Should I use PostgreSQL if I don't know my final data model yet?

Yes—PostgreSQL is usually a good choice even if you don’t know the final data model yet.

Why it works well in that situation:

• Flexible enough to evolve: You can start with a simple schema and add tables, columns, constraints, and indexes later.
• Handles semi-structured data: JSONB is very useful when parts of your model are still uncertain.
• Strong migration support: Schema changes are manageable with tools like Flyway, Liquibase, Alembic, Prisma, etc.
• You can still enforce structure gradually: Start loose, then add constraints as the model stabilizes.
• Widely supported and reliable: Good default choice for many projects.

When PostgreSQL is especially a good fit:

• You expect your data model to change over time
• You want to prototype quickly
• You may need both relational queries and document-like flexibility
• You care about correctness, transactions, and future scalability

When you might choose something else:

• If you truly have no idea what the data will look like and want maximum schema freedom, a document DB like MongoDB may feel easier initially
• If you need extremely simple key-value storage or event logs, another storage layer might be better
• If your project is tiny and short-lived, you might optimize for developer speed over long-term structure

A practical approach:

1. Start with PostgreSQL.
2. Keep the early schema minimal.
3. Use JSONB for uncertain fields.
4. Add migrations as the model becomes clearer.

So the short answer is: yes, PostgreSQL is often one of the best choices when the final model is still unknown.

If you want, I can also help you decide between PostgreSQL vs MongoDB vs SQLite for your specific project.