A/B tests

Tests are judged on revenue per visitor, not clicks. A version that gets fewer signups but sells bigger plans wins here.

Start a test with one line

JavaScript

// After the tracking script has loaded (for example in a React useEffect):
const variant = dc.variant("pricing-page", ["control", "annual-first"]);
if (variant === "annual-first") showAnnualPlanFirst();

// Uneven split: 80% keep the current page, 20% see the new one.
dc.variant("checkout-copy", { control: 80, "short-copy": 20 });

Each visitor always gets the same variant; nothing is stored beyond the existing visitor cookie.
The test appears here as soon as the first visitor sees it. Results count purchases within 30 days of that first view.
On cookieless sites, a visitor keeps their variant for one visit only, so results are noisier.