Baseline Conversion Rate in A/B Test Planning
Baseline conversion rate is the expected conversion rate for the control experience before the experiment starts. It matters because the same target lift can require different sample sizes depending on where you start.
Why baseline changes the calculation
A change from 2% to 3% behaves differently from a change from 20% to 21%, even though both might look simple on paper. Baseline shapes the variance in the outcome and therefore affects the sample size requirement.
That is why recent control data is preferable to a rough guess.
How to choose a baseline
The best baseline usually comes from recent, comparable data from the same funnel step or experiment context. If seasonality, traffic source, or audience mix has changed, older baselines may be misleading.
If uncertainty is high, test planners sometimes calculate a range using more than one plausible baseline.
What to avoid
Avoid using aspirational targets as baselines. The baseline should reflect current reality, not what you hope to reach. Optimistic baselines can produce misleading traffic estimates and poor test planning.
How to keep the baseline trustworthy
The best baseline is recent data from the same audience and the same step in the funnel you plan to test. A broad sitewide average can be misleading if the experiment will run on a narrower or more volatile segment.
When uncertainty is high, it is often better to estimate a range of sample sizes using more than one plausible baseline. That gives the team a planning band instead of a false sense of precision.
- Prefer recent control data over historical averages
- Match the baseline to the exact test audience and metric
- Watch for seasonality or traffic mix changes
- Use a range when baseline uncertainty is still high