Feature: Sale/Price Tracking for WardrobeAI Recommendations #15

New issue

Closed

opened 2026-05-04 03:51:45 +00:00 by agent-engineering · 1 comment

agent-engineering commented

2026-05-04 03:51:45 +00:00

Summary

WardrobeAI should track sales, price changes, and availability across retailers so it can recommend items that are actually in stock, at the best price, and alert users when items go on sale.

Problem

While building the wardrobe cart feature, we discovered three critical gaps:

Stock unavailability: Items recommended by the AI were not available in the user's size on Amazon. We couldn't verify real-time stock, leading to multiple rounds of failed cart attempts.
No price tracking: Users miss sales (e.g., Lucky Brand's 30% off VIP30 code). The AI can't tell users "buy now, this is on sale" or "wait, this will likely drop."
No cross-retailer comparison: The same item may be cheaper on one site vs. another, or available in a size that's out of stock elsewhere. Currently we can't surface this.

This is a killer feature for the product — it transforms WardrobeAI from a styling tool into a shopping companion that saves users money and frustration.

Proposed Solution

Phase A: Real-Time Stock & Price Verification

Integrate with product APIs (Amazon Product Advertising API, or affiliate/partner APIs for Quince, Lucky Brand, etc.) to check:
- Availability: Is size X in color Y currently in stock?
- Price: What's the current price? Is it on sale?
- Shipping: Prime eligible? Free shipping threshold?
Cache results with short TTL (1-4 hours) to avoid API rate limits
Surface a "Verified ✓ In Stock" badge on recommendations vs. "⚠️ Check availability"

Phase B: Price History & Sale Alerts

Track price history per item+size+color over time
Detect sale events (e.g., "Lucky Brand 30% off everything this weekend")
Alert users: "This item just dropped 30% — buy now" or "This item rarely goes on sale — current price is typical"
Support promo code detection ("Use code VIP30 for 30% off")

Phase C: Cross-Retailer Comparison

For each recommended item, search across multiple retailers (Amazon, brand direct, Zappos, Nordstrom, etc.)
Show: "Available at 3 retailers: Lucky Brand $49 (30% off w/ VIP30), Amazon $55 (Prime), Zappos $79 (free returns)"
Link to the best available option for the user's size+color combo

Phase D: Size & Fit Intelligence

Build a size chart database per brand (we already started this manually for Quince vs Lucky Brand)
Auto-map user measurements to brand-specific sizes
Surface: "Quince XL = bust 42.5", Lucky Brand XL = waist 35-37" — based on your measurements, Lucky Brand XXL will fit best for tops"

Technical Notes

Amazon Product Advertising API requires an affiliate account (free to register)
Many DTC brands (Quince, Lucky Brand) have public product pages we can scrape or query
Consider using a price tracking service (e.g., Keepa, CamelCamelCamel) as a data source for Amazon price history
Rate limiting and caching are essential — product pages change slowly, no need to hit APIs on every recommendation

Acceptance Criteria

WardrobeAI can verify stock availability for a specific size+color before recommending
Current price and sale status are shown for each recommended item
Users see "Verified In Stock" or "Check Availability" labels
Size chart data is stored per brand and used to recommend the right size
At least 3 retailers are compared for each item (where available)

Labels

feature

## Summary WardrobeAI should track sales, price changes, and availability across retailers so it can recommend items that are actually in stock, at the best price, and alert users when items go on sale. ## Problem While building the wardrobe cart feature, we discovered three critical gaps: 1. **Stock unavailability**: Items recommended by the AI were not available in the user's size on Amazon. We couldn't verify real-time stock, leading to multiple rounds of failed cart attempts. 2. **No price tracking**: Users miss sales (e.g., Lucky Brand's 30% off VIP30 code). The AI can't tell users "buy now, this is on sale" or "wait, this will likely drop." 3. **No cross-retailer comparison**: The same item may be cheaper on one site vs. another, or available in a size that's out of stock elsewhere. Currently we can't surface this. This is a **killer feature** for the product — it transforms WardrobeAI from a styling tool into a shopping companion that saves users money and frustration. ## Proposed Solution ### Phase A: Real-Time Stock & Price Verification - Integrate with product APIs (Amazon Product Advertising API, or affiliate/partner APIs for Quince, Lucky Brand, etc.) to check: - **Availability**: Is size X in color Y currently in stock? - **Price**: What's the current price? Is it on sale? - **Shipping**: Prime eligible? Free shipping threshold? - Cache results with short TTL (1-4 hours) to avoid API rate limits - Surface a "Verified ✓ In Stock" badge on recommendations vs. "⚠️ Check availability" ### Phase B: Price History & Sale Alerts - Track price history per item+size+color over time - Detect sale events (e.g., "Lucky Brand 30% off everything this weekend") - Alert users: "This item just dropped 30% — buy now" or "This item rarely goes on sale — current price is typical" - Support promo code detection ("Use code VIP30 for 30% off") ### Phase C: Cross-Retailer Comparison - For each recommended item, search across multiple retailers (Amazon, brand direct, Zappos, Nordstrom, etc.) - Show: "Available at 3 retailers: Lucky Brand $49 (30% off w/ VIP30), Amazon $55 (Prime), Zappos $79 (free returns)" - Link to the best available option for the user's size+color combo ### Phase D: Size & Fit Intelligence - Build a size chart database per brand (we already started this manually for Quince vs Lucky Brand) - Auto-map user measurements to brand-specific sizes - Surface: "Quince XL = bust 42.5\", Lucky Brand XL = waist 35-37\" — based on your measurements, Lucky Brand XXL will fit best for tops" ## Technical Notes - Amazon Product Advertising API requires an affiliate account (free to register) - Many DTC brands (Quince, Lucky Brand) have public product pages we can scrape or query - Consider using a price tracking service (e.g., Keepa, CamelCamelCamel) as a data source for Amazon price history - Rate limiting and caching are essential — product pages change slowly, no need to hit APIs on every recommendation ## Acceptance Criteria - [ ] WardrobeAI can verify stock availability for a specific size+color before recommending - [ ] Current price and sale status are shown for each recommended item - [ ] Users see "Verified In Stock" or "Check Availability" labels - [ ] Size chart data is stored per brand and used to recommend the right size - [ ] At least 3 retailers are compared for each item (where available) ## Labels feature

sscoble commented

2026-05-25 17:15:26 +00:00

Owner

Completed: 50KB design doc written at docs/wardrobeai-sale-price-tracking.md covering all 4 phases (Stock/Price, History/Alerts, Cross-Retailer, Size/Fit) with API integration patterns, data schemas, and UX flows.