Frameworks, templates, and operating models I use as a technical PM working across data infrastructure, BI, and AI tooling in warehouse operations.
Most PM templates are written for consumer product managers. This repo is for PMs who work at the intersection of data engineering, operations, and AI — where the "product" is often an internal tool, a BI system, or an automated workflow.
These are the templates I actually use at ShipMonk, adapted from practice across:
- B2B wholesale operations product work
- Data infrastructure and BI ownership
- AI tooling evaluation and deployment
- Cross-functional ops/engineering collaboration
Use this when scoping a new BI dashboard, data pipeline, or internal analytics tool.
# [Product Name] — Data Product PRD
**Owner**: [PM Name]
**Status**: Draft / In Review / Approved
**Last Updated**: [Date]
---
## Problem Statement
What operational or business question is currently unanswered or answered poorly?
Who is affected, and what decisions are blocked as a result?
---
## Success Metrics
| Metric | Baseline | Target | How Measured |
|---|---|---|---|
| [e.g. Time to produce weekly report] | [e.g. 2 hours] | [e.g. 10 minutes] | [e.g. Team survey] |
| [e.g. Data freshness] | [e.g. T+2 days] | [e.g. T+0 daily] | [e.g. Metabase timestamp] |
---
## Users & Stakeholders
| Role | Need | Usage Frequency |
|---|---|---|
| [e.g. Ops Director] | [e.g. Weekly scorecard review] | [e.g. Weekly] |
| [e.g. Site Lead] | [e.g. Daily exception management] | [e.g. Daily] |
---
## Data Requirements
**Source systems**:
- [ ] System name — table/entity — freshness requirement
- [ ] System name — table/entity — freshness requirement
**Grain**: What is one row in the output?
**Filters / Parameters**: What should the user be able to filter by?
**Calculated fields**: List any non-trivial derived metrics with their formulas.
---
## Functional Requirements
1. [Requirement 1]
2. [Requirement 2]
3. [Requirement 3]
## Non-Functional Requirements
- **Refresh cadence**: [Daily / Weekly / Real-time]
- **Access control**: [Who should see this?]
- **Alerting**: [Should this trigger alerts? On what condition?]
---
## Open Questions
- [ ] [Question 1 — Owner — Due Date]
- [ ] [Question 2 — Owner — Due Date]
---
## Out of Scope
[Explicitly list what this product will NOT do, to prevent scope creep]Use this when investigating a financial discrepancy, billing gap, or EDI compliance failure.
# RCA: [Incident Name]
**Investigator**: [Name]
**Date Opened**: [Date]
**Date Closed**: [Date]
**Financial Impact**: $[Amount]
---
## Incident Summary
[2-3 sentences: what happened, who was affected, how it was discovered]
---
## Affected Records
| PO / Order # | Merchant | Warehouse | Violation / Issue | Amount |
|---|---|---|---|---|
---
## Timeline of Events
| Timestamp | Event | System | Notes |
|---|---|---|---|
---
## Root Cause
**Primary root cause**: [One sentence]
**Contributing factors**:
1. [Factor 1]
2. [Factor 2]
**Root cause classification**:
- [ ] Fulfillment delay
- [ ] EDI / notification system failure
- [ ] Configuration error
- [ ] Process gap
- [ ] Third-party failure
---
## Corrective Actions
| Action | Owner | Due Date | Status |
|---|---|---|---|
---
## Prevention
What systemic change would prevent this class of issue from recurring?
---
## Data Systems Investigated
| System | Database | What Was Checked |
|---|---|---|Use this before deploying an AI tool (Claude, Gemini, GPT, etc.) into an internal workflow.
# AI Tool Evaluation: [Tool Name] for [Use Case]
**Evaluator**: [Name]
**Date**: [Date]
---
## Use Case Summary
What workflow is this tool being applied to?
What is the current state (manual, semi-automated, etc.)?
What is the target state?
---
## Evaluation Criteria
| Criteria | Weight | Score (1-5) | Notes |
|---|---|---|---|
| Output quality | 30% | | Does the AI output require significant editing? |
| Reliability | 25% | | Does it produce consistent results on similar inputs? |
| Latency | 15% | | Is response time acceptable for the workflow? |
| Integration effort | 15% | | How hard is it to embed in the existing workflow? |
| Cost | 15% | | What is the per-use cost vs. time saved? |
---
## Test Cases
| Input | Expected Output | Actual Output | Pass/Fail | Notes |
|---|---|---|---|---|
---
## Prompt Engineering Notes
What prompt structure produced the best results?
What constraints or output format instructions were needed?
What edge cases caused failures?
---
## Recommendation
- [ ] Deploy as-is
- [ ] Deploy with modifications: [describe]
- [ ] Pilot with limited users first
- [ ] Do not deploy: [reason]
---
## Risks & Mitigations
| Risk | Likelihood | Impact | Mitigation |
|---|---|---|---|Standard agenda structure for a multi-site warehouse operations sync.
# Weekly B2B Ops Sync — [Date]
## SQCD Tips
| Safety | |
| Standard Work | |
| Delivery / Cost | |
| Success Stories | |
## Special Project Update
[Owner]
## Previous Meeting Items
[Carry-forward from last week]
## Network Scoreboard
- OTRS: [W-1 % vs W-2 %]
- OTS: [W-1 % vs W-2 %]
- Pick Audit: [W-1 % vs W-2 %]
- Pack Audit: [W-1 % vs W-2 %]
- Aging Orders: [Count by warehouse]
- Put-Backs: [Count by warehouse]
## Building Headlines
*What is red or at risk? What help is needed?*
## Network Risk This Week
*At-risk customers, volume spikes, inventory or system risks*
## Change Announcements
*New accounts, system or process changes impacting operations*
## Help Requests & Commitments
*Owner | Item | Due Date*Use this to manage incoming data requests from ops, finance, and commercial teams.
# Data Request Triage
| Field | Value |
|---|---|
| **Requester** | [Name / Team] |
| **Request Date** | [Date] |
| **Business Question** | [What decision does this answer?] |
| **Urgency** | High / Medium / Low |
| **Data Source(s)** | [Snowflake / Metabase / ACCI / Other] |
| **Estimated Effort** | [Hours] |
| **Output Format** | Dashboard / Ad-hoc query / Export / Automated report |
| **Recurrence** | One-time / Weekly / Monthly |
| **Owner** | [Analyst / PM] |
| **Due Date** | [Date] |
| **Status** | Backlog / In Progress / Done / Blocked |- PRD template → Every new Metabase card or data pipeline I scope starts here
- RCA template → Used for every chargeback investigation and billing gap finding
- AI evaluation framework → Used before deploying Claude or Gemini into any automated workflow
- Ops meeting structure → The backbone of the weekly B2B wholesale ops sync (automated via
ai-ops-tooling) - Data request triage → Running backlog in Asana for data/analytics requests across ops teams
warehouse-ops-intelligence— The BI system these templates governb2b-chargeback-rca-framework— The RCA process documented in template #2ai-ops-tooling— The AI tools evaluated using template #3