The current recommendation engine is based on mood, pace, and genre preferences, which is a great starting point. But those categories are really broad. I've read 45 books tagged "fae" and rated them everywhere from 1.75 to 4.75. Telling me I like fae books doesn't help me find my next great read — the tag doesn't predict quality for me.

What actually predicts whether I'll love a book is whether readers who rate the same books I do also loved it. That's collaborative filtering, and StoryGraph already has the building blocks:

• You already have Similar Users matching
• You already have everyone's ratings data
• You already know which books users have in common

What I'd love to see:

Take the users who are my strongest taste matches (based on actual rating alignment on shared books, not mood/genre profiles) and surface the books they rated 4+ stars that I haven't read yet.

Something like:

"3 readers with 80%+ taste match rated this book 4.5 or higher"

Or even just a recommendation category called "Loved by readers like you" that pulls from taste-matched users' top-rated books.

This would be a massive leap in recommendation quality because it doesn't rely on broad categories that don't predict individual taste. It relies on the most reliable signal there is: people who agree with me about specific books tend to agree about other books too.

Tags, genres, and moods describe what a book is. Taste-matched readers tell you whether a book is good for you specifically. That's the difference.