The quality of your quiz depends directly on the quality of your source material. A few minutes of preparation before you upload can make the difference between vague, repetitive questions and sharp, accurate ones.
The AI extracts meaning from paragraphs, headings, and bullet points. Documents with logical structure — numbered sections, headers, short paragraphs — produce better questions. Avoid wall-of-text layouts and decorative columns that break the reading flow.
Cover pages, table of contents pages, bibliographies, legal boilerplate, and repeated headers/footers add noise. Trim them before uploading or use the page-range selector when uploading a PDF.
A PDF exported from Word or Google Docs contains selectable text and gives the highest accuracy. Scanned images of pages go through OCR, which can introduce errors. If you must use a scan, make sure it is high-resolution (300 dpi+) and well-lit.
Quiz Master processes documents in chunks. Very long files (100+ pages) may produce questions skewed toward the beginning. For best coverage, split the document into logical parts (chapters, topics) and create one quiz per part.
Requesting 40 questions from a 2-page document forces the AI to repeat itself. A good rule of thumb is 1–3 questions per substantial paragraph. Use the question-count slider to stay within that range.
If your document is in Hebrew, set the output language to Hebrew before generating. The AI will produce questions and answers in the selected language, even if the source document mixes languages.
"Easy" produces factual recall questions. "Hard" produces inference and analysis questions. Match the difficulty to your audience — a beginner-level training document shouldn't use hard inference questions.
Instead of guessing how many questions to request, enable the Auto-detect toggle in the generation form. Quiz Master will analyze your document's structure — paragraphs and tables — and calculate an optimal number of questions per section. A summary is shown for your review before generation begins, and credits are charged based on the detected total, not the slider value.
These three formatting elements are what the AI relies on most. Tables map directly to quiz questions — each row is typically a fact or comparison that becomes a natural question target. Lists (bulleted or numbered) produce clean, discrete questions with minimal ambiguity — each item stands on its own. Paragraphs provide the context and reasoning that enable deeper inference and analysis questions. A document that combines all three gives the AI richer material and produces a more varied, balanced quiz.