Setting up and maintaining a script with realistic seed data is unrewarding and tedious. But as a system grows it starts paying dividends.
Testing and showing different use cases is a matter of seconds instead of filling in unrealistic dummy data that doesn’t cover edge cases.
Acceptance tests can use it as starting point and no additional fixture or factory library is needed.
A good seed script can be run multiple times to create increasing amounts of data. Capacity planning a system just got a whole lot easier.
Invest in a good seed script with realistic data that covers all different use cases of the system, it will pay off in the near future.