How accurate is the RevisionDojo EE grader?

There is not yet a robust public dataset that establishes an exact accuracy rate against official final IB EE marks. The grader can still be useful because it applies a selected rubric and provides criterion-level explanations. Its score is most credible when the draft is complete, the correct assessment model is selected, and each judgement cites visible evidence. Accuracy is likely to vary by subject, document quality, rubric version, and the complexity of the argument. Treat the displayed grade as an estimate and the detailed feedback as a set of claims to verify.

Can I trust a predicted A from the RevisionDojo EE grader?

A predicted A indicates that the tool found features associated with the highest markbands, but it does not guarantee an official A. Check how close the estimated raw mark is to the assumed boundary because a borderline prediction carries more risk. Read every criterion explanation and confirm that the cited strengths genuinely appear throughout the essay rather than in isolated passages. Ask your supervisor about disputed subject-specific judgements, particularly those involving method, disciplinary terminology, and evaluation. Plan revisions around defensible weaknesses instead of treating the predicted letter as secured.

Why did RevisionDojo and my teacher give different EE grades?

The grader and your teacher may interpret qualitative descriptors differently or may be assessing different versions of the essay. Your teacher also knows your subject context, research process, and school guidance, while an automated system primarily evaluates what it can extract from the uploaded document. Differences can arise from missing reflections, unreadable figures, an incorrect rubric selection, or borderline best-fit decisions. Ask both sources to justify their criterion scores using exact passages and descriptor language. Where the disagreement concerns subject conventions, your supervisor's contextual expertise should carry more weight than an unexplained automated score.

Should I upload a draft or only my final Extended Essay?

A substantial draft is usually more useful than an outline because the grader needs visible argument, evidence, analysis, evaluation, and structure to assess the criteria. An early upload can still identify broad problems with the research question or method, but its predicted grade will be especially uncertain. Later, upload a near-final version containing all material the tool is expected to evaluate, including readable figures and relevant reflections where supported. Keep the document version clearly labelled so you do not compare feedback from different drafts as if it assessed identical work. Use repeated uploads to monitor whether specific revisions improve the rubric evidence, not merely whether the headline score rises.

Does RevisionDojo use the new 2027 Extended Essay criteria?

RevisionDojo provides a dedicated grader for the Extended Essay model used from first assessment in 2027. That model has five redesigned criteria totalling 30 marks, whereas the preceding model totals 34 marks. Students must select the framework that matches their examination session rather than the year in which they began preliminary research. Confirm your session and pathway with your EE coordinator before interpreting the result. Scores produced under the two models are not directly interchangeable.

Is using an AI EE grader allowed by the IB?

The IB does not impose a blanket ban on AI tools, but all submitted work must remain the student's own. Using a grader to receive feedback is different from copying AI-generated analysis, evidence, or rewritten paragraphs into the essay. Students must critically evaluate AI output and follow their school's rules concerning disclosure, citation, and permitted assistance. Any AI-generated material incorporated into assessed work must be handled transparently under applicable IB and school guidance. The safest approach is to use the grader for diagnosis, then conduct the reasoning and rewriting yourself.

Back to all posts

RevisionDojo EE Grader Review: How Accurate Is It?

Reviews Site Grader Reviews IB IB Extended Essay

• 19 Sept 2026•14

RevisionDojo EE Grader Review: How Accurate Is It?

Join 650k+ students breezing through the IBReady to breeze through the IB?

RevisionDojo EE Grader Review: How Accurate Is It? | RevisionDojo

Back to all posts

RevisionDojo EE Grader Review: How Accurate Is It?

Reviews Site Grader Reviews IB IB Extended Essay

• 19 Sept 2026•14

This RevisionDojo EE grader review reaches a qualified conclusion: the grader can provide useful rubric-based feedback, but its predicted score should not be treated as a reliable forecast of your official Extended Essay grade. As of 17 September 2026, no robust public dataset matches RevisionDojo predictions with final IB EE marks across a clearly defined sample and date range.

That absence matters. Rubric alignment and consistent scoring can make an AI grader useful for revision, but neither proves that its predictions match independent IB examiners. Students should therefore trust the grader most as a diagnostic tool that identifies possible weaknesses, not as confirmation that an A, B, or C is secured.

Is there a verified predicted-vs-final EE dataset?

No sufficiently detailed public validation dataset could be verified for this review. RevisionDojo publishes information about its IB Coursework Grader, examples of grader output, student testimonials, and broader survey findings, but these do not provide the information needed to calculate RevisionDojo EE grading accuracy against official results.

