# Learning Loop Design

> Designs feedback loops, indicators, review cadence, and decision checkpoints to determine whether a system intervention is working. Use after recommendations, pilots, controls, governance changes, process repairs, or technology changes where sustained improvement must be monitored.

- Skill: `bkjohn2018/learning-loop-design` (Agent Skill)
- Install (CLI): `npx skillmds@latest add bkjohn2018/learning-loop-design`
- Raw SKILL.md: https://api.skillmd.com/api/skills/bkjohn2018/learning-loop-design/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: bkjohn2018 (https://skillmd.com/u/bkjohn2018)
- Updated: 2026-09-21
- Page: https://skillmd.com/skills/bkjohn2018/learning-loop-design

---


# Learning Loop Design

## When to Use
- A recommendation needs follow-up evidence to confirm improvement.
- A pilot, control, governance change, process repair, or technology change must be monitored.
- The issue is recurring, systemic, high-risk, or tied to adoption and sustainability.

## Critical Governance Rule
If the intervention affects controls, governance, compliance, audit evidence, financial reporting, regulated data, formal ownership, approvals, or risk acceptance, define monitoring, evidence, review cadence, and decision checkpoints before considering the recommendation complete.

Policies, procedures, controls, standards, processes, and governance artifacts may evolve, but changes should be routed through the appropriate lifecycle, approval, evidence, and ownership process rather than introduced ad hoc during diagnosis. Use `finance-documentation-lifecycle` when governed document changes require controlled revision, publication, versioning, retention, or retirement.

## Human-in-the-Loop Control Gate
If a learning loop, checkpoint, or recommendation may create, revise, retire, reinterpret, or materially affect a control or governance artifact, do not proceed on assumptions. Request explicit user instruction or affirmation before drafting or changing policy, procedure, process, control, ownership, approval, evidence, or governance language. If intent is ambiguous, ask for clarification.

## Workflow
1. Define the expected behavior or system change.
2. Select leading indicators that show early movement.
3. Select lagging indicators that show durable outcome improvement.
4. Define stakeholder feedback loops.
5. Define data quality or process measures where relevant.
6. Set review cadence and decision checkpoints.
7. State what to revisit if evidence contradicts the diagnosis.

## Output Template
```markdown
## Learning Loop
- Expected behavior change:
- Leading indicators:
- Lagging indicators:
- Stakeholder feedback:
- Data quality or process measures:
- Review cadence:
- Decision checkpoints:
- Revisit triggers:
```

## Quality Checklist
- Measures are tied to the intervention's intended effect.
- Leading and lagging indicators are distinct.
- Review cadence is realistic for the decision horizon.
- Governance or control changes include evidence and monitoring expectations.