A/B testing works as follows: a company drafts several different versions of a particular webpage. Perhaps they try different colors or images, or different headlines for a news article, or different arrangements of items on the screen. Then they randomly assign incoming users to these various pages, usually in equal numbers. One user may see a red button, while another user may see a blue one; one may see DONATE and another may see CONTRIBUTE. The relevant metrics (e.g., click-through rate or average revenue per visitor) are then monitored. After a period of time, if statistically significant
...more
