Split testing is part of the 🧪 Pro Tools add on. It lets one tracking URL feed two or more destinations, then measures which one actually earns.
Creating A Test:
Split tests are created from the link, not from the Split Tests tab. Go to the 🔗 Links tab and click the 🧪 A/B Test button on the link you want to test. The 🧪 Split Tests tab then lists every test you have running.
The Test Form:
- Test Name. Your own label for the experiment.
- Variants. Variant A is the original destination. Add Variant B, and more if you want a multi way test, each with its own Variant URL.
- Declare winner after conversions on. Choose Any Variant, Variant A (Original), or Variant B (Challenger).
- Min conversions to trigger. The number of conversions required before a winner is declared, so a lucky early sale cannot end the test.
Save with Create Split Test, or Update Split Test when editing an existing one.
How The Leader Is Chosen:
While a test runs, WCT shows a leader based on sales conversion rate, meaning sales divided by clicks. If no variant has produced a sale yet, it falls back to the all conversions rate so a leads only test still shows a leader. Variants with fewer than ten clicks are ignored for leader calculation, which stops a single click looking like a hundred percent conversion rate. Once a winner is formally declared, that declaration takes precedence over the calculated leader.
Managing Tests:
The Split Tests tab has a search box for finding a test by name, and each row offers actions to edit, pause or resume, reset the collected data, and delete the test. Resetting is useful when you have changed a landing page mid test and want the numbers to start again from a clean baseline.
What Gets Tracked Per Variant:
Each variant accumulates its own clicks, conversions, leads, sales, gross revenue and net revenue, so you are comparing money and not just click through.
Running A Test That Tells You Something:
- Change one thing at a time. Two different headlines, or two different offers, not both at once.
- Set Min conversions high enough that the result is not noise. A handful of conversions is not a decision.
- Let it run across a full weekly cycle if your traffic varies by day.
- Judge on sales conversion rate and revenue per click rather than raw clicks.
- When a winner is declared, point the original link at the winning destination and start the next test.