# One appeal test, ready to run

Use this sheet to agree on an audience, two messages and a review before sending. Keep normal receipts, acknowledgements and answers to donor questions in place.

## The decision

Decision we need to make:
Audience and why this question matters to them:
Research source and the part that applies here:
One change between A and B:
What the source measured, and what we will measure:

## People and dates

Campaign owner:
Data and assignment owner:
Message/fact reviewer:
Email or mail operations owner:
Who answers replies and by when:
Send date and time:
Fixed observation window:
Review date:

## Eligibility and assignment

Assignment unit (one person, household or another defined unit):
Eligible population and date frozen:
Segments, defined before assignment:
Exclusions (contact preferences, recent gifts, scheduled payments, existing conversations, duplicate identities):
How we will assign comparable units at random within each segment:
Where the original A/B assignment and exact message versions will be kept:
How we will prevent the same unit receiving both versions:

Honor any new stop-contact request immediately. Keep that person's original assignment in the report, with delivery/contact status recorded separately. Do not reclassify people by whether they replied or gave after sending.

## The two messages

A, current version:

B, proposed version:

Facts verified, by whom and on what date:
What a donor's choice does (gift designation, planning input or future updates):
Any contact promise and its exact end date:
Campaigns, reminders and automated sequences checked against that promise:
Operations owner's confirmation:

## Outcomes, agreed before sending

Primary outcome: net donation revenue per originally assigned eligible unit during the fixed window.
Qualifying gifts and attribution/counting rule:
Costs and fees included in net revenue:
Number of gifts, units with a gift and giving rate (report separately):
Unsubscribe, complaint and stop-contact measures, with denominators:
The condition that will make us pause and who acts:
What we will do if results are too sparse or mixed to support a decision:
Who will assess uncertainty before a wider rollout:

| Segment | Version | Originally assigned | Delivered | Units with a gift | Net gift revenue | Unsubscribes | Complaints/stop requests |
| ------- | ------- | ------------------: | --------: | ----------------: | ---------------: | -----------: | -----------------------: |
| Recent  | A       |                     |           |                   |                  |              |                          |
| Recent  | B       |                     |           |                   |                  |              |                          |
| Older   | A       |                     |           |                   |                  |              |                          |
| Older   | B       |                     |           |                   |                  |              |                          |

Giving rate = assigned units with at least one qualifying gift ÷ originally assigned eligible units. Count each unit once, even if it makes several gifts. In an individual-donor test, one unit is one donor; in a household test, one unit is one household.
Net revenue per unit = net gift revenue ÷ originally assigned eligible units.
Relative giving-rate change = (B rate − A rate) ÷ A rate; undefined if A is zero.
Percentage-point change = 100 × (B rate − A rate), with rates entered as fractions.
Record complaints and unsubscribe rates separately from money. A reply about future giving is a stated intention, not a pledge or received gift.

## The review

What happened by segment:
Do a few large gifts explain the apparent difference:
Any deliverability, identity or version problem:
Any preference or contact promise breached, and what we changed:
Decision (keep A, test again, limited rollout after review, or stop):
Evidence for that decision, including uncertainty:
Next action, owner and date:

_Companion: [Before You Borrow the Winning Email](https://gratona.com/atlas/reading?source=before-you-borrow-the-winning-email&page=kit)._