Product Design: Fit & Style Feedback Loop #6

New issue

Closed

opened 2026-05-02 18:46:12 +00:00 by agent-cos · 1 comment

agent-cos commented

2026-05-02 18:46:12 +00:00

Objective

Design the fit feedback loop that continuously improves the AI's understanding of each customer's body and preferences. After every box, the customer provides specific feedback on each item.

Why This Matters

The #1 complaint about Stitch Fix and Wantable is that the AI never learns. A tight feedback loop with structured data collection is our differentiator.

Feedback Per Item (Required)

For each item received, customer must provide:

Keep or Return (binary)
Fit Rating (1-5)
Style Rating (1-5)
Color Rating (1-5)
Fabric/Quality Rating (1-5)
Free-form comment: What worked or didn't work?

Feedback Per Box (Optional)

Overall satisfaction (1-5)
Would you wear these together? (Y/N/Maybe)
Anything you wish we'd sent instead? (free form)
Any life changes? (new job, weight change, season change)

AI Learning System

Short-term (per customer, per item)

Track fit accuracy by brand, size, and cut type
Learn color, silhouette, and pattern preferences
Learn price sensitivity

Long-term (cross-customer)

Aggregate fit data by brand → "brand fit profiles"
Identify style clusters
Train curation model on aggregate keep rates
Predict preferences based on similar profiles

Feedback UX

Mobile-first micro-feedback: swipe left/right like Tinder
1-tap emoji ratings for speed
Deep feedback optional for power users
Gamification: feedback completion = $5 credit toward next box

Data Schema (Initial)

ItemFeedback {
  itemId: string; customerId: string; boxId: string;
  action: 'kept' | 'returned';
  fitRating: 1-5; styleRating: 1-5; colorRating: 1-5; qualityRating: 1-5;
  comment: string;
  reasons?: string[]; // 'too-small', 'not-my-color', 'love-it', etc.
  createdAt: timestamp;
}

Deliverable

Design doc at product/feedback-system.md — per-item flow, AI learning model, aggregation strategy, data schema.

## Objective Design the fit feedback loop that continuously improves the AI's understanding of each customer's body and preferences. After every box, the customer provides specific feedback on each item. ## Why This Matters The #1 complaint about Stitch Fix and Wantable is that the AI never learns. A tight feedback loop with structured data collection is our differentiator. ## Feedback Per Item (Required) For each item received, customer must provide: 1. **Keep or Return** (binary) 2. **Fit Rating** (1-5) 3. **Style Rating** (1-5) 4. **Color Rating** (1-5) 5. **Fabric/Quality Rating** (1-5) 6. **Free-form comment**: What worked or didn't work? ## Feedback Per Box (Optional) 7. **Overall satisfaction** (1-5) 8. **Would you wear these together?** (Y/N/Maybe) 9. **Anything you wish we'd sent instead?** (free form) 10. **Any life changes?** (new job, weight change, season change) ## AI Learning System ### Short-term (per customer, per item) - Track fit accuracy by brand, size, and cut type - Learn color, silhouette, and pattern preferences - Learn price sensitivity ### Long-term (cross-customer) - Aggregate fit data by brand → "brand fit profiles" - Identify style clusters - Train curation model on aggregate keep rates - Predict preferences based on similar profiles ### Feedback UX - Mobile-first micro-feedback: swipe left/right like Tinder - 1-tap emoji ratings for speed - Deep feedback optional for power users - Gamification: feedback completion = $5 credit toward next box ## Data Schema (Initial) ``` ItemFeedback { itemId: string; customerId: string; boxId: string; action: 'kept' | 'returned'; fitRating: 1-5; styleRating: 1-5; colorRating: 1-5; qualityRating: 1-5; comment: string; reasons?: string[]; // 'too-small', 'not-my-color', 'love-it', etc. createdAt: timestamp; } ``` ## Deliverable Design doc at `product/feedback-system.md` — per-item flow, AI learning model, aggregation strategy, data schema.

agent-cos referenced this issue

2026-05-02 18:52:49 +00:00

Business: Phased Roadmap (Concept → Launch → Scale) #14

agent-cos referenced this issue

2026-05-02 19:35:17 +00:00

Project Management: Milestone-Based Issue Organization #28

sscoble referenced this issue

2026-05-03 08:11:42 +00:00

MVP: Personal Use Engine — Core Curation System #20

sscoble commented

2026-05-25 02:12:39 +00:00

Owner

✅ Design/planning ticket with substantial content. Contains complete fit feedback loop design with per-item ratings, AI learning system (short-term per customer, long-term cross-customer), feedback UX, and data schema. Keeping open as a feature.

✅ **Design/planning ticket with substantial content.** Contains complete fit feedback loop design with per-item ratings, AI learning system (short-term per customer, long-term cross-customer), feedback UX, and data schema. Keeping open as a feature.