docs(seed): update content guide to v3 — product standard, readability test, final approval

This commit is contained in:
null 2026-06-25 18:56:48 -05:00
parent 4686a2c200
commit 164acf415d
1 changed files with 215 additions and 139 deletions

View File

@ -1,6 +1,6 @@
# Closer Question Writing Guide v2
# QUESTION_CONTENT_GUIDE.md
**See also:** [QUESTION_SCHEMA.md](QUESTION_SCHEMA.md) — JSON schema & validation rules | [QUESTION_REWRITE_PLAN.md](QUESTION_REWRITE_PLAN.md) — rewrite checklist & category order | [QUESTION_QUALITY_CHECKLIST.md](QUESTION_QUALITY_CHECKLIST.md) — quality gate before commit
# Closer Question Writing Guide v3
## Mission
@ -10,26 +10,29 @@ Closer is not therapy homework.
Closer is not a personality quiz.
Closer should feel like a conversation game couples voluntarily keep playing because they are having fun.
Closer should feel like a conversation game couples voluntarily keep
playing because they are having fun.
Every question should move the couple toward at least one of these:
* laughing
* flirting
* learning something new
* remembering something
* planning something together
* feeling understood
* feeling appreciated
* creating a future memory
- laughing
- flirting
- learning something new
- remembering something
- planning something together
- feeling understood
- feeling appreciated
- creating a future memory
If a question does none of those things, delete it.
## Consumer First
------------------------------------------------------------------------
# Consumer First
Every question must pass this test.
Would a real couple answer this on a Friday night?
Would a real couple willingly answer this on a Friday night?
If not, rewrite it.
@ -37,203 +40,276 @@ Never write for psychologists.
Write for normal couples.
## Question Mix
------------------------------------------------------------------------
# Product Standard
Closer should feel premium.
Every category should feel like it was hand written by an experienced
relationship coach and game designer, not generated by AI.
If a user can predict the next question, the category fails.
------------------------------------------------------------------------
# Question Mix
Per 250 questions:
* 140 multi_choice
* 50 single_choice
* 35 scale
* 15 this_or_that
* 10 written
- 140 multi_choice
- 50 single_choice
- 35 scale
- 15 this_or_that
- 10 written
At least 76 percent must be choice based.
Typing should feel special.
Typing should feel rare and meaningful.
Never force users to type constantly.
------------------------------------------------------------------------
## Emotional Mix
# Emotional Mix
Every category should roughly include:
Each category should roughly contain:
* 35 percent playful
* 25 percent everyday life
* 20 percent meaningful
* 10 percent future
* 10 percent deeper
- 35 percent playful
- 25 percent everyday relationship
- 20 percent meaningful
- 10 percent future focused
- 10 percent deeper vulnerability
Alternate emotional weight naturally.
Never stack several heavy questions together.
Do not stack heavy questions.
Alternate emotional intensity naturally.
## Conversation Goals
------------------------------------------------------------------------
Every question should create at least one reaction:
# Conversation Goals
* I did not know that.
* That is adorable.
* We should do that.
* I cannot believe you picked that.
* I never thought about that.
* That is actually a good idea.
Every question should create at least one of these reactions:
If none fit, rewrite it.
- "I didn't know that."
- "That's adorable."
- "We should actually do that."
- "I can't believe you picked that."
- "I've never thought about that."
- "That's a really good point."
## Variety
If none apply, rewrite it.
Avoid repeating openings.
------------------------------------------------------------------------
Use varied openings such as:
# Variety
* Imagine...
* Suppose...
* If we suddenly...
* Pick every answer...
* Finish this thought...
* Would you rather...
* When do you...
* What is one thing...
* Which tiny habit...
* If we had one free hour...
* What would make...
* What feels most like us...
Never repeat sentence openings.
Never allow obvious patterns.
Mix structures constantly.
## No AI Writing
Examples:
Reject wording like:
- Imagine...
- Suppose...
- If we suddenly...
- Pretend...
- Pick every answer...
- Finish this thought...
- What's one thing...
- Which tiny habit...
- If we only had one hour...
- What would make...
- When do you feel...
- Which sounds most like us...
* Describe...
* Reflect on...
* Discuss...
* Evaluate...
* In what ways...
* How satisfied are you...
* What boundary around...
* Explore your feelings about...
* Rate the effectiveness of...
* Identify the ways...
No opening should dominate a category.
------------------------------------------------------------------------
# No AI Writing
Reject questions that sound like:
- Describe...
- Reflect on...
- Discuss...
- Evaluate...
- In what ways...
- How satisfied are you...
- What boundary around...
- Explore your feelings...
