Experimentation isn't just for scientists. I run experiments all the time to see what works and what doesn't. Some become habits or new processes, but most don't. I've found that even the ideas that don't stick can be just as valuable.
Experimentation is especially important in software development, where things change quickly, and adaptation is essential.
A Probationary Period for Your Ideas
An experiment is a trial run that lets you walk away if something isn't truly helpful. Being "a little helpful" usually isn't enough. If a new process saves time in one place but costs time in another, that's not real progress.
Giving your experiment a time limit takes away the pressure to make it work. We all want our ideas to succeed, so it's easy to hold onto experiments longer than we should. A deadline helps you avoid sunk-cost thinking. If something isn't helping, move on.
At RDG, we apply this to software features: ship something small, see how it lands, keep what works, cut what doesn't.
Humility Is Built In
Trying new things means admitting you don't have all the answers, and that's okay. Companies that keep learning and changing usually do better than those that stop after finding one thing that works.
At RDG, we focus on solving the problem right in front of us, not just repeating what worked for someone else years ago. We use our experience without getting stuck in old ways.
Evaluate Retrospectively
Every experiment needs an honest review: what worked, what didn't, and what new ideas came up. If the result is just okay, it's not much better than failing. If you don't point out average processes, they can pile up over time.
I've noticed that teams are always trying new things, often without realizing it. If you aren't intentional, these accidental experiments can let bad habits slowly become normal.
Keep the Foundation Stable
You can't change everything at once. Teams need a steady base to experiment from, or else it's hard to know what's working.
For me, this means keeping experiments focused on a small area. We use this same idea with our clients. Small, step-by-step releases are easier to review, easier to change, and less likely to mess up what's already working. Clients can decide when they're done at any time.