What minimum detectable effect means
The minimum detectable effect (MDE) is the smallest true difference between variants that an experiment is adequately powered to detect at a given significance level and power. It is set before launch alongside the baseline conversion rate, the alpha (typically 0.05), and the power (typically 0.80). A test is not designed to detect any effect; it is designed to reliably catch effects at or above the MDE.
Why the curve flattens with sample size
The MDE shrinks with sample size, but not linearly. Because statistical precision improves with the square root of the number of observations, detectable effect size scales roughly with one over the square root of n. The practical consequence is that doubling traffic lets you detect effects about 30 percent smaller (since one over the square root of two is about 0.71), and chasing very small lifts requires disproportionately large samples. The curve therefore steepens at small samples and flattens as traffic grows, which is why detecting a tiny lift can be prohibitively expensive.