Plan a format test by writing what stays constant, what changes and what decision the comparison can support. Different formats may require different execution choices. Document those choices before interpreting the result.
The aim is to make a useful next decision without pretending the test isolates a cause that its design does not establish.
Build a message record, assign a communication job to the new format and retain a change list. Separate production acceptance from performance review, and limit the conclusion to the evidence available.
Write the message record
Capture the intended buyer, situation, supported promise, proof and action. Add the evidence that makes this angle worth exploring and the context of any prior result.
Avoid treating a successful asset as proof of one isolated ingredient. Its message, presenter, offer and delivery may all contribute to the outcome.
For a hypothetical consumer-product example, an existing execution could explain a verified feature. A new treatment might show that feature in an approved demonstration. This example contains no real product or campaign claim.
The message record gives the two briefs a shared boundary. It should state what information is supported and what remains a strategic interpretation, so production does not silently expand the promise.
State the format's communication job
Describe what the new treatment lets the buyer understand or inspect. An explanation can add context. A demonstration can make proof visible. A comparison can organize choices.
The job should connect to the buyer question. A format selected for novelty can still be tested, but name that rationale rather than inventing a proof purpose it does not serve.
Specify the required evidence and production resources. A demonstration may need approved conditions and new footage. A comparison may need exact verified attributes.
The format does not earn an unlimited claim boundary because it is visually compelling. Inspect the full intended execution, including what the sequence and imagery imply, before accepting the brief.
Make a constants-and-changes list
Write the elements you intend to preserve and the elements you change. Include message, proof, opening, structure, presenter, length and call to action where relevant.
Record surrounding variables such as offer, audience, placement and delivery too. The same approved promise can still appear in different testing conditions.
Several changes may be necessary to create the format. That is not automatically a defect. It limits how specifically you can explain the result.
Use the list in the review. If the new execution performs differently, the comparison may support a decision about the combined treatment. A causal statement about format alone needs evidence that can distinguish it from the other changes.
Build the new production specification
Translate the communication job into scenes or elements. Identify which part supplies proof, which part explains the question and which part provides the action.
Keep the exact approved wording and relevant conditions with the brief. Distinguish mandatory elements from choices left to production.
Ask whether the available footage and assets can meet the requirements. Founder-led execution can handle many treatments, but it still needs a clear task and accurate product information.
This production specification connects a strategic format hypothesis to usable creative work. References can suggest presentation mechanics. They do not replace the buyer evidence, approved claim or explanation of why the new version deserves to be made.
Agree how the result will be reviewed
Define the objective and relevant outcomes with the responsible growth team. Record what evidence and delivery context are needed for a decision and which conditions would be inconclusive.
Use account-specific criteria instead of an unsupported universal benchmark. Preserve attribution and measurement limits.
The possible next actions might include continuation, a focused revision, stopping or further investigation. State how those actions relate to the question you are exploring.
Acceptance should first establish that the asset meets the approved brief. Performance review follows distribution. Keeping those stages separate makes it easier to identify whether an issue concerns execution fidelity, test context or the broader strategic hypothesis.
Write a conclusion the design can support
After the result, record the observation, interpretation and next action. Compare the executed version with the original change list. Note any unplanned differences.
If the evidence supports only a broader execution decision, say so. You can choose to explore the treatment further without claiming it proves a universal format advantage.
If the result is too limited, preserve the uncertainty and the practical decision about resources. The next brief should state what changes because of the review.
This gives the team a useful account of the test. It also keeps the next strategist from treating a complex comparison as a simple rule that every future creative should follow.
A worked example: make the handoff concrete
A fictional brand compares a presenter explanation with a tabletop demonstration of the same verified feature. The demonstration also changes length, visual sequence and proof visibility. The comparison can explore the combined treatment. It cannot attribute every observed difference to format alone without evidence capable of separating those changes.
The planning sheet names the common buyer question and approved claim, then lists each changed element. The growth owner separately records the distribution design and intended outcome. Production signs off the accepted versions against their briefs before the test begins. This prevents an unapproved execution difference from disappearing inside a “format test” label.
| Decision field | Practical instruction | Check before handoff |
|---|---|---|
| Message and claim | Keep approved wording consistent | Record any deviation |
| Proof and treatment | Specify the communication job | Name required changes |
| Distribution | Record audience, offer and placement | Inspect actual delivery |
| Conclusion | Match specificity to the design | Retain alternative explanations |
Choose the conclusion language in advance
Write two or three possible review statements before launch. One may support continuing the combined treatment. Another may call for an isolated follow-up on proof visibility. A third may say the evidence is insufficient because delivery or measurement did not answer the question. These are decision categories, not promises about the outcome.
Afterward, compare the executed assets and actual campaign conditions with the plan. Record unexpected differences even if the result looks encouraging. If several factors changed, retain that fact beside the selected action. This lets the next strategist use the finding without inheriting a universal claim that the test never established.
Current reference: record the actual experiment conditions
Google Ads’ current custom-experiment documentation explains how a trial compares with an original campaign and how traffic and budget allocation form part of setup. It also warns that changes to the original do not automatically carry into the experiment. That supports recording the actual comparison context. The test-card and review methods here are original operating guidance, not a universal Google Ads testing recipe or a claim about another platform. Read Google Ads’ custom-experiment guidance (checked October 16, 2026).
Expert Q&A
Do we have to change only one element?
No. Some creative treatments require several changes. Document them and match the strength of the conclusion to the comparison. A broader execution test can guide a next action. If you need to isolate a particular cause for a material decision, plan an appropriate follow-up with the responsible testing team.
Is matching the message enough for a fair comparison?
It helps define a common starting point, but delivery, offer, audience, proof and other execution elements may still differ. Record the relevant context. Avoid claiming a format effect solely because the wording remained similar. The design and available measurement determine what the result can support.
What if production changes the plan?
Update the change list and obtain the necessary review. A revised scene, proof or claim can affect the hypothesis. Preserve the exact accepted version so the result is evaluated against what actually ran. Do not continue describing the test as if the initial brief was followed unchanged.
How should we choose the next version?
Use the supported observation and unresolved question to define a specific change. Include the evidence and production instructions in the next brief. Avoid selecting a new treatment only to keep output moving when nobody can explain what it is intended to help the team decide.
Create a message record and change list before the next format test. Afterward, check whether your conclusion is as specific as the comparison can reasonably support.