# Content Attribution Assumption Log

Record how a team connects content exposure to outcomes, including model rules, data gaps, inferred relationships, alternative causes, and uncertainty.

## Usage note

Use this log to make an attribution interpretation inspectable before it becomes a budget, optimization, or performance claim. It does not prove that content caused an outcome.

## How to use this template

1. Define the attribution question, asset, audience, exposure, outcome, period, and decision.
2. Record the model rule, data sources, identity logic, exclusions, and known blind spots.
3. Separate observed facts, model-assigned credit, causal evidence, and analyst inference.
4. Test plausible alternatives and document how sensitive the conclusion is to assumptions.
5. Choose proportionate wording and action, then assign a date or event for reevaluation.

## Blank template

### Question and measurement boundary

- **Log ID, analyst, and date:** [Enter]
- **Decision this analysis informs:** [Enter]
- **Content asset or program:** [Enter]
- **Eligible audience and exposure:** [Define]
- **Outcome event:** [Define]
- **Observation and attribution windows:** [Enter]
- **Unit of analysis:** [Person / Session / Account / Other]
- **Exact attribution claim under review:** [Write]

### Attribution method and evidence

- **Model and version:** [First touch / Last touch / Multi-touch / Incrementality / Other]
- **Credit assignment rule:** [Describe]
- **Data sources and extraction reference:** [List]
- **Identity and deduplication rules:** [Describe]
- **Observed association:** [State without causal wording]
- **Causal evidence available:** [Experiment / Natural test / None / Other]
- **Model-assigned contribution:** [Enter]
- **Reason this model fits the decision:** [Explain]

### Assumptions, alternatives, and action

- **Required assumptions:** [List]
- **Missing or untracked behavior:** [List]
- **Competing explanations:** [List]
- **Sensitivity checks and results:** [Enter]
- **Confidence:** [Low / Medium / High, with reason]
- **Permitted wording:** [Enter]
- **Wording to avoid:** [Enter]
- **Decision and owner:** [Enter]
- **Review trigger and date:** [Enter]