A/B testing landing pages with a single short link

How split testing at the link works

You attach several destination URLs to one short link with percentage weights. Each new visitor is assigned a variation, and a cookie keeps them on the same one so their experience is consistent across visits.

Stickiness matters more than people expect

Without it, a visitor who returns sees a different page, your conversion data becomes meaningless, and the experience feels broken. Any serious implementation sets a cookie on first assignment.

Rotation is not a test

Round-robin rotation sends clicks evenly across destinations regardless of visitor identity. It is right for load distribution or rotating affiliate offers, and wrong for measuring which page converts better.

Sample size discipline

Stopping the moment one variation looks ahead is the most common way to get a wrong answer. Decide the sample size and the duration before you start, run for whole weeks so weekday and weekend traffic are both represented, and do not peek-and-stop.

Test one thing

If variation B changes the headline, the image and the price, a win tells you nothing about which change caused it. Change one element per test unless you have the traffic for a proper multivariate design.

Where the numbers come from

Per-variation click counts come from the link platform; conversions need the destination to report back, either with a conversion pixel or a server-side call carrying the click identifier. Without that second half you are testing clicks, not outcomes.