Put this into practice
Run A/B tests and personalize your store without code. 14-day free trial, no credit card.
Start free trial →Run A/B tests and personalize your store without code. 14-day free trial, no credit card.
Start free trial →A lookback window is the historical time period you examine to establish baseline metrics before launching an A/B test or analyzing conversion performance. In ecommerce experimentation, the lookback window determines what data you use to calculate your baseline conversion rate, average order value, and traffic volume — the inputs that feed sample size calculations and experiment planning. Choosing the right lookback window is critical: too short, and seasonal anomalies or one-time traffic spikes can produce a misleading baseline; too long, and gradual trend changes in visitor behavior get diluted.
The right lookback window depends on your store's traffic patterns and business cycle:
Standard recommendations:
What to exclude from the lookback window:
An incorrect baseline from a poorly chosen lookback window cascades into every downstream decision: you'll calculate the wrong sample size, run the test for the wrong duration, and potentially misinterpret the results. A Shopify brand that measures baseline conversion rate during a 20%-off sale period will set an inflated baseline — then struggle to replicate those numbers in the actual test, which runs under normal pricing. The result looks like a negative test when the real cause was an overstated baseline.
Boat Lifestyle was preparing to test a new product carousel on their homepage. Their initial lookback window covered the previous 30 days — which happened to include their anniversary sale, during which conversion rates were 40% above normal. Using that period as their baseline would have made any normal-period test look like a failure. Their analyst extended the lookback to exclude the sale period, using the 30 days before the sale began. The revised baseline conversion rate was 2.1% rather than 2.9%. The subsequent test, sized on the 2.1% baseline, reached significance at the expected time and produced a credible 8% lift.
The lookback window feeds the baseline metric, which feeds the sample size calculation, which determines run time. Errors in the lookback window propagate through the entire test planning process. Building a standard process for establishing and documenting baselines — including the lookback window dates, any excluded periods, and the reason for exclusions — is a hallmark of a mature experimentation program.
Run smarter A/B tests with CustomFit.ai — 14-day free trial, no credit card required.