UX

    A/B Testing

    Term #89

    Comparing two versions with equal-sized groups. Version A is the existing one, version B is the change. Every major store change is worth testing before full rollout.

    In practice

    A/B testing is a scientific experiment. Without it, you're guessing. With it, you know. 40% of A/B tests reveal that the 'improvement' you assumed would help actually reduces conversions.

    Common misconception

    'A one-week A/B test is enough.' Wrong. You need 100-200 conversions PER VARIANT. CR = 2% with 200 daily visits gives 4 conversions/day — 50 days to a result.

    How to fix it

    Not reaching significance? (1) Continue until you hit the required sample size, (2) check for external factors (holidays, unusual promotions), (3) don't change anything else during the test.

    Example

    An oils store tested 'Buy Now' vs. 'Add to Cart.' Over 3 weeks and 4,000 visits: 'Add to Cart' had a 3.1% CR, 'Buy Now' had 2.7%. A 14.8% difference at 97% significance.

    Priority

    Highest-ROI order to test: (1) email subject line, (2) product-page CTA, (3) main image, (4) free-shipping threshold, (5) review placement.

    Jimmy tracks your A/B Testing, automatically

    WhatsApp alerts when something important changes. No dashboards, no hassle.