The Incrementality Test Design Template

Test Name: [SHORT DESCRIPTIVE NAME, e.g. "Meta Q3 Incrementality Test"]

Business Question Being Tested: [Write the exact question in plain English. Example: "Does running Meta ads cause more people to buy our [PRODUCT/SERVICE], or do those buyers find us anyway through organic search, word of mouth, or direct?"]

Channel Being Tested: [e.g. Meta Ads / Google Search / LinkedIn / Email]

Campaign or Ad Set Name: [Exact campaign name in your ad account]

Test Start Date: [DD/MM/YYYY]

Test End Date: [DD/MM/YYYY]

Minimum Test Duration: [Recommended: at least 14 days. Note here if your purchase cycle is longer and adjust accordingly.]

Person Responsible for Running This Test: [NAME]

Person Responsible for Reading the Results: [NAME]

Baseline Period: [DD/MM/YYYY to DD/MM/YYYY] (use the same length of time as your test period, from directly before it starts)

Baseline Revenue From This Channel: [£/$ AMOUNT over baseline period]

Baseline Conversion Volume: [NUMBER of purchases / sign-ups / leads over baseline period]

Baseline Conversion Rate: [% calculated from your own data, not platform benchmarks]

Baseline Average Order Value or Deal Size: [£/$ AMOUNT]

Organic and Direct Conversion Volume During Baseline (no paid ads running): [NUMBER]

Notes on Anything That Was Different During Baseline: [e.g. seasonal events, promotions, website changes, competitor activity that could skew the comparison]

Data Source Used for Baseline: [e.g. Google Analytics 4, HubSpot, Shopify, CRM. Be specific. Do not use platform-reported numbers as your baseline.]

Test Group Definition (exposed to ads):
[Describe who will be in the test group. Example: "All users in [GEOGRAPHIC REGION or AUDIENCE SEGMENT] matching [TARGETING CRITERIA]. Estimated size: [NUMBER of people]."]

Control Group Definition (not exposed to ads):
[Describe who will be withheld from ads. Example: "A [X]% holdout of the same audience, excluded at the ad account level using [PLATFORM HOLDOUT FEATURE / GEO SPLIT / MANUAL EXCLUSION LIST]." Estimated size: [NUMBER of people].]

Control Group Size as % of Total Audience: [Recommended: 10 to 20%. Note your choice here and why.]

Method Used to Create the Control Group: [e.g. Meta Conversion Lift holdout / Geo split / CRM suppression list / Platform-level audience exclusion]

Risk of Contamination Between Groups: [Describe any ways the control group might still see your ads, e.g. via a partner, retargeting, or organic social. State what you have done to reduce this.]

Are Both Groups Exposed to the Same Seasonal Conditions? [YES / NO. If NO, explain.]

Primary Metric: [The one number that determines whether this test is a pass or a fail. Example: "Number of confirmed purchases attributable to incremental lift."]

Secondary Metrics (optional, do not let these override the primary): [e.g. Revenue per user in test vs control / Cost per incremental conversion / ROAS calculated on incremental revenue only]

Minimum Detectable Lift to Be Considered Meaningful: [X%. Choose a number that would actually change your spend decision. If 5% lift would not change anything, set the bar higher.]

Lift Calculation Formula:
Incremental Lift = (Conversion Rate in Test Group) minus (Conversion Rate in Control Group)
Incremental Revenue = Incremental Lift x [TOTAL AUDIENCE SIZE] x [AVERAGE ORDER VALUE or DEAL SIZE]

Target Incremental ROAS (the ROAS calculated only on incremental revenue, not total attributed revenue): [X.Xx]

Note: Platform-reported ROAS includes conversions your control group would have made anyway. Incremental ROAS is always lower. If incremental ROAS drops below [YOUR THRESHOLD], the campaign is not earning its budget.

Statistical Confidence Threshold Required Before Acting on Results: [Recommended: 90% minimum. Note your threshold here.]

Planned Promotions or Sales During the Test Period: [LIST ANY. Promotions inflate both groups and can mask or exaggerate lift. If a promotion is unavoidable, note how you will account for it.]

Seasonal Events That Overlap With the Test: [e.g. bank holidays, school terms, industry events, competitor launches]

Website or Product Changes Planned During the Test: [e.g. pricing changes, new landing pages, checkout updates. If possible, freeze these during the test.]

Other Active Campaigns Running in Parallel: [LIST ALL. Cross-channel exposure can make it impossible to isolate this channel's effect. Note how you will separate them.]

Data Collection Method and Verification: [Where will you pull results from? How will you verify the data is not double-counting or pulling from the platform's own attribution model?]

Test Pause Conditions: [Describe the specific conditions under which you would pause or cancel the test early. Example: "If spend in the test group exceeds [£/$X] with zero incremental conversions recorded by day 7, pause and review."]

Sign-Off Required Before Pausing or Extending the Test: [NAME and ROLE]

Pre-Committed Decision Rules (fill in BEFORE the test ends):

IF incremental lift exceeds [X%] AND incremental ROAS exceeds [X.Xx] THEN: [STATE YOUR ACTION. e.g. "Scale budget by [X]% and expand to [NEW AUDIENCE or REGION]."]

IF incremental lift is between [X%] and [X%] AND incremental ROAS is between [X.Xx] and [X.Xx] THEN: [STATE YOUR ACTION. e.g. "Hold budget flat and run a second test with revised creative and targeting."]

IF incremental lift is below [X%] OR incremental ROAS is below [X.Xx] THEN: [STATE YOUR ACTION. e.g. "Reduce budget by [X]% and reallocate to [CHANNEL] where lift has been demonstrated."]

---
Fill in after the test:

Test Group Conversion Rate: [X%]
Control Group Conversion Rate: [X%]
Measured Incremental Lift: [X%]
Statistical Confidence Level Achieved: [X%]
Incremental Revenue Generated: [£/$ AMOUNT]
Incremental ROAS: [X.Xx]
Platform-Reported ROAS (for comparison): [X.Xx]
Difference Between Platform ROAS and Incremental ROAS: [X.Xx]

Decision Made Based on Results: [WRITE THE DECISION EXACTLY AS STATED IN YOUR PRE-COMMITTED RULES ABOVE]

Anything That Compromised Test Integrity: [DESCRIBE or write "None identified"]

Next Test Recommended: [DATE and BRIEF DESCRIPTION]