Category
Experiments
Variants, traffic split, statistical significance.
Pause, resume, or complete an experiment
When something looks off, pause. When you have a winner, complete. Here's the safe way to control a running A/B test.
Choose a goal for your experiment
Pick the right conversion goal so your A/B test measures something useful — pageviews, clicks, or custom events.
Understanding statistical significance
What "95% confidence" actually means, and when to call a winner.
How traffic split works
Each visitor is hashed into a stable bucket so they always see the same variant.