A defensible validation study would need to publish:

The grader's predicted raw mark and letter grade before results were released
The official IB raw mark and final EE grade for the same submitted essay
The number of matched essays
The examination sessions and date range covered
The EE subject or pathway for each essay
The rubric version and grader version used
The proportion of exact matches and predictions within one grade
Average raw-mark error and any tendency to overpredict or underpredict

The current State of Learning Survey reports self-reported changes in coursework performance and platform use. It does not publish a matched dataset of grader predictions against official final EE results. Those findings should not be converted into an accuracy percentage because they measure a different outcome.

Consequently, there is no verified sample size or date range to disclose for a predicted-vs-final EE validation study. Claims such as “the grader is 90% accurate” or “usually within two marks” would be unsupported unless accompanied by auditable matched data.

What the RevisionDojo EE grader actually does

The grader is part of RevisionDojo's coursework feedback system for EEs, IAs, and TOK work. Students upload a draft, select the relevant assessment framework, and receive criterion-level marks, annotations, explanations, and suggested priorities for revision.

RevisionDojo states that the tool uses subject-specific IB rubrics and three-shot averaging, meaning multiple grading passes contribute to the displayed result. This may improve repeatability by reducing the effect of one unusually generous or severe output. It does not, by itself, demonstrate agreement with official IB examiners.

The distinction is important:

Type of performance	What it asks	What is currently supported?
Rubric alignment	Does feedback refer to the relevant criteria?	Supported by the product's criterion-based design
Repeatability	Does the same essay receive broadly similar results repeatedly?	Three-shot averaging is intended to improve this
Diagnostic usefulness	Does the feedback identify issues worth checking?	Plausible and supported by student reviews, but case-dependent
Predictive validity	Does the score match the official final EE result?	No robust public matched dataset was found

A grader can perform well in the first three areas while remaining imperfect at the fourth. For most students, its strongest function is showing where an argument becomes descriptive, where evidence is insufficiently evaluated, or where the method does not clearly answer the research question.

Why an EE prediction can differ from the final grade

The IB explains that Extended Essays are externally assessed by appointed examiners. An automated estimate is therefore being compared with a later human judgement made within the IB's assessment and standardisation process.

The criteria require qualitative judgement

Terms such as “effective,” “coherent,” “critical,” and “evaluative” cannot be measured by counting keywords. An examiner considers how well the essay works as a whole and uses a best-fit judgement within each criterion.

Two readers may agree that an essay contains evaluation while disagreeing about its depth or consistency. This is especially likely near the boundary between adjacent mark levels.

Subject conventions affect the meaning of quality

A History EE requires methods and source evaluation appropriate to historical inquiry. A Mathematics EE may depend on the student's understanding of mathematical reasoning, while an English A EE requires sustained literary or linguistic analysis.

An AI system may recognise general academic qualities but still misread specialised methods, notation, visual evidence, experimental design, or an unconventional yet valid argument. Selecting the wrong subject or rubric makes that risk substantially greater.

The uploaded draft may not be the assessed submission

A predicted-vs-final comparison is meaningful only if the grader assessed the exact version submitted to the IB. If the student subsequently changes the research question, analysis, conclusion, citations, or reflections, the two marks do not describe the same work.

The quality of the file also matters. Missing figures, poorly rendered equations, incomplete appendices, or an omitted reflection document can prevent the grader from seeing evidence that an examiner later receives.

Grade boundaries add another layer of uncertainty

A raw-mark prediction and a letter-grade prediction are not identical claims. The IB sets boundaries through its assessment process, so a one-mark raw-score error may change the letter grade for an essay near a boundary while a larger error may leave another essay in the same band.

For that reason, “predicted B, final B” does not necessarily mean highly precise marking. Likewise, a one-grade difference does not reveal how many raw marks separated the two judgements.

Check that you are using the correct EE rubric

The EE assessment model changes for first assessment in 2027. The official IB Extended Essay update describes five redesigned criteria totalling 30 marks. Earlier sessions use the five-criterion model totalling 34 marks.

Assessment model	Total	Main criterion structure
Up to the 2026 assessment cycle	34 marks	Focus and method; knowledge and understanding; critical thinking; presentation; engagement
First assessment 2027	30 marks	Framework for the essay; knowledge and understanding; analysis and line of argument; discussion and evaluation; reflection

RevisionDojo provides a dedicated 2027 EE grader and explains the 2027 assessment criteria. Before interpreting any score, confirm your examination session with your coordinator and check that the selected grader uses the applicable model.

