When to stop an A/B test without guessing
Stopping rules, peeking habits, and how to decide whether a lift is ready for a product call.
Field notes
Practical writing on A/B testing performance analytics — stopping rules, samples, guardrails, and segment conflicts.
Sample size myths that quietly ruin app experiments
Why “we had thousands of users” is not the same as a trustworthy comparison between variants.
Guardrail metrics worth watching during A/B tests
Primary metrics win headlines. Guardrails catch the quiet damage that makes a “winning” variant expensive.
Reading conflicting segment results without cherry-picking
New users love the new flow; returning users hate it. Here is how to talk about that without rewriting the hypothesis after the fact.