What Is A Split URL Experiment?
A Split URL experiment is an A/B experiment where each variation is a separate page hosted on your own website, rather than a modified version of one page.You build the variation pages yourself and give FigPii their URLs. When the experiment launches, FigPii splits your visitors between those URLs and tracks the conversions each one generates.
Say you want to test your homepage at https://www.mysite.com/, comparing two new designs against the current one.
With a split URL test, you (or your designer or developer) build the two new designs as real pages on your site:
https://www.mysite.com/— your original homepagehttps://www.mysite.com/new_design_1.html— first variationhttps://www.mysite.com/new_design_2.html— second variation
All three pages exist on your website. When you create the split URL experiment, you enter all three URLs. On launch, FigPii divides incoming visitors between them and reports how each one performs.
When to use a split URL test
Split URL tests suit situations where the change is too large for the visual editor to handle well:
- Testing a completely redesigned page rather than individual elements
- Testing new functionality that needs its own code, not just styling changes
- Avoiding the flicker effect — because each variation is a real page, visitors never see the original flash before the variation loads. See The Flicker Effect.
The trade-off is build time. An A/B test can be set up in the visual editor in minutes; a split URL test needs the variation pages built and hosted first.
For a fuller comparison, see What Is The Difference Between AB Tests And A Split URL Test? and Which Is Better: AB Tests Or Split URL Tests?.
Before you launch
Install the FigPii tracking code on every variation page. Each variation is a separate page, so each one needs the tracking code in its <head> — the same as your original page. If a variation is missing the code, FigPii cannot record visits or conversions for it, and the results will look as though that variation is underperforming when it simply isn't being measured.
Check that all your variation pages are live and reachable before launching. FigPii sends real visitors to these URLs, so a page that returns a 404 will cost you traffic.