Creating A Quiz In Power BI

This blog lays out the core patterns - data model, slicers, measures, and UX - so anyone can adapt the idea to their dataset.

Core Concept

Pattern: keep your quiz logic in disconnected tables (no relationships), then use measures to pull the “current question” and evaluate answers.

Why:

• Disconnected tables are immune to other page slicers; your quiz won’t break if the report has filters elsewhere.
• Measures read from your question bank deliberately, so behaviour is predictable.
• Easy to swap question types (year, trivia, anything) without rebuilding relationships.

Data Model: Minimal, Flexible, Reusable

1) Question Bank (calculated/imported table)

One row = one question. Keep it simple and generic:

Columns:

QuestionID (number) used for Stable ID indexing

QuestionText (text) is the prompt shown to the user

QuestionType (text) is the multiple choice questions / truefalse / shortext logic

CorrectAnswer (text/number) is what counts as correct

2) Quiz Controls (disconnected)

• Question Picker: either a What-if parameter table (1..N) or a simple GENERATESERIES(1,N,1) table for a slider/slicer.
• Reveal Toggle: tiny manual table with values Hide / Reveal.
• User Choice (for MCQ/TrueFalse): a small table the user selects from - e.g., AnswerChoices[Answer] with {A,B,C,D} or {TRUE,FALSE}.

No relationships needed. You’ll filter the question bank via measures.

UI Flow You Can Reuse Anywhere

1. Pick a question (slider or “Next” buttons).
2. Show the prompt (optionally masked).
3. Let the user choose an answer (buttons/tiles).
4. Reveal (and score).

This works for: guessing a year, picking a prime, choosing a definition - anything.

Generic Measures (Templates with Placeholders)

Replace table/column names in 'Questions'[], 'Picker'[], 'Reveal'[], 'AnswerChoices'[] with yours.

1) Read the selected QuestionID

2) Pull the current question and answer

Why use FILTER + CALCULATE? It clearly targets one row by QuestionID, and adapts to any question bank shape.

3) Masking (optional “reveal later” pattern)

Mask any sensitive token (e.g., a year, a keyword) only when hidden.

Why mask? Makes a “guess first, reveal later” flow without bookmarks.

4) Reveal switch

5) User’s selected answer (for MCQ / True/False)

6) Correctness & feedback

Why store correctness as a measure? It recalculates instantly with slicer changes; no state to manage.

Multiple-Choice Layout (Generic)

• Slicer (tiles) → AnswerChoices[Answer] (A/B/C/D or TRUE/FALSE).
• Card (prompt) → QuizPromptToShow.
• Buttons → Reveal[Mode] (Hide/Reveal).
• Card (feedback) → FeedbackText.
• (Optional) Options table visual if your options come from a child table:
  • Add a table visual with options filtered by QuestionID = [CurrentQuestionID].

Random vs Deterministic

• Deterministic (recommended for demos): slider/parameter picks QuestionID. Clean and repeatable.
• Random: RANDBETWEEN(1,N) for CurrentQuestionID. Fun, but it re-rolls whenever the page recalculates.

Scoring (Session-style)

Power BI doesn’t persist user state. A lightweight pattern is to score current selection only:

If you want a short quiz of K questions, create a small IndexList table (1..K), and store a per-index choice in a disconnected table, then sum correctness across indices. (That’s more advanced; the key rule still holds: keep it disconnected and compute with measures.)

Accessibility & UX Tips

• Tiles for slicers (bigger hit targets).
• Edit interactions so other report slicers don’t interfere with the quiz page.
• Help panel with bookmarks (Data = Off) for a clean show/hide “How to use”.
• Readable fonts (Segoe UI / Calibri), high-contrast colours, and enable word wrap on cards.
• Instructions at the top: “1) Pick a question. 2) Choose an answer. 3) Reveal.”

Quick Build Checklist

1. Create a QuestionBank (one row per question, with QuestionID, QuestionText, CorrectAnswer).
2. Add a Picker table (1..N) for the slider; no relationships.
3. Add Reveal and (if needed) AnswerChoices tables; no relationships.
4. Write measures to:

• read current QuestionID,
• fetch prompt and correct answer,
• mask text (optional),
• evaluate correctness,
• show feedback.

5. Build the page (tiles, card, reveal toggle).
6. Test edge cases (no selection, wrong selection, masking targets).