Any single-change uplift above ~25% is more likely a bug than a breakthrough.

Any single-change uplift above ~25% is more likely a bug than a breakthrough.
Twyman's law: figures that look interesting are usually wrong. A 40% checkout lift from a button color almost always traces to broken tracking, a leaked variant, or contaminated assignment.

— Investigate big wins harder than flat results.
— Replicate before you celebrate; extraordinary deltas need extraordinary QA.

Read the number, not the story. [implausibility = investigate]