Book Data Sources

Where we get book metadata (titles, authors, covers, descriptions, ISBNs).

Requirements

• Accurate metadata (title, author, publication date)
• High-quality cover images
• ISBN-10 and ISBN-13 support
• Good coverage of both popular and niche titles
• Reasonable API costs at scale

Options

Source	Pros	Cons
Hardcover	Modern API, good data quality, book-focused community	Newer, smaller catalog
ISBNdb	Comprehensive ISBN database, excellent coverage	Paid API (~$10/mo for 1k requests)
Open Library	Free, open source, large catalog	Inconsistent data quality, missing covers
Google Books	Large catalog, free tier	Rate limits, inconsistent covers, TOS restrictions

Evaluation Criteria

Hardcover

• Built by book lovers, for book lovers
• API designed for reading apps
• Community-contributed data
• Worth evaluating for primary source

ISBNdb

• 32+ million books
• REST API with JSON responses
• Includes covers, descriptions, subjects
• Paid but reasonable pricing
• Good for comprehensive ISBN lookups

Open Library

• Part of Internet Archive
• Fully open data
• Covers API separate from book API
• Data quality varies significantly
• Good as fallback, not primary

Google Books

• Largest catalog
• Free tier available
• Cover quality inconsistent
• Rate limits can be restrictive
• TOS may restrict some use cases

Strategy

User searches for book
        │
        ▼
┌─────────────────┐
│ Primary Source  │ ← Hardcover or ISBNdb (TBD)
└────────┬────────┘
         │ Not found?
         ▼
┌─────────────────┐
│ Fallback Source │ ← Open Library
└────────┬────────┘
         │ Not found?
         ▼
┌─────────────────┐
│  Manual Entry   │ ← User provides details
└─────────────────┘

Caching

• Cache book metadata in our database
• Refresh on-demand or periodically
• Store covers in S3/CDN
• Reduce API calls and improve performance

Open Questions

• [ ] Which source has best cover image quality?
• [ ] How do we handle books with multiple editions?
• [ ] Do we need to support non-English books initially?
• [ ] What's the cost projection at 10k, 100k, 1M users?