Metrics Dashboard Workflow
This skill provides a structured workflow for designing KPI dashboards using the metrics-analyst agent.
Workflow Overview
The metrics dashboard workflow follows 5 phases:
| Phase |
Name |
Description |
| 1 |
Requirements |
Define dashboard objectives and audience |
| 2 |
KPI Design |
Define metrics and methodology |
| 3 |
Data Architecture |
Map data sources and calculations |
| 4 |
Visualization |
Design visual presentation |
| 5 |
Implementation |
Build and validate |
Phase 1: Requirements
MANDATORY: Define dashboard objectives before building
Questions to Answer
| Question |
Purpose |
| What decisions will this support? |
Ensures relevance |
| Who is the primary audience? |
Tailors complexity |
| What frequency of update? |
Sets refresh requirements |
| What level of drill-down? |
Scopes depth |
| What benchmark comparisons? |
Defines targets |
Dashboard Types
| Type |
Audience |
Focus |
| Executive |
C-Suite |
High-level, strategic |
| Operational |
Managers |
Detailed, actionable |
| Departmental |
Department heads |
Function-specific |
| Board |
Directors |
Governance, strategic |
Blocker Check
If ANY of these are unclear, STOP and ask:
- Dashboard purpose
- Primary audience
- Key decisions supported
- Update frequency required
Phase 2: KPI Design
MANDATORY: Define all metrics with methodology
KPI Definition Standard
| Element |
Requirement |
| Name |
Clear, concise name |
| Definition |
Precise description |
| Formula |
Exact calculation |
| Unit |
Measurement unit |
| Target |
Performance target |
| Owner |
Accountable person |
| Frequency |
Update cadence |
KPI Categories
| Category |
Example KPIs |
| Financial |
Revenue, margin, EBITDA, cash flow |
| Operational |
Throughput, cycle time, utilization |
| Customer |
Retention, NPS, LTV, CAC |
| Growth |
ARR growth, customer growth, expansion |
KPI Selection Principles
| Principle |
Description |
| Relevance |
Supports specific decisions |
| Measurable |
Can be quantified objectively |
| Actionable |
Drives specific actions |
| Timely |
Available when needed |
| Owned |
Clear accountability |
Phase 3: Data Architecture
MANDATORY: Document data lineage completely
Data Source Mapping
| Element |
Documentation |
| Source system |
Where data originates |
| Extraction method |
How data is obtained |
| Transformation |
Any calculations or adjustments |
| Refresh frequency |
How often updated |
| Data quality |
Validation checks |
Data Quality Requirements
| Check |
Validation |
| Completeness |
All required data present |
| Accuracy |
Data matches source |
| Timeliness |
Data is current |
| Consistency |
Data consistent across sources |
Phase 4: Agent Dispatch
Dispatch to specialist with full context
Agent Dispatch
Task tool:
subagent_type: "metrics-analyst"
model: "opus"
prompt: |
Design metrics dashboard per these specifications:
**Purpose**: [from Phase 1]
**Audience**: [from Phase 1]
**Update Frequency**: [from Phase 1]
**KPIs Required**:
[From Phase 2 - list with definitions]
**Data Sources**:
[From Phase 3 - source mapping]
**Required Output**:
- KPI definitions with formulas
- Data source documentation
- Calculation methodology
- Visualization specifications
- Anomaly thresholds
- Implementation guide
Required Output Elements
| Element |
Requirement |
| Metrics Summary |
Dashboard overview |
| KPI Definitions |
Complete definitions |
| Data Sources |
Source documentation |
| Calculation Methodology |
Formula details |
| Dashboard Design |
Visual specifications |
| Anomaly Analysis |
Threshold definitions |
| Recommendations |
Enhancement suggestions |
Phase 5: Implementation
MANDATORY: Validate before deployment
Implementation Checklist
| Check |
Validation |
| Data feeds working |
All sources connected |
| Calculations verified |
Outputs match expected |
| Visuals rendering |
Display correctly |
| Refresh working |
Updates as expected |
| Access controlled |
Right users have access |
Validation Tests
| Test |
Description |
| Data reconciliation |
Dashboard ties to source |
| Historical comparison |
Trends make sense |
| Edge cases |
Handles nulls, zeros |
| Performance |
Loads in acceptable time |
Pressure Resistance
See shared-patterns/pressure-resistance.md for universal pressures.
Dashboard-Specific Pressures
| Pressure Type |
Request |
Agent Response |
| "Just show the numbers" |
"Numbers without methodology cannot be trusted. I'll include documentation." |
|
| "Pick the most important KPIs" |
"KPI selection requires business input. Which decisions should these support?" |
|
| "Skip the data quality checks" |
"Unreliable data undermines dashboard value. I'll validate all sources." |
|
| "Copy the existing dashboard" |
"Each dashboard needs fresh design. I'll validate requirements." |
|
Anti-Rationalization Table
See shared-patterns/anti-rationalization.md for universal anti-rationalizations.
Dashboard-Specific Anti-Rationalizations
| Rationalization |
Why It's WRONG |
Required Action |
| "Everyone knows what revenue means" |
Definitions vary |
DEFINE specifically |
| "Data source is obvious" |
Lineage needs documentation |
DOCUMENT source |
| "Calculation is standard" |
Standard still needs documentation |
SHOW formula |
| "Refresh frequency doesn't matter" |
Stale data causes bad decisions |
SPECIFY frequency |
Execution Report
Upon completion, report:
| Metric |
Value |
| Duration |
Xm Ys |
| KPIs Defined |
N |
| Data Sources Mapped |
N |
| Visualizations Designed |
N |
| Anomaly Thresholds |
N |
| Result |
COMPLETE/PARTIAL |
Quality Indicators
| Indicator |
Status |
| All KPIs defined |
YES/NO |
| All sources documented |
YES/NO |
| All calculations shown |
YES/NO |
| Data validated |
YES/NO |
| Refresh tested |
YES/NO |
majiayu000