This is not a minor technicality. A score out of 34 cannot be compared directly with a score out of 30, and the revised model redistributes emphasis across the criteria. Students should also avoid using unofficial fixed grade boundaries as guarantees for a future session.

What public RevisionDojo EE grader reviews show

Public comments are mixed. Some students report that coursework predictions were close to their later marks, while others describe overprediction, inconsistent comments, or substantial differences from final results. Examples appear on Trustpilot's RevisionDojo review page and in student discussions about RevisionDojo coursework grading and EE grader accuracy.

These accounts are useful for identifying possible strengths and failure modes, but they are not a representative accuracy study. Reviewers self-select, may use different product versions, and do not always provide the exact draft, predicted raw mark, final raw mark, subject, or session. Some comments also combine IAs, TOK work, and EEs even though those tasks involve different criteria and assessment processes.

The responsible conclusion is not that positive reviews prove accuracy or that negative reviews prove uselessness. They show that results vary and reinforce the need to inspect the feedback rather than accepting the displayed grade automatically.

How to test the feedback on your own essay

You cannot validate a final prediction before receiving your official result, but you can test whether the report is well founded.

Audit every significant deduction

For each low criterion score, ask:

What exact descriptor is being applied?
Which passage in the essay supports the judgement?
What evidence does the grader say is absent?
Does the proposed change fit the conventions of my EE subject?
Would the change improve reasoning, or merely add rubric vocabulary?

A useful comment should survive this audit. “Needs more critical thinking” is too vague; identifying an unsupported inference and explaining how to evaluate the evidence is actionable.

Compare the diagnosis, not just the number

Run the grader on a stable draft and record the criterion-level reasoning. Compare it with your own rubric annotation and any permitted supervisor feedback. Agreement about a weakness is more informative than agreement on an overall letter.

If the grader and supervisor disagree, do not automatically follow either one. Locate the relevant descriptor, ask what evidence each judgement relies upon, and let your supervisor clarify subject-specific expectations within the limits of permitted EE support.

Use a score range

Treat a result such as 23 marks as an uncertain estimate, not a measurement precise to one mark. A cautious interpretation might be that the work appears to sit within a nearby range, with greater uncertainty where the feedback depends on subtle evaluation or disciplinary expertise.

The range should be wider when the draft is incomplete, the essay sits close to a grade boundary, or the tool cannot read important evidence. It can be narrower only when several independent reviews broadly agree and clearly justify their judgements.

Revise one criterion at a time

The most effective workflow is diagnostic:

Upload a complete, readable draft using the correct rubric.
Identify one or two high-impact weaknesses.
Verify each criticism against the official descriptors.
Rewrite the relevant section in your own words.
Ask Jojo AI to explain unclear comments rather than write the submission.
Regrade the revised draft and check whether the reasoning changed.
Discuss major changes in question, method, or argument with your supervisor.

RevisionDojo's EE feedback guide provides a similar criterion-led approach. Its broader guide to RevisionDojo Extended Essay support also explains how the grader fits alongside exemplars, planning resources, and supervisor guidance.

Academic integrity when using an AI grader

The IB does not prohibit AI tools outright, but students remain responsible for submitting their own work. Its guidance on AI in learning and assessment says students must critically review AI output and transparently acknowledge AI-generated material used in assessed work.

Using feedback to identify an unclear argument is different from copying a generated replacement paragraph. The safest practice is to interpret the comment, return to your evidence, and rewrite independently. Follow your school's AI policy because it may impose more specific disclosure or usage requirements.

Do not let a grader invent citations, evidence, calculations, quotations, or reflections. These are especially serious risks because apparently polished output can still be false, inappropriate to the subject, or inconsistent with your actual research process.

Final verdict

The RevisionDojo EE grader is best treated as a rubric-aligned second reader, not a substitute examiner. Its criterion breakdowns and annotations can help students find weaknesses earlier, but there is currently no robust public predicted-vs-final EE dataset establishing an exact accuracy rate.

Trust comments that identify specific evidence and align with the correct rubric. Be cautious with the overall predicted grade, especially near boundaries or when the judgement depends on subtle subject expertise. For a balanced revision process, use RevisionDojo's Coursework Grader and Jojo AI alongside the official criteria, your own critical judgement, and your school-appointed supervisor.

Sources and referenced URLs

Add as a preferred source on Google

Get your coursework graded against the IB criteria

Upload an IA, EE, or TOK draft and get criterion-by-criterion feedback in minutes.

Grade my coursework Browse coursework examples