A/B testing compares two versions of one thing, such as a page, ad or email, by showing each to a random half of visitors and measuring which converts better.
Worked0.6 ÷ 2.4 = 0.25
A clinic's booking page converts 2.4 of every 100 visitors. A new version with the doctor's photo and a shorter form converts 3.0. The difference is 0.6, so the lift is 0.6 divided by 2.4, which is 0.25, or 25 percent, but only worth trusting if enough people saw each version.
A/B testing replaces opinions with evidence. Instead of arguing over whether the button should be green or the headline shorter, you split traffic at random, send half to the original (the control) and half to the change (the variant), and see which gets more of the action you care about. Because both run at the same time, season, ad spend and day of the week affect them equally. Google Ads offers the same logic as campaign experiments.
Done properly, it turns small gains into compounding ones. Done badly, it turns noise into confident wrong decisions.
If a page gets 600 visitors a month and converts 3 in 100, a test would take many months to detect anything but a huge change. A school or clinic with low traffic gains more from fixing obvious problems, using customer calls and session recordings, and then comparing before and after.
BridgeE runs this as part of conversion rate optimisation, and says plainly when the traffic is too low to test.