CRO · November 2024 · 7 min read

    A/B Testing for eCommerce: What to Test First (and Why)

    A prioritized testing roadmap for Shopify and WooCommerce stores: what to test first, how to run a valid test, and how to read the result.

    Split testing comparison on a whiteboard

    By Cristian Daron

    Most eCommerce stores test the wrong things first. They'll spend weeks testing button colors when the real revenue is hiding in checkout flow changes, pricing psychology, and product page layout. Here's the A/B testing priority framework I use with every client.

    The Revenue Impact Matrix

    I rank every potential test on two axes: expected revenue impact (high/medium/low) and implementation effort (easy/medium/hard). Tests that are high-impact and easy to implement go first. This sounds obvious, but most teams skip this step and test whatever the CEO suggests.

    Tier 1: Checkout Tests (Highest ROI)

    The checkout touches 100% of your converting traffic. Small improvements here have outsized impact. Tests I always run first: one-page vs. multi-step checkout, express payment button placement, trust badge positioning, shipping cost visibility (show early vs. late), and progress indicator presence.

    Tier 2: Product Page Tests

    Product pages are where purchase intent is formed. High-priority tests: image gallery layout (grid vs. carousel vs. zoom), CTA button copy and placement, social proof format (stars vs. review count vs. photo reviews), pricing display (with/without crossed-out original price), and sticky add-to-cart bar on mobile.

    Tier 3: Collection & Category Pages

    These pages influence browsing behavior and product discovery. Test: grid columns (3 vs. 4 on desktop), product card information density, filter sidebar position and format, sort default (bestsellers vs. newest vs. price), and quick-add-to-cart functionality on hover.

    Running Valid Tests

    Invalid tests waste time and money. Rules I follow: decide the sample size before the test starts with a calculator such as Evan Miller's, run for at least two full business cycles, never change several variables in one test, and judge the result by its interval rather than its headline number.

    Documenting & Compounding Wins

    Every test result, win, lose, or inconclusive, gets documented with the hypothesis, variant details, traffic split, duration, and results. Winners are immediately implemented. Winners compound, because each one becomes the control the next test has to beat. Losing and inconclusive tests are reported too: a programme that only shows its winners is not a programme.

    Tools I Use

    For Shopify: Intelligems, Convert.com or VWO (Google Optimize was shut down in September 2023). For custom setups: I build lightweight A/B testing using Checkout Champ's built-in split-testing. For email: Klaviyo's built-in A/B testing for subject lines, send times, and content blocks.

    Sources

    1. Evan Miller, A/B test sample size calculator (opens in a new tab)
    2. Google, Sunset of Google Optimize (September 2023) (opens in a new tab)
    3. This site: the Puur Smile engagement analytics and A/B tests