Split testing answers a simple question: which destination works better? Send 80% of clicks to your current page and 20% to an experiment, then let the click counts tell you.
Setting up a split
Open the link in the editor and set two fields together:
- Split target URL — the alternative destination.
- Split percentage — the share of clicks (1–100) that go there. The rest still go to the link's normal target.
The two fields are a pair: set both, or clear both. Every visitor is randomly assigned on the spot — roughly speaking, a 20% split sends about one in five clicks to the variant.
Reading the results
Each click is recorded with which destination it received (primary or split), and the stats page breaks clicks down by variant. Compare human clicks against whatever the destination measures — signups, sales, replies — and keep the winner.
Running a fair test
- Use a 302 while testing. Permanent redirects may be briefly cached, which can skew who sees which variant. Flip to 301 only after you've retired the experiment — see 301 vs 302.
- Let it collect enough clicks. A 50/50 split on ten clicks tells you nothing; give it real traffic or a bigger share of it.
- Change one thing at a time. The split target should differ from the primary in the one way you're testing.
Ending the test
When you have a winner, clear both split fields — the link then sends everyone to its normal target (edit that target first if the variant won). The historical clicks keep their primary/split labels, so your before/after comparison stays intact.