A/B Testing Guide

What Is A/B Testing?

A/B testing (split testing) sends different percentages of your traffic to different destinations. For example, you might send 50% of clicks to Landing Page A and 50% to Landing Page B to see which converts better.

GeoRedir handles this with traffic split rules. Each rule has a weight (percentage), and GeoRedir randomly distributes visitors according to those weights.

Setting Up a Test

To create an A/B test, add two or more traffic split rules to your smart link:

Rule Type: Traffic Split
Weight: 50%
Destination: https://landing-page-a.com

Rule Type: Traffic Split
Weight: 50%
Destination: https://landing-page-b.com

You can test more than two variants. For example, a 3-way test with 33%/33%/34% splits, or an 80%/20% split to test a new variant with limited risk.

Statistical Significance

GeoRedir uses the chi-squared test to determine if one variant is statistically better than another. A test is considered significant when:

• The p-value is less than 0.05 (95% confidence)
• Each variant has received at least 100 clicks

You can view test results on the link detail page. GeoRedir shows a confidence bar for each variant and declares a winner when significance is reached.

Auto-Optimization

Enable auto-optimize on your smart link to let GeoRedir automatically shift traffic to the winning variant once statistical significance is reached. This maximizes your conversion rate without manual intervention.

Best Practices

• Test one variable at a time. Change only the landing page, the offer, or the CTA — not all at once.
• Use conversion tracking. Enable conversion tracking to measure actual conversions, not just clicks.
• Run tests for at least 7 days. Daily and weekly traffic patterns can skew short tests.
• Need at least 100 clicks per variant. Small sample sizes produce unreliable results.
• Document your tests. Use the link name/description to record what you're testing and the hypothesis.