A campaign learning log records what ran, how the result was measured, what you observed, what else could explain it, and what you will create or check next. It preserves the reasoning behind a decision without treating every change in a dashboard as proof of a winning creative.
Keep the observation attached to the next move.
The starting record is fictional. Replace it with your own evidence. Your entries stay in this browser session. Download the editable log before leaving; nothing is automatically saved or uploaded.
Import adds records without replacing the current ones. Maximum 50 records; JSON files up to 2 MB. Imported text is displayed as text and cannot change the page.
A worked record, without a winner claim.
| Fictional coffee example | What belongs in the record |
|---|---|
| Question | Does explaining the brewer-to-grind choice help a shopper decide? |
| Reported observations | Image A: 8 purchases from 200 recorded landing-page visits. Carousel B: 10 from 200. |
| Arithmetic | Observed rates are 4% and 5%, a difference of one percentage point. |
| What is unresolved | The event definition, attribution window, audience allocation, reporting maturity, and other changes must be checked. |
| Next creative instruction | Prepare a clear grind-choice explanation using verified product guidance. Treat its effectiveness as a hypothesis. |
These invented figures demonstrate record keeping, not Aitoma performance. The calculation does not establish statistical significance or show that the format caused the difference. A campaign reviewer still needs to evaluate the experiment design and uncertainty.
A useful next brief can follow an inconclusive observation. The important distinction is whether you are creating a candidate to test or repeating something supported by a suitable comparison. Write that distinction into the brief so it survives the handoff.
Record enough context to reopen the decision.
- Identify the exact creative versions and destination. A revised hook or changed product page can change what the comparison means.
- Record the account, platform, date range, timezone, report extraction time, and whether the work was paid or organic. Keep private identifiers in your own records.
- Define the event and denominator. Include the counting rules and attribution window where relevant, plus a reference to the source report.
- Separate the observation from the explanation. “More recorded purchases” describes a report; “the hook increased purchases” claims a cause.
- List other changes or missing data that limit the comparison. Assign an owner and the next review action rather than leaving a vague “check later.”
Choose an action your evidence can support.
| Review decision | What it means here |
|---|---|
| Observation only | Preserve a signal before deciding what it means. |
| Needs measurement review | Resolve definitions, tracking gaps, changed conditions or incomplete reporting before interpreting the result. |
| Next brief proposed | Prepare a creative instruction for review. This label does not certify the evidence or approve a budget change. |
| Closed with no change | Record why no new production work is justified, and what would trigger a fresh review. |
The log does not choose a winner, calculate statistical confidence, connect to advertising accounts, or recommend spend. You choose the review status. A complete-looking record can still contain an unsupported conclusion.
For organic content, record the post’s intended job and relevant audience response without treating it as a controlled ad experiment. Keep feedback about the production itself, such as a voice correction, separate from audience-performance evidence.
Write a brief someone can actually produce.
Replace “make more like B” with a specific instruction: who the piece is for, which question it answers, the proposed change, the product evidence to preserve, the format, and the next action. Link to the approved references and the current export being discussed.
For the coffee example: “Explain how a shopper matches a brewer to the store’s available grind options. Use the current product guide. Open with the brewer choice, show one readable comparison, and close on the actual selection page.” That is a production direction. Whether it improves results remains a question for an appropriate evaluation.
Keep the log with the brief and review the completed work against it. When new evidence changes the decision, add a new record with a reference to the earlier one. This preserves how the reasoning developed instead of silently rewriting history. When the next brief is ready, Aitoma’s image and video ad creative workflow shows the inputs and review steps for producing it. The campaign test and spending decision remain separate.
Use the right reporting definitions.
Google Ads recommends a clear hypothesis, a limited experimental change, success metrics chosen before the run, and retained experiment records. These principles inform the record structure; this worksheet does not replace the platform’s experiment analysis. Google Ads experiment guidance.
In Google Ads, a conversion window determines how long after an ad interaction a conversion can be recorded. Record the relevant setting when comparing reports; a shared label alone does not ensure matching definitions. Google Ads conversion-window guidance.
Sources reviewed October 19, 2026. Other platforms have their own reporting definitions. The fictional example and editable worksheet are Aitoma’s explanatory materials, not a platform-endorsed methodology.