Executive Dashboards: Trust Indicators with Source Links

# trust_layer.yaml
version: 1.3
owners:
  primary_contact: "jane.park@company.com"
  backup_contact: "ops-analytics-oncall@company.com"
  escalation_channel: "#exec-metrics-trust"
regions:
  default: "us-east-1"
  data_residency: ["US", "EU"]
rbac:
  roles:
    - name: "ELT-Viewer"
      privileges: ["view_tiles", "view_lineage_panel"]
    - name: "Board-Viewer"
      privileges: ["view_tiles", "view_lineage_panel"]
    - name: "Data-Steward"
      privileges: ["edit_trust", "approve_anomaly_overrides"]
warehouses:
  primary: "snowflake://EDW_PROD"
  catalog_fallbacks:
    - "bigquery://corp-edw"
    - "databricks://unity-catalog/prod"
indicators:
  confidence_thresholds:
    green: ">=0.80"
    yellow: "0.60-0.79"
    red: "<0.60"
  freshness_slo_default: "2h"

kpis:
  - name: "Net Retention Rate"
    tile_id: "look:elt_overview.nrr"
    semantic_view: "nrr.view"
    source_links:
      explore: "look://explore/revops/nrr?filters=fq%3Dcurrent_qtr"
      warehouse_sql: "snowflake://EDW_PROD/FINANCE_MART/NRR.sql"
    ownership:
      owner: "revops_owner@company.com"
      on_call: "revops-analyst@company.com"
    freshness:
      last_load_field: "_ingest_ts"
      slo: "2h"
      breach_alert: "#revops-alerts"
    anomaly_bands:
      method: "seasonal_decompose"
      training_window_days: 540
      band_sigma: 2.2
    data_quality_tests:
      - name: "customer_dim_key_not_null"
        severity: "high"
      - name: "invoice_line_sum_matches_gl"
        severity: "high"
    confidence_score:
      inputs: ["dq_pass_rate", "freshness_ok", "anomaly_distance"]
      formula: "0.4*dq_pass_rate + 0.3*freshness_ok + 0.3*(1-anomaly_distance)"
    approval_workflow:
      required_for_narrative: true
      approvers: ["Data-Steward", "Finance-Controller"]
      evidence_url: "https://governance.company.com/ledger/nrr"

  - name: "Pipeline Coverage (Next 2Q)"
    tile_id: "pbi:elt_rev.pipeline_coverage"
    semantic_view: "pipeline_coverage.model"
    source_links:
      lineage_pane: "pbi://lineage/elt_rev/pipeline_coverage"
      warehouse_sql: "bigquery://corp-edw/revops/pipeline_coverage.sql"
    ownership:
      owner: "sales_analytics@company.com"
      on_call: "oncall-analytics@company.com"
    freshness:
      last_load_field: "_loaddate"
      slo: "4h"
      breach_alert: "#sales-ops-alerts"
    anomaly_bands:
      method: "prophet"
      training_window_days: 365
      band_sigma: 2.0
    data_quality_tests:
      - name: "sf_account_fk_valid"
        severity: "medium"
      - name: "opp_stage_in_enum"
        severity: "medium"
    confidence_score:
      inputs: ["dq_pass_rate", "freshness_ok", "anomaly_distance"]
      formula: "0.5*dq_pass_rate + 0.2*freshness_ok + 0.3*(1-anomaly_distance)"
    approval_workflow:
      required_for_narrative: false
      approvers: ["Data-Steward"]
      evidence_url: "https://governance.company.com/ledger/pipeline_coverage"

