Experiments

For anyone comparing two funnels head-to-head on the same traffic. You'll have randomized variant traffic and side-by-side analytics. You need at least two funnels.

What an experiment tests#

A single variable at a time: which funnel converts better, or which variant of the same funnel wins. Multivariate experiments are not supported; keep one variable per experiment.

Create an experiment#

  1. Open Dashboard → Experiments.
  2. Click Create.
  3. Name the experiment.
  4. Define the experiment goal — what you're testing and how you'll know which variant wins.
  5. Pick the funnels to include as variants. Usually two: control and treatment.
  6. Save.

Traffic distribution#

By default, Zellify distributes traffic evenly across all funnels in the experiment and ensures the total adds up to 100%. You can adjust the per-funnel split manually as long as the total still equals 100%.

Connect to traffic via a campaign#

Experiments don't drive traffic themselves. Create a campaign that points at the experiment. The campaign link distributes visitors randomly across variants.

Analytics#

Per-variant: revenue, number of conversions, conversion rate, total visits, paywall conversion, and revenue per visitor. For each metric, Zellify highlights the leading funnel so you can spot the winner at a glance.

Page-level comparison#

For deeper analysis, compare drop-off by page across variants. This is where you see not just "variant B wins" but "variant B wins because the quiz is shorter".

Renaming a page resets its analytics: Page-level metrics are keyed off the page name. Renaming a page mid-experiment starts its history over. Lock in page names before launching traffic.

Picking a winner#

Revenue beats conversion rate: A funnel with higher revenue at a lower conversion rate often wins. Conversion rate alone is misleading if the product price differs. Zellify highlights the leader per metric; compare revenue and conversion rate together.
Campaigns
What drives traffic to an experiment.
Funnels
The variants in an experiment.