The build broke after the last deploy. You trace it back. The schema changed. One missing column in the database throws the entire service into chaos. A new column can stabilize everything—if you add it the right way.
Adding a new column in production is simple in theory and dangerous in practice. The process is more than running ALTER TABLE. It’s about anticipating queries, indexes, null defaults, and migration speed. Get it wrong and you block writes, lock tables, or corrupt data integrity. Get it right and your feature ships without a hitch.
Start with your migration strategy. For large datasets, use techniques that avoid locking. Break the change into multiple deploy steps. Create the column with a default state and fill it in asynchronously. Test it against a replica before touching production. Measure query plans after the change—indexes that worked before might produce different performance profiles now.