A Shopify A/B test should have one decision metric tied to the business goal, supported by diagnostic metrics and guardrails that reveal where an effect occurred and whether it caused harm.
Pick the decision metric
Completed-order conversion rate is useful for purchase frequency, while revenue per visitor includes both conversion and order value. Choose the measure that matches the decision before results are visible.
Use funnel metrics diagnostically
Product views, add-to-cart rate, checkout starts, and payment steps help locate behavior changes. They should not replace the primary metric merely because one looks more favorable.
Set guardrails
Monitor page speed, error rate, returns, discount cost, average order value, and other measures the change could damage. A variant can improve one funnel step while reducing total commercial value.
Define the measurement window
Decide whether a conversion is credited in the same session or within a longer attribution window. Apply the same rule to every variant and confirm that wallet and third-party checkout orders remain linked.
Run the test in Shopify
Sofcy supports URL, template, theme, and custom JavaScript API experiments.
Start free trial