Measure the sale, not the click
The easy number on a presell is click-through to the order page. The number that matters is sales, or revenue, per presell visitor. A curiosity headline can raise clicks and lower sales by sending readers who were never going to buy, so judge every variant on what happens after the click. That means tracking the visitor from presell to order page to order, with the variant and the traffic source attached.
Split-testing a presell
- Change one thing at a time: the headline, the lead, the offer placement. Two changes at once tell you nothing about either.
- Split traffic at random and keep the split stable for each visitor, so a returning reader sees the same version.
- Wait for enough sales before calling a winner. A dozen orders on each side is noise; decide your stopping rule before the test starts.
- Keep the ad's destination URL the same where you can. Changing the URL in a live ad can send it back through review and reset the platform's delivery learning.
What to test first, how to split traffic, how long to run and when to stop are covered step by step in landing page split testing for presells.
The numbers to keep per traffic source
Presell visitors, click-through to the order page, orders, revenue and refunds, each by variant and by source. The same presell often performs differently on social, native and search traffic, and a blended average hides which source is paying for the test. The full setup, for offer owners and for affiliates who only control the presell, is in how to measure presell-to-sale conversion by traffic source.