  - name: "Headcount Actuals vs Plan"
    tile_id: "look:elt_ops.hc_actuals_vs_plan"
    semantic_view: "hc_actuals.view"
    source_links:
      explore: "look://explore/people/hc_actuals?drill=dept"
      warehouse_sql: "databricks://unity-catalog/prod/hr/hc_actuals.sql"
    ownership:
      owner: "people_analytics@company.com"
      on_call: "hris-oncall@company.com"
    freshness:
      last_load_field: "_update_ts"
      slo: "24h"
      breach_alert: "#people-ops-alerts"
    anomaly_bands:
      method: "rolling_mad"
      training_window_days: 730
      band_sigma: 3.0
    data_quality_tests:
      - name: "workday_worker_id_not_null"
        severity: "high"
      - name: "dept_mapping_complete"
        severity: "low"
    confidence_score:
      inputs: ["dq_pass_rate", "freshness_ok", "anomaly_distance"]
      formula: "0.4*dq_pass_rate + 0.4*freshness_ok + 0.2*(1-anomaly_distance)"
    approval_workflow:
      required_for_narrative: true
      approvers: ["Data-Steward", "HR-Controller"]
      evidence_url: "https://governance.company.com/ledger/hc_actuals"

narrative_settings:
  show_citations: true
  min_confidence_for_publish: 0.80
  prompt_logging: true
  never_train_on_client_data: true
observability:
  metrics:
    - name: "tile_clicks"
    - name: "time_to_answer_minutes"
    - name: "freshness_breaches"
    - name: "anomaly_overrides"
  owners_report: "https://analytics.company.com/reports/trust-weekly"

{
  "title": "Executive Dashboards: Trust Indicators with Source Links",
  "published_date": "2025-11-26",
  "author": {
    "name": "Elena Vasquez",
    "role": "Chief Analytics Officer",
    "entity": "DeepSpeed AI"
  },
  "core_concept": "Executive Intelligence and Analytics",
  "key_takeaways": [
    "Trust indicators on every KPI tile convert skepticism into clicks and conversations.",
    "Lineage, freshness SLOs, and owner attribution cut executive decision latency dramatically.",
    "Instrument a semantic trust layer across Looker/Power BI with Snowflake/BigQuery/Databricks sources.",
    "Adoption rises when every chart links to the source explore, SQL, and owner—governed with RBAC and audit trails.",
    "A 30‑day audit → pilot → scale motion delivers measurable usage gains and fewer escalations."
  ],
  "faq": [
    {
      "question": "How do we avoid exposing raw SQL to execs while still providing transparency?",
      "answer": "Use read‑only links to the lineage pane and a human‑readable semantic view map. The SQL path is accessible to stewards; execs see the Explore and owner name."
    },
    {
      "question": "What if two teams own similar metrics?",
      "answer": "Designate a single authoritative KPI per area with a clear owner; allow alternative cuts as ‘views’ that inherit lineage but are not ELT‑level KPIs."
    },
    {
      "question": "Will this slow down dashboards?",
      "answer": "No. Trust metadata is cached alongside semantic definitions. We track query latency and only fetch lineage panel details on demand."
    }
  ],
  "business_impact_evidence": {
    "organization_profile": "Public SaaS company (~1,800 FTE) with Snowflake + Looker; Salesforce and Workday as systems of record.",
    "before_state": "Exec dashboards under‑used; conflicting NRR definitions; analysts fielded 20+ Slack threads/week asking for sources; decisions deferred to ad‑hoc spreadsheets.",
    "after_state": "Trust badges, lineage, and source links added to 26 ELT KPIs; narratives included citations; anomalies routed to owners with a decision ledger.",
    "metrics": [
      "Decision time for KPI variances dropped from 2 days to 2 hours.",
      "Slack “what’s the source?” threads fell 58% within 6 weeks.",
      "Weekly ELT dashboard adoption (unique viewers) up 42%."
    ],
    "governance": "Legal/Security approved because all narratives log prompts and citations, RBAC enforced at the semantic layer, data residency honored (US/EU), and models never train on client data; decision overrides recorded with approver and reason."
  },
  "summary": "Chiefs of Staff: add trust badges, lineage, and source links to executive dashboards in 30 days—boost adoption and speed decisions with audit-ready signals."
}