Free Incrementality Testing Template: Test Plan and Excel Results

Plan your test before launch, then calculate the revenue difference between randomized test and control groups. This workbook includes an editable plan and a complete numerical example.

What is in the free Excel file?

Results: editable audience sizes, group revenue, spend, attributed revenue and margin. Formulas calculate revenue per person, estimated incremental revenue, iROAS and contribution after ads.
Test plan: fifteen planning questions, example guidance and a blank column for your own plan. Cover assignment, outcomes, timing, power, contamination, decision rules and ownership.

How to use the template

Complete the Test plan sheet before assigning participants. Specify a primary outcome, follow-up period and business decision rule.
Check statistical power using historical data and your intended allocation. This workbook does not calculate sample size.
Randomly assign eligible individuals and retain everyone in their assigned group for the analysis. Check suppression and missing outcomes.
Enter group sizes and consistently defined revenue in the amber cells of Results. Use the same currency and period for revenue and spend.
Review the arithmetic alongside your experiment's uncertainty and any deviations before changing a budget.

Worked example: a 10,000-person test and 2,000-person control

The fictional test group earns $33,600 and the control earns $5,600. Control revenue per person is $2.80. Applied to 10,000 people, that implies $28,000 expected revenue without treatment. The estimated increment is $5,600, giving 0.70x iROAS on $8,000 spend.

Relative revenue lift is 20%, while the incremental share of test revenue is 16.7%. The denominators differ. At 60% contribution margin before ads, the example produces a $4,640 contribution loss after ad spend.

Try the same numbers in the online iROAS calculator, or follow the formula walkthrough.

Which tests does this support?

The results sheet uses a difference in revenue per assigned individual. It assumes randomized groups, comparable follow-up and no material interference between groups. It is not a matched-market or geo model. Revenue totals alone are insufficient for a confidence interval.

For regional experiments, use the geo-experiment guide and a design-specific analysis. The geo volume screening tool is an initial volume check, not proof that a particular regional design has enough power.

Can I use it in Excel or Google Sheets?

The download is an Excel workbook with ordinary formulas. You can import it into Google Sheets; check the example outputs after import. The workbook has been recalculated and its sample outputs checked during preparation. It contains no macros.

Can iROAS be negative?

Yes. If test revenue per person is lower than control revenue per person, the estimate is negative. Keep that result visible and investigate uncertainty and validity. A blank required input is shown as unavailable, not zero.