Inspiration

Most people decide whether a new serum "works" by gut feeling, and by the time they are sure, the 30-day return window has closed. Shoppers lose money; retailers get late, angry returns and no signal about why.

What it does

Keep or Return turns the return window into a small, objective skin trial.

  1. Register a purchase – product, price, purchase date, return window, the concern it claims to fix (acne, redness, wrinkle, moisture, pore, spots...) and the improvement you expect.
  2. Day 0 baseline – a selfie is scored by the YouCam Skin Analysis API.
  3. Check-ins on day 7, 14, 21 – only the concern the product promises is tracked, the other scores are shown as context.
  4. Verdict before the deadline – KEEP / RETURN / KEEP TESTING, with the trend chart, days left, the return-by date and the money at stake. ## Verdict logic
  5. Improvement on the target concern ≥ your bar → KEEP
  6. Flat or worse trend, or 7 days or fewer left and the projected trend won't reach your bar → RETURN (while you still can)
  7. Otherwise → KEEP TESTING, with the projected score at the deadline

How we built it

A single-file web app (HTML/CSS/vanilla JS, no backend). Live mode calls the YouCam Skin Analysis REST API: upload the selfie → create a skin-analysis task → poll the result → read per-concern scores. The API key stays in the browser (localStorage) and only scores are stored, never photos. Demo mode uses sample scores so anyone can try the full flow without a key.

Challenges

  • Skin scores move slowly and depend on lighting, so the app compares only the claimed concern against a user-set bar instead of an overall score.
  • Making the verdict arrive before the deadline, not after it.

What we learned

Skin AI is most useful when it answers one concrete decision the shopper already has – keep or return – rather than showing a dashboard of numbers.

Consumer & eCommerce value

  • Shoppers: an objective answer before money is lost.
  • Retailers: earlier, data-backed returns and per-product evidence of what works for which skin concern.

What's next

Lighting-normalized check-in guidance, regimen comparison (two products side by side), and a retailer dashboard of anonymized outcomes.

Built With

Share this project:

Updates

Submission history