I probably wouldn't migrate a project, but I switched my default db to PostgreSQL after seeing the many ways MySQL can silently screw up data. Truncating strings, truncating numbers, allowing NULLs in a NOT NULL, etc. Much (all?) of this can be fixed with settings these days: https://dev.mysql.com/doc/refman/5.7/en/constraint-invalid-d...

PostgreSQL has a hard bias toward correctness. If you give it invalid data, it will blow up loudly every time.

PostgreSQL also has a boatload of useful features. Need full-text search? It has it, and it's good enough for lots of stuff: http://rachbelaid.com/postgres-full-text-search-is-good-enou...

Need to search for locations with coordinates < N kilometers from a coastline? PostGIS does it.

Need to store and query JSON? Partition tables by a date range? Index only the rows matching a WHERE condition? Use a CHECK constraint? Prevent race conditions from creating reservation rows with overlapping datetime ranges (exclusion constraints)? Run a query periodically and cache the results as a queryable table (materialized views)? The list goes on and on.

More correct data + fewer reasons to add another tool to the stack - especially when that would mean having to keep data in sync - is pretty compelling to me.