The new column was live before the migration finished. You could see it in the schema, blank but ready. No one asked for a wiki page. No one opened a ticket. It was decided, executed, and deployed in minutes.
Adding a new column should be simple. In most stacks, it is not. You fight with schema drift. You wait on migrations to run in production. You hope the ORM behaves. You pray indexes don’t lock tables in the wrong environment. Every extra step increases the risk of downtime and data loss.
A clean workflow treats each new column as an atomic change. Define the column with explicit data types. Align nullability and defaults to avoid inconsistent rows. Ensure indexes support future queries without overloading writes. Roll out in stages—first create the new column, then backfill data in controlled batches, then switch application writes to fill it. This reduces contention and rollback complexity.