Convert Word Document to Quiz Questions
using the headings already in your .docx.

Convert Word document to quiz questions one heading at a time

A .docx already knows its own structure. When someone applied Heading 1 to a chapter title and Heading 2 to the parts under it, Word stored that as a property of the paragraph, which is how the navigation pane and the automatic contents page work. Flattening that into one long block of text on the way in is the normal way to do this, and it is what every other input page here does.

Keeping it changes what you can ask for. The outline that appears after you drop the file is the document's own, and each row is a checkbox. Tick the two sections you are behind on and the questions come from those two, because the text sent for generation is those two sections and nothing else. Untick the appendix and the reading list and they are not in the material at all, rather than being in it and hopefully ignored.

That distinction matters more than it sounds. A generator asked to write ten questions from a forty page handout will spread them across the whole thing, and roughly half will land on material you already know. Scoping the input is the only reliable way to aim them, and a Word document is the one format that tells you where its own seams are.

None of it requires an account, and the file itself never moves. The unzipping and the XML reading happen in the tab you have open, which is also why the outline can appear before you have committed to anything.

What inside a .docx becomes a question, and what quietly does not

Body paragraphs are the material. A sentence that states a subject, a condition and an outcome is the shape a multiple choice question gets built from, which is why a handout written in prose converts well and a page of bullet fragments does not. This is the same limit the PDF version of this tool runs into, and it has nothing to do with the file format on either side.

Tables, text boxes and the parts people forget are in there

Table cells come through as their own short lines, one per cell, which is honest but rarely useful: a two column table of terms and definitions reads as a list of disconnected fragments once flattened, and questions built from it tend to be trivia. Text boxes and callouts are read too, and they are attached to the paragraph they were anchored to, which is usually where they belong. Headers, footers and page numbers are in other parts of the file and are left alone, so the phrase running across the top of every page does not turn up forty times in your material.

Documents exported from Google Docs

Downloading a Google Doc as .docx gives you a real .docx, and its heading styles survive the trip, so the outline works the same way. What does not survive is anything that only existed in Google Docs, such as suggestions that were never accepted. If your document lives in Google Docs, File, Download, Microsoft Word is the whole preparation step.

Convert Word document to quiz: five files, five results

The shape of the document decides the result before the generator reads a word. These are five document shapes people actually drop in, with the kind of return each one gives, and two of them come back thin on purpose. Filter by whether heading styles were used, or sort by how many questions each returns, and open a card to see the shape of a question it produces.

Tracked changes, comments, and the noise a Word file carries

This is the failure people hit when they copy out of Word by hand. Select all in a document with markup showing, paste it somewhere, and you can end up with both the old wording and the new one in the same sentence, plus the margin comments as stray text. Feed that to any generator and you get a question whose correct answer is a number the author deleted a week ago.

Reading the file directly avoids the whole problem, because the two things live in different places inside it. Struck out text is stored in its own element, distinct from ordinary text, and only ordinary text is collected. Comments are not in the document part at all, they sit in a separate entry inside the zip, and that entry is never opened. What the generator receives is the document as it reads with every change accepted, which is what a marker would read.

When a document does carry either, the picker says so above the button rather than staying silent about it. If the open revisions were the point, and you wanted questions on the draft rather than the current version, this is the wrong tool for that specific job and the note is there so you find out before generating.

Word document to Canvas, Moodle or Quizlet: what the export does

Plenty of people looking for a Word to quiz converter are really after an LMS import. Canvas imports question banks as QTI packages, and Blackboard exports pools as QTI 2.1, but nothing here writes QTI. Moodle is a different case worth knowing about: its own importers read Moodle XML, GIFT and Aiken, and the last two are plain text formats rather than packages. The exports here are an Anki text file and a CSV, both produced after generation, both free.

A CSV is a reasonable middle step. It opens in Excel or Sheets, Quizlet's import takes pasted text with a delimiter you choose, and Moodle's Aiken format is close enough to a two column list that a spreadsheet can produce it. What it is not is one click, and if one click into Canvas is what you need, a dedicated QTI converter will serve you better than this will. The Anki side of the export is the path that needs no reshaping at all, because the format matches what Anki expects.

Where this stops

Only .docx, so a .doc has to be resaved first and a password protected document cannot be opened at all. No images and no OCR: if the content of your handout is a photograph of a page, there is no text in the file to read, and a diagram's caption comes through while the diagram does not. The free tier reads 25,000 characters of whatever you selected, so a very long document either needs fewer sections ticked or gets trimmed with a visible note. An equation written in Word's equation editor is stored in different markup from ordinary text, so it is dropped rather than mangled: the sentence around it comes through and the formula does not.

What is left, reading a Word document accurately, keeping its outline, and letting you aim the questions at one part of it, is the part worth getting right.

Next: once the questions are on screen, turn the same selection into a study guide of the same sections, or send them to a deck you can review tomorrow.