The control is the original, unmodified version of a webpage, feature, or experience that serves as the benchmark in an A/B test. It represents the status quo — whatever is currently live and generating your baseline conversion rate. Every variant in an experiment is measured against the control to determine whether the proposed change produces a statistically meaningful improvement or decline in the target metric.
Why Control Matters for Ecommerce
Without a control, you have no reference point. The control anchors your experiment and prevents you from attributing random traffic fluctuations to a design change. For D2C brands on Shopify, the control is often a page that has been accumulating incremental optimizations over months — it represents your current best practice. Running a well-isolated control means you can be confident that any difference in performance between groups comes from the variant change, not from a flash sale, a seasonal spike, or a broken promo code affecting one group differently.
Real-World Example
Sugar Cosmetics ran an experiment on their checkout page to test whether adding a trust badge near the "Place Order" button would reduce checkout abandonment. The control was the existing checkout page with no badge. The variant added a "Secure Payment" icon with a short reassurance line. At the end of the test, the control had a 64% checkout completion rate; the variant reached 69%. The control's rate became the denominator for calculating the 7.8% relative lift.
How to Protect Your Control
- Never modify the control mid-test — any change to the live page during the experiment invalidates the comparison.
- Ensure the control and variants receive traffic simultaneously, not sequentially, to avoid time-of-week or seasonal confounds.
- Verify that your testing tool is correctly bucketing users so the same visitor always sees the same experience (control or variant).
- Document the control state with screenshots or version snapshots before the test begins so you can reference it during analysis.
- Monitor the control's metrics independently — a sudden drop in control conversions might signal a site issue unrelated to the test.
Control in A/B Testing
In every A/B test, the control is Group A. It receives its defined share of traffic — typically 50% in a standard two-way split — and its conversion rate becomes the baseline against which the variant's performance is assessed. Statistical significance is calculated by comparing the distribution of outcomes in the control group versus the variant group.
Run smarter A/B tests with CustomFit.ai — 14-day free trial, no credit card required.