- Identify...
- Rate the effectiveness...
These sound like surveys.
Not conversations.
## Written Questions
------------------------------------------------------------------------
Written questions must earn the keyboard.
# Multiple Choice
Use them only when choices cannot create the same value.
Should make people pause because several answers feel right.
Good:
* What's something tiny I do that means more than I realize?
Bad:
* Describe your communication style.
## Multiple Choice
Options should be interesting.
Use 4 to 6 options.
Avoid obvious answers.
Avoid yes or no disguised as choices.
Avoid filler.
Allow multiple selections whenever realistic.
Whenever realistic, allow multiple selections.
Use prompts like:
------------------------------------------------------------------------
* Pick every answer that fits.
* Choose your top three.
* What sounds good to you?
* Which of these feel true?
# Single Choice
## Single Choice
There should never be one obvious answer.
There should be no obvious correct answer.
Every option should feel believable.
The user should hesitate because several options feel plausible.
------------------------------------------------------------------------
## Scale
# Scale
Only use scales when intensity matters.
Use scales only when measuring intensity genuinely improves the
conversation.
Do not use scales just because they are easy to generate.
Never use a scale because it is easier than writing good options.
Good scale topics:
------------------------------------------------------------------------
* comfort
* confidence
* closeness
* interest
* frequency
* importance
# This Or That
## This Or That
Keep them:
Keep these:
- fast
- funny
- memorable
- shareable
* fast
* funny
* shareable
* easy
Most should take under three seconds.
They should take under three seconds.
------------------------------------------------------------------------
## Fun Rule
# Written Questions
Written questions must earn the keyboard.
Only use them when typing creates something choices cannot.
Good written questions create:
- stories
- memories
- appreciation
- future plans
Never ask users to write essays.
------------------------------------------------------------------------
# Fun Rule
Every category should include:
* inside jokes
* tiny challenges
* food debates
* traditions
* bucket lists
* silly scenarios
* friendly competitions
* future plans
* low pressure date ideas
- inside jokes
- tiny challenges
- food debates
- traditions
- bucket lists
- silly scenarios
- future adventures
- friendly competitions
- low pressure date ideas
- unexpected situations
Even serious categories need relief.
Even serious categories need moments to smile.
## Relationship First
------------------------------------------------------------------------
# Relationship First
Anchor every question to:
* you
* your partner
* your relationship
* your memories
* your future
- us
- we
- you
- your partner
- our memories
- our future
Avoid generic questions that could belong in a workplace survey.
Avoid generic questions that could belong in any survey.
## Duplicate Prevention
------------------------------------------------------------------------
# Duplicate Prevention
Never repeat:
* sentence structure
* concepts
* option lists
* conversation goals
- sentence openings
- concepts
- option lists
- conversation goals
- emotional outcome
If two questions feel similar, delete one.
## Premium Test
------------------------------------------------------------------------
Reject any question if the answer to any of these is no:
# Readability Test
* Would I pay for this?
* Would this start a real conversation?
* Would this create a memory?
* Would I screenshot this?
* Would I send this to my partner?
After completing a category:
## Final Rule
Randomly read 30 questions.
Reject the category if:
- the writing feels repetitive
- the openings repeat too often
- the tone feels robotic
- the next question becomes predictable
- multiple questions blend together
- it sounds AI generated
This test is mandatory.
------------------------------------------------------------------------
# Premium Test
Before approving every question ask:
- Would I pay for this?
- Would this start a real conversation?
- Would this create a memory?
- Would I screenshot this?
- Would I send this to my partner?
- Would I smile while answering it?
If any answer is no, rewrite it.
------------------------------------------------------------------------
# Final Approval
A category is complete only if:
- schema validation passes
- duplicate checks pass
- readability test passes
- category identity is obvious
- emotional mix feels balanced
- consumer test passes
- premium test passes
- it does not sound AI generated
------------------------------------------------------------------------
# Final Rule
Conversation quality always beats quantity.
One unforgettable question is worth more than twenty average ones.
Never optimize for finishing faster.
Optimize for creating conversations couples will remember.