All guides
    Links
    Updated 7 October 2026

    How to A/B test and rotate link destinations

    Split one Smart Link’s traffic across two or more destinations by weight — rotate offers or A/B test landing pages — and see per-variant click counts in analytics. Country and device rules still take priority; rotation covers everyone else.

    A/B testing splits a link’s traffic across two or more destinations so you can see which converts best; rotation cycles clicks evenly between destinations. Both let one short link feed several endpoints, so you can experiment without ever changing what you have shared.

    1. 1

      Open the form and turn on “A/B testing & rotation”

      In the link builder, open Advanced options, expand A/B testing & rotation and switch it on.

    2. 2

      Add your destinations and weights

      Click + Add Destination for each variant, paste its URL, add an optional label, and set a weight (its share of traffic).

      • 1 — Two or more destinations, each with a label and weight (e.g. 50% / 50%).
      Two A/B rotation destinations, Landing A and Landing B, split 50/50 by weight
    3. 3

      Create your link and read the results

      Click Create Smart Link. Each visitor is sent to one destination based on the weights, and you’ll see clicks per variant in Analytics under the A/B tab.

    Frequently asked questions

    What's the difference between A/B testing and rotation?

    A/B testing sends a share of traffic to each variant so you can compare performance; rotation distributes clicks evenly, which is handy for load-sharing or alternating offers. Both show up in your analytics.

    Can I see which variant performs better?

    Yes. The A/B results in your analytics show clicks per variant, so you can pick a winner and point the whole link at it — without re-sharing anything.

    Does A/B testing change my short URL?

    No. You share one gsl.li link and the split happens behind it. You can change the variants or end the test at any time and the link keeps working.