One short link can serve several destinations. There are two modes, and they answer different questions.
Weighted split testing
Add up to ten variations, each with a destination and a percentage. The main destination receives whatever is left: three variations at 20% each means the main destination gets 40%. The editor shows the running total and warns when the variations exceed 100%.
A visitor is assigned once and kept on the same variation by a cookie for thirty days, so their experience stays consistent across visits. That is what makes the comparison meaningful — a visitor who sees a different page on every click tells you nothing.
Clicks are recorded with the variation they were sent to, and the statistics page shows an A/B variations card breaking clicks down by destination. Combine that with conversion tracking and you can compare variations by conversions rather than by clicks, which is the number that actually matters.
Rotation mode
Rotation ignores visitor identity and cycles through the destinations in order, round-robin. No cookie is set. This is right for rotating offers, rotating affiliate destinations, or distributing traffic across several servers; it is wrong for measuring which page converts better, because the same person gets a different page each time and your two samples are not comparable.
Practical notes
- Variations are per link. Duplicating a link copies them.
- A matched geo rule takes precedence: a visitor who matches a geo rule is never assigned a variation.
- Bots and preview crawlers hit the same logic. If a link is shared somewhere with an aggressive preview fetcher, some of your early traffic is not people. Filter to human clicks before drawing conclusions.
- Changing percentages mid-test does not reassign existing visitors — their cookie still points at the variation they were given.
Reading the result
Wait for enough clicks that a swing of a few dozen would not flip the winner, and prefer human clicks and conversions over total clicks. If two variations are within a few percent of each other after a genuine sample, the honest conclusion is usually that the page is not the constraint.
Find this in your dashboard
- Sign in and select the workspace and domain you want to manage.
- Open Branded Links. Review the article’s steps and your plan’s available controls.
- Save your changes, reopen the record and verify the saved result. For routing changes, check the short link with a test visit.
