Define the question and hold the offer steady
A format test compares two or more ways to communicate a similar message. Write a narrow question such as “Can viewers understand this feature more clearly in a reaction-plus-demo or a text-led video?” Keep the product fact, audience and intended next action as comparable as practical. If you change the offer, audience and message at the same time, a difference in results cannot be attributed to format alone.
This is a planning example, not a completed experiment. The fictional feature is a project call note with a next-step section.
Create the two concepts
Version A: a short reaction hook says “the next step is right here,” followed by a real recording that opens the note and shows its section. Version B: a wall-of-text overlay describes the same task and supported feature over appropriate footage. Each version should communicate the action accurately and use assets you have permission to publish.
Keep a simple record of exact wording, duration, account, destination and dates. The versions do not need identical editing, but differences that matter should be documented. Do not call these drafts completed or claim a winner before actual evidence exists.
| Item | Reaction plus demo | Wall-of-text |
|---|---|---|
| Product fact | Same verified feature | Same verified feature |
| Opening | Short reaction tied to the feature | Written situation and feature |
| Proof | Real app recording | Supported text and suitable footage |
| Measure | Recorded platform definition | Same measure if available |
| Decision | Interpret with audience context | Do not claim a winner from sparse data |
Decide what you will observe
Before publishing, select a primary question and the platform’s available measure. For comprehension, a small review with unfamiliar viewers can ask what the app does after one viewing. For platform performance, record the platform-defined views and available retention or engagement metrics, together with the audience and distribution context. Metric definitions vary, and views do not equal customers or revenue.
Avoid setting an arbitrary universal threshold. Decide in advance what evidence would justify another iteration, and record sample size. Organic distribution, timing and unequal audiences can make comparisons noisy.
Interpret and iterate cautiously
After the agreed observation window, record both results and the conditions. If data is sparse or one format reached a different audience, describe the outcome as inconclusive rather than declaring a winner. If reviewers misunderstand the same feature in both versions, improve the explanation before testing another format.
Choose one next revision and keep the previous scripts and exports for comparison. Do not invent numbers or assume a format that worked for another account will work for yours. Secure any permissions and approvals before actual publication.
Copy the planning worksheet
Use this worksheet in your own notes. Fill it with verified product details and actual review observations before sharing an asset.
Question: __. Shared feature: __. Version A: __. Version B: __. Audience and destination: __. Measure and observation window: __. Limitations: __.Browse the founder guide library for the next task in your workflow.