Sample Size Calculator Based on Conversion Rate

Calculator Inputs

Baseline conversion rate (%)

Lift type

Expected lift value

Confidence level (%)

Statistical power (%)

Test direction

Variation to control allocation ratio

Number of variations

Daily eligible visitors

Traffic used for test (%)

Expected data loss (%)

Value per conversion

Example Data Table

Scenario	Baseline Rate	Lift Type	Lift	Confidence	Power	Planning Note
Landing page test	4.50%	Relative	12%	95%	80%	Good for a moderate page improvement.
Checkout button test	8.00%	Absolute	1.20 points	95%	90%	Needs stronger power for a key funnel step.
Signup form test	2.20%	Relative	20%	90%	80%	Useful when traffic is limited.

Formula Used

The calculator uses a normal approximation for a two proportion conversion test. It estimates the visitors needed in the control group and each variation.

Control visitors = ((z alpha × square root of pooled variance) + (z beta × square root of group variance)) squared, divided by the squared rate difference.

Baseline rate is p1. Variation rate is p2. The effect is p2 minus p1. The allocation ratio is variation visitors divided by control visitors.

For several variations, alpha is divided by the number of control versus variation comparisons. This is a Bonferroni style adjustment.

The final target is inflated by expected data loss. Run time is total adjusted visitors divided by daily visitors assigned to the test.

How to Use This Calculator

Enter the current conversion rate from reliable analytics data.
Choose whether the expected lift is relative or absolute.
Set the confidence level and statistical power.
Select one sided or two sided testing based on your decision rule.
Add allocation, variation count, traffic share, and data loss.
Press the calculate button and review the result above the form.
Export the result as CSV or PDF for test planning records.

Why Sample Size Matters

A conversion test needs enough visitors before it can prove a real change. Small samples move quickly, but they also create noisy results. A landing page may look better after one busy day. It may fade after more traffic arrives. This calculator helps you set a practical target before the test starts.

Core Planning Ideas

The baseline conversion rate is the current rate for the control page. The expected lift is the smallest improvement worth detecting. A low baseline needs more visitors than a high baseline. A tiny lift also needs more visitors. Confidence controls false positive risk. Power controls the chance of finding the lift when it truly exists. Both settings raise sample size when they increase.

Advanced Test Settings

Many teams compare more than one variation against a control. That increases the risk of a lucky winner. The calculator adjusts the alpha level by the number of planned comparisons. This keeps the experiment more disciplined. You can also choose one sided or two sided testing. Two sided tests are safer when either a gain or a loss matters. One sided tests need fewer visitors, but they fit narrower decisions.

Traffic And Duration

A sample target is only useful when matched with traffic. Daily eligible visitors and traffic share estimate run time. A data loss setting inflates the requirement for bot filters, tracking gaps, consent limits, or exclusions. This prevents the final analysis from falling short after cleaning.

Using The Result

Use the total adjusted sample as the main launch target. Use the control and variation rows for allocation planning. Check expected conversions to judge data stability. Review the estimated days before starting. Avoid stopping early because the current winner looks exciting. Early stopping can increase false positives. A fixed target keeps the decision cleaner.

Good Experiment Habits

Plan one primary conversion metric before launch. Keep page changes stable during the run. Segment results after the main decision, not before it. Watch data quality each day. Do not mix campaigns with different intent unless that traffic is planned. A strong sample size does not fix a weak test design. It simply gives a fairer chance to measure the change you care about.

Reduce decision doubt.

FAQs

What is a conversion rate sample size?

It is the visitor count needed to compare a control rate with a variation rate. The target depends on baseline rate, lift, confidence, power, and test direction.

Why does a smaller lift need more visitors?

A small lift is harder to separate from normal random noise. More visitors make the estimate steadier and improve the chance of detecting that smaller effect.

Should I use relative or absolute lift?

Use relative lift when improvement is stated as a percent increase over the baseline. Use absolute lift when the target is measured in percentage points.

What does statistical power mean?

Power is the chance of detecting the planned lift when it truly exists. Higher power reduces missed wins, but it increases the sample size requirement.

When should I use a two sided test?

Use a two sided test when both improvement and harm matter. It is the safer default for most conversion experiments and product decisions.

Why adjust alpha for several variations?

More variations create more chances for a lucky result. Alpha adjustment controls that extra risk when every variation is compared against the same control.

Can I stop the test early?

Stopping early can raise false positive risk. Use the calculated target unless you have a planned sequential testing method or a serious data quality issue.

What does data loss mean?

Data loss covers excluded visitors, bot filtering, tracking gaps, consent limits, and invalid sessions. The calculator inflates the sample target to preserve enough usable data.