Methods are checked through questions
Wrong methods return sooner. Stable formulas and methods can return after more days.
Real mistakes get priority
Questions that already cost marks are imported into the correct Topic and scheduled for another attempt.
Seven days remove false familiarity
Wrong questions wait before the second review, so recently seen options and lucky guesses are less useful.
03
Learning can happen before scheduling.
This step is optional. A student can use Rankri only for revision, or first use the learning screen. Questions marked Already Known do not add unnecessary work. Questions that are new or weak enter the scheduler.
OPTIONALLearn firstMeaning, rule, explanation and examples.
DECIDEAlready KnownSkip scheduling and reduce revision load.
DECIDEAdd to revisionRankri begins tracking that Question.
REVIEWOne to five starsTell Rankri how well you remembered.
RESULTNext returnThe timing changes after every review.
04
The application layer around the scheduler.
A strong scheduling formula is only one part of Rankri. The product also needs reliable offline storage, synchronization, content pipelines and analysis that connects back to revision.
React desktop application
The browser application gives students a large workspace for Topics, quizzes, mock analysis and planning.
React Native with Expo
The mobile application supports daily study and offline use on the device students carry everywhere.
Persistent encrypted storage
Study can continue without a connection. Local progress is kept safely until synchronization is available.
Supabase and PostgreSQL
Authentication, server data, RPCs, storage and cross-device progress are managed through the backend.
Question-level merge rules
Review history, unlock state and other progress types follow explicit merge rules so stronger study progress is not replaced by weaker state.
Exam content and mock imports
Raw questions and mock results are converted into structured data that Rankri can schedule, analyse and show across devices.
05
What comes from research, and what Rankri adds.
Spaced repetition and FSRS are established areas of research. Rankri does not claim to invent them. Rankri's work is in how that research is applied to government-exam subjects, learning workflows, mock mistakes, quiz retention and the complete study experience.
Modern FSRS scheduling
Uses review history and memory ratings to estimate when a Question should return.
Subject-aware targets
Vocabulary, GK, Grammar, Maths and mock mistakes do not all receive identical timing.
Everything feeds revision
Learning, quizzes, mock imports and analysis connect to the next study action instead of remaining separate tools.
Rankri's supplied development document reports internal testing with more than 10,000 algorithmic flashcards and a validation cohort from the 2025 SSC cycle. Public evidence links can be added when available.