How We Create Quizzes

Every Hitohira quiz is built through a deliberate process designed to ensure content quality, scoring fairness, and a consistent user experience. This page walks through each stage of that process.

1. Choosing a Topic

We select topics based on three criteria: genuine self-exploration value for the user, no involvement of officially licensed IP that would carry legal risk, and the ability to present the subject in a warm, non-diagnostic tone. We deliberately avoid frameworks that leave users feeling judged or reduced to a label.

2. Designing the Scoring Model

Before we write a single question, we define the scoring model:
- Dimensional: Scores accumulate along multiple bipolar axes (for example, Extraversion vs. Introversion). The final combination of axis outcomes determines the result code. Best suited for personality type quizzes.
- Tally: Each answer option casts votes directly toward one or more result candidates. The candidate with the most votes wins. Best suited for "which type of X are you?" formats.

Once the model is defined, we enumerate all possible result codes and set tie-break preferences for each axis, so the scoring engine always produces a clear result even when scores are perfectly balanced.

3. Writing Questions and Options

Question design principles:
- Each question focuses on a single clear dimension or scenario
- Option wording avoids telegraphing an obvious "correct" answer
- Each option's scoring weights are defined in advance
- Chinese and English versions of questions and options are written separately to ensure cultural appropriateness across languages, rather than mechanical translation

4. Writing Result Copy

Each result includes a title, subtitle, body copy (minimum 500 words in English, covering trait description, strengths, blind spots, relationships, work style, and self-reflection prompts), share text, and a disclaimer. All copy undergoes human editorial review before publication (see Editorial Policy).

5. Creating Original Artwork

Result card illustrations for each quiz are original works designed in line with the site's aesthetic. Artwork is stored as PNG files and used on result pages and social media share cards.

6. AI Assistance and Human Review

We use AI writing tools to assist with initial drafts, which speeds up content production. However, every AI-generated draft must pass human editorial review to confirm language quality, scoring accuracy, and content appropriateness. We do not publish AI content that has not been verified by a human editor.

7. Technical Architecture: Engine Once, Theme Many

Every quiz shares the same scoring engine but has its own theme configuration — colors, illustration style, and tone of voice. This "engine once, theme many" architecture lets us maintain multiple quizzes efficiently while ensuring that each has a distinct personality rather than feeling like a generic template.

8. Ongoing Updates

After a quiz is published, we track user feedback and content versions. When result copy or scoring logic sees significant improvements, we increment the version number and conduct a fresh review. Our goal is for every quiz to remain accurate and readable over the long term.