Students and parents examining RevisionDojo student results should know the central limitation immediately: as of September 2026, RevisionDojo has not published a robust dataset matching its students' predicted grades or platform estimates with their official final IB grades. Available outcomes data shows self-reported changes in predicted grades during the May 2026 cycle, not predicted versus final IB grade accuracy.
An increase in a school prediction may be encouraging, but it does not prove that the prediction was accurate, that the final result increased, or that RevisionDojo caused the change. This article explains what the evidence supports, what remains unknown, and how final-grade accuracy should be validated.
What evidence is currently available?
The current RevisionDojo 2026 State of Learning Survey reports results from 364 May 2026 students across 258 schools, including 215 complete and 149 partial responses. It is a company-run observational survey based on students' self-reported grades and experiences.
The most relevant findings are:
-
Across 1,168 matched subject ratings, mean predicted grades increased from 4.81 to 5.58.
-
Among 253 students with at least one matched subject rating, 211, or 83.4%, reported a higher average predicted subject grade.
-
The mean increase among these students was 0.77 grade points.
-
Among 105 students reporting total diploma predictions at both check-ins, the mean increased from 31.10 to 34.95 points out of 45.
-
Sciences recorded the largest reported subject-group increase, at 1.01 grade points.
The reporting groups subjects into broad areas, but does not publish a complete subject-by-subject breakdown for all matched ratings. The public results page also does not clearly disclose the precise fieldwork dates for both check-ins.
These are genuine published survey results, but they are not a predicted-versus-final dataset. They cannot establish final scores, university offer attainment, or the accuracy of RevisionDojo's estimates.
Predicted-grade improvement is not prediction accuracy
The IB defines a predicted grade as a teacher's estimate of the grade a candidate is expected to achieve, based on evidence of the student's work and knowledge of IB standards. The IB's assessment guidance asks schools to predict accurately without systematic overprediction or underprediction.
A final IB grade is awarded after assessed components have been marked, weighted, moderated where applicable, and converted through the relevant grade boundaries. The IB assessment overview confirms that Diploma Programme subjects receive grades from 1-7, while TOK and the Extended Essay can contribute up to 3 core points.
Consider a student whose prediction rises from 5 to 6:
-
A final grade of 6 makes the later prediction exactly accurate.
-
A final grade of 5 means it overpredicted by one grade.
-
A final grade of 7 means it underpredicted by one grade.
The reported improvement is identical in every case, but the accuracy conclusion changes. Prediction growth therefore cannot answer a question about final-grade accuracy.
What the current results can and cannot support
The survey supports the narrow conclusion that many participants reported improved predicted grades while using RevisionDojo. It also indicates that improvements appeared across several broad subject groups.
However, the research design has important limitations:
-
Grades were self-reported, not independently imported from official results documents.
-
Students chose whether to respond, creating possible selection and response bias.
-
There was no randomized comparison group of similar non-users.
-
Teaching, tutoring, mock examinations, coursework and independent revision were not controlled.
-
Later predicted grades were not publicly matched to official final grades.
What a robust predicted-versus-final dataset should contain
A credible study of predicted vs final IB grade accuracy would register predictions before final results became available. Collecting estimates after results day would introduce a serious risk of hindsight bias.
With informed consent, researchers should match frozen estimates to verified final result records, remove identifying information before analysis, and follow safeguards such as those described in RevisionDojo's student privacy guidance.
The published dataset should disclose:
-
The number of students and subject-grade pairs.
-
Examination sessions and data-collection dates.
-
Exact subjects, levels and syllabus versions.
-
The source and timestamp of every prediction.
-
The method used to verify final grades.
-
Inclusion criteria, exclusions and missing-data rates.
-
Platform usage intensity.
-
Any independent analysis or audit.
Results should cover the full eligible sample, not only students who improved. Researchers should also examine whether students who withheld final results differed from those who completed the study.
Which accuracy metrics should be published?
No single statistic gives a complete account of prediction quality. A useful validation report should include:
-
Exact-match rate: the percentage whose predicted and final grades are identical.
-
Within-one-grade rate: the percentage whose prediction is no more than one grade above or below the result.
-
Mean absolute error: the average size of prediction errors, regardless of direction.
-
Mean signed error: whether predictions systematically run high or low.
-
Subject-level accuracy: separate findings for each subject and level with an adequate sample.
-
Diploma-total accuracy: differences between estimated and official totals out of 45.
-
Calibration: how often students assigned each estimated grade actually receive it.
Results should include confidence intervals, particularly for smaller subjects. A headline accuracy percentage without uncertainty or sample breakdowns can appear more precise than the evidence permits.
How should students and parents evaluate RevisionDojo now?
Until matched final-grade evidence is published, evaluate RevisionDojo as a learning and practice platform rather than a guaranteed score predictor. The practical question is whether it helps the student produce stronger independent work under realistic conditions.
The RevisionDojo Grade Calculator can translate entered practice marks into an indicative grade using selected boundaries. Its output depends on the quality of those inputs and is not an official school prediction.
A sensible evaluation process is:
-
Complete a timed baseline paper or substantial section independently.
-
Identify recurring content, method and timing errors.
-
Use RevisionDojo's study features, including Study Notes, Flashcards, Questionbank practice and Jojo AI, to target those weaknesses.
-
Complete a comparable timed assessment without assistance.
-
Compare raw marks, error patterns and consistency, not only estimated grades.
The RevisionDojo mock-exam workflow is most useful when conditions resemble the real examination. A stable pattern across several timed papers is stronger evidence than one unusually high mark.
Common mistakes when interpreting outcome claims
Students and parents should remember that predicted-grade growth is not final-grade growth. Correlation between platform use and improvement also does not demonstrate causation, especially when other academic support is not controlled.
Aggregate findings cannot determine what will happen in a particular subject or level. RevisionDojo similarly advises readers to treat predictive outputs as guidance in its accuracy review, rather than as official IB decisions.
Conclusion
Current RevisionDojo student results provide encouraging self-reported evidence of higher predicted grades among participating May 2026 students. They do not provide a verified comparison between frozen predictions and official final IB results, so no defensible final-grade accuracy rate can yet be stated.
A future matched study needs preregistered predictions, verified results, transparent samples, missing-data reporting and independent analysis. For now, families should judge RevisionDojo through changes in independent timed performance, using the Grade Calculator cautiously and combining Questionbank practice, Mock Exams and Jojo AI with teacher feedback.
