Codebases evolve quickly at high growth startups as new changes are pushed out to production daily. There are a lot of moving pieces and it’s nearly impossible to be able to catch every bug before it hits production. These systems grow in complexity over time to the point where there will be side-effects that no one is able to predict, especially when the code runs at scale. At some point, an incident will happen. 

Incidents are inevitable—there’s no way to fully prevent them. At some point you’ll be pulled into an incident because some code is causing issues in production, whether it’s code you wrote or someone else’s. In order to prevent confusion and miscommunication during high stress situations, engineering organizations often define incident severity and priority levels ahead of time to align and prepare for these situations. You won’t be able to prevent incidents from happening, but how you respond to them can make a big difference.