# Signal Validation & Action Log

Ruff Justice | Recreated educational template

Separate an observation from a diagnosis, then turn the investigation into a next action.

## Observed signal
What changed? Include the source, date, and measurement window.

[Your notes]

## How it is produced
Whose actions count? Could automation, internal access, missing data, or a changed definition affect it?

[Your notes]

## Initial hypothesis
What might it mean for the customer? Keep this separate from fact.

[Your notes]

## Alternative explanation
What else could produce the same signal?

[Your notes]

## Corroborating evidence
What does the customer say? What other evidence agrees or conflicts?

[Your notes]

## Assessment and uncertainty
What do you currently believe, and what is still unknown?

[Your notes]

## Action, owner, review
What happens next, who owns it, and when will the assessment be revisited?

[Your notes]

## Use responsibly
This is a reasoning worksheet, not a validated predictive model. Do not turn its fields into universal risk weights without testing them in your own context.

Newly created for Ruff Justice. Not a former-employer document. Keep confidential information out of shared copies.

## When to use it
Use before escalating a health-score change or treating an activity metric as evidence of value.

## How to use it
1. Record exactly what changed and how the signal was generated.
2. Write down at least one alternative explanation and seek corroborating customer context.
3. Choose the next useful action, assign an owner, and record what would change your assessment.

## Fictional worked example
customer logins decline, but an automated report still supports a weekly planning meeting. Confirm that the intended recipients receive and use the report before deciding whether the account is disengaged.

## Related case
The Metric That Lied