Why baselines matter before UK change windows
Friday afternoon deployments still happen. A short baseline window beforehand makes Monday’s review far less speculative.
Change windows concentrate risk. They also concentrate opinion. Without a baseline, the Monday stand-up becomes a collection of anecdotes from whoever was on call.
A baseline does not need to be elaborate. Two quiet mid-week days of the same traffic pattern you expect after the release is often enough for customer-facing applications. For batch-heavy systems, match the day-of-week pattern instead of chasing a perfect seven-day sample.
Agree the window in writing with the release owner. Note seasonal effects — month-end processing, school holidays, or marketing campaigns — so nobody later claims the baseline was unfair.
When instrumentation is thin, say so. A documented gap is more honest than a fabricated precision. You can still compare support volume, manual checks, and a few high-trust counters.
Teams that set baselines tend to shorten their post-release debates. The conversation moves from “did anything break?” to “which signal moved, and does that match the intent of the change?”