expert · incidents

One Week to Make It Faster

You have one week to make this system meaningfully faster. How do you decide what to work on?

Practical scenario

Leadership has allocated one engineering week to "performance" on a service with a 1.8 s p95. There are twelve open suggestions in a document, none with measurements attached.

What it tests

  • • Prioritisation from measurement rather than intuition
  • • Understanding of critical path and bottleneck migration
  • • Whether cost and regression risk enter the decision

Lessons behind this question

Measure Before You OptimizeThe Critical Path Is the Only Path That PaysThe Bottleneck Moves After Every FixEvery Optimization Buys Something and Sells SomethingRegression or Tuesday? Telling a Real Change from Noise