How we know when to ship things — does it make things worse?

We’ve started to take much smaller steps with our merges into master and subsequent deploys to production. Oftentimes we’ll use a feature flag to hide any in-progress work from the user.

It can sometimes be hard to know when to turn on the new experience to the user. How many requirements do we need to get in before its good enough to give to our customers?

The question I find useful is

Does it make things worse?

If it does make things worse for the end customer then don’t ship it (unless you’ve got good reasons). However, if it makes things better or even keeps them on a par with the current experience turn it on and let real people use it. You’ll find out about any bugs faster, you’ll learn about the impact faster and you can continue adding additional features.

I would always advocate for getting new things to your users as fast as possible (aiming to not make things worse).