Creative and testing
Peeking
Checking a running test repeatedly and stopping it as soon as the result looks significant.
What it is
Every look is another chance for random noise to cross the line, so the real error rate is much higher than the threshold suggests.
Why it matters
For lead generation and e-commerceDashboards make peeking easy, and it is one of the main ways false winners enter ad accounts.
In practice
Set the sample size in advance and read once, or use a sequential method designed for repeated looks.
A common mistake
Treating a dashboard's live significance badge as a verdict.
Read the evidenceThe winner is smaller than it looks All terms