# ChatGPT Ads test plan

Advarde · Editable template · 11 September 2026

Test name / version:
Account ID / currency / timezone:
Owner / approver:
Start / review / end dates:

## Hypothesis

We believe [customer situation] will respond to [supported offer / reason to buy].
We will evaluate [primary business outcome] using [source and definition].
The result will inform [specific next decision].

## Economics and limits

Contribution per customer:
Click-to-outcome rate assumption:
Lead qualification / close rate assumption (if applicable):
CPC assumption:
Media budget ceiling:
Other acquisition costs included:
Break-even CPC calculation:
Downside scenario:
Assumptions source / date:

Expected outcomes are scenario averages, not guaranteed counts.

## Design

| Element | Definition |
| --- | --- |
| Offer and destination | |
| Geography and eligibility | |
| Campaign objective | |
| Context groups | |
| Creative A / version | |
| Creative B / version | |
| Variables intentionally different | |
| Variables held consistent | |
| Primary metric and denominator | |
| Supporting diagnostic metrics | |
| Attribution definition | |
| Follow-up period for customer quality | |

## Readiness

- [ ] Advertiser access and billing checked.
- [ ] Successful business action measured correctly.
- [ ] Failed action and retry / refresh cases tested.
- [ ] Creative, destination, units, and budget reviewed.
- [ ] Exact draft version approved.
- [ ] Created objects inspected while paused.
- [ ] Activation decision and monitoring owner recorded.

## Decision rules written before launch

Continue if:
Revise and retest if:
Pause immediately for repair if:
Stop this hypothesis if:
Insufficient evidence if:
Maximum spend / time before another review:

## Illustrative planning example — no campaign results

Question: Can a studio consultation offer produce qualified bookings affordably?
Contribution per customer: $800.
Qualified-lead close rate assumption: 10%.
Click-to-qualified-lead assumption: 4%.
Implied break-even CPC before other costs: $800 × 10% × 4% = $3.20.
The test must define what “qualified” means and record follow-up. These figures are assumptions, not an Advarde benchmark.

## Review record

Actual dates / spend / clicks:
Completed outcomes and qualification:
Tracking or delivery issues:
Changes during the test:
What the evidence supports:
What remains uncertain:
Decision / approver:
Next action / owner / date:

Reference: https://advarde.com/templates/chatgpt-ads-test-plan