How Link Split Testing Works
You create one tracking link and give it two or more destinations, such as page A and page B. Each visitor is assigned one version, and keeps seeing it if they return. Your tracker records conversions for each version, so you can see which one wins.
The beauty of doing this at the link level is that it works with almost any page builder, offer or traffic source. You don't need testing code on the page itself.
What's Worth Testing
Offers
Two affiliate offers, or two price points, for the same audience.
Headlines and Promise
The biggest lever on most landing pages.
Page Length
A short opt-in page against a long sales letter.
Funnel Order
Pre-sell page first versus straight to the offer.
Big changes show results faster. A new button colour needs far more traffic to prove anything than a new offer does.
Running a Test, Step by Step
- Make sure conversions are trackedEvery version must report the same conversion. See our conversion tracking guide.
- Write down your hypothesis"A shorter page will raise opt-ins because the offer is simple."
- Work out the sample sizeUse a calculator with your current conversion rate and the smallest improvement you care about.
- Split traffic evenly50/50 for two versions, so the comparison is fair.
- Run at least a full weekWeekday and weekend visitors behave differently.
- Stop at your planned sample sizeNot when one version first looks ahead.
- Roll out the winner and test againEach test builds on the last.
A Worked Example
Illustrative numbers: 2,000 visitors split evenly between two pages.
| Version | Visitors | Sales | Conversion rate | Revenue per visitor |
|---|---|---|---|---|
| Page A | 1,000 | 20 | 2.0% | $0.94 at $47 |
| Page B | 1,000 | 28 | 2.8% | $1.32 at $47 |
Page B looks better, but whether 1,000 visitors each is enough depends on the sample size you calculated. If your calculator said 3,000 per version, keep going.
Mistakes That Pick the Wrong Winner
- Stopping at the first "significant" peek. Early leads often fade.
- Testing too many things at once so you can't tell what caused the change.
- Comparing clicks instead of sales.
- Changing the traffic mid-test, such as launching a new ad halfway.
- Skipping an A/A check. Sending traffic to two identical pages confirms your tracking splits fairly.
Split Testing in ClickMagick
ClickMagick includes split testing with conversion tracking, so one link can send visitors to several pages or offers and report sales for each. Returning visitors see the same version, and results sit alongside your ad and funnel data. See ClickMagick funnel tracking.
Test pages and offers by sales, not guesses.
Try ClickMagick FreeFrequently Asked Questions
What is split testing with links?
You share one tracking link that sends visitors to two or more versions of a page or offer, then compare conversions to see which version performs better.
How long should a split test run?
Long enough to reach the sample size you worked out before starting, and at least a full week so every day of the week is included. Don't stop early because one version looks ahead.
How much traffic do I need?
It depends on your current conversion rate and the size of improvement you want to detect. A sample size calculator gives you a number per version before you start.
What should I split test first?
Start with changes likely to matter: the offer, the headline, the main promise or the page length. Small cosmetic tweaks need far more traffic to show a difference.
Can I split test affiliate offers?
Yes. Point the split test at two offers or two pre-sell pages and track commissions with postbacks, so you compare earnings rather than clicks.