Voice-of-Customer Summaries: 30-Day Executive Intelligence

```yaml
version: 1.3
name: voc_exec_brief
owners:
  analytics_owner: "J. Patel (Chief of Staff Analytics)"
  business_owners:
    support: "M. Green (Support Ops)"
    product: "A. Kim (PM Lead)"
    revops: "D. Ross (Revenue Ops)"
regions:
  - us-east-1
  - eu-west-1
residency:
  us-east-1: allow_datasets: ["CASES", "CALLS_US", "SFDC_ACTIVITIES"], retention_days: 365
  eu-west-1: allow_datasets: ["CASES_EU", "CALLS_EU", "SFDC_ACTIVITIES_EU"], retention_days: 365
sources:
  tickets:
    type: snowflake
    database: COMPANY_ANALYTICS
    schema: SUPPORT
    table: CASES
    id_field: CASE_ID
  calls:
    type: databricks
    path: dbfs:/voice/transcripts/
    id_field: CALL_ID
  sales_notes:
    type: salesforce
    objects: ["Task", "Note", "Event"]
    id_field: SFID
taxonomy:
  themes: ["product_quality", "onboarding", "billing", "integrations", "performance", "ux", "pricing"]
  rule_overrides:
    - if: {source: "tickets", field: "CASE_SUBJECT", contains: "invoice"}
      then: {theme: "billing"}
classification:
  model: "vpc_llm_v2"
  confidence_score_field: CONFIDENCE
  sentiment: ["negative", "neutral", "positive"]
  human_review_thresholds:
    high_severity: 0.80
    standard: 0.70
anomaly_detection:
  baseline_window_days: 56
  alert_thresholds:
    theme_change_pct: 0.25   # 25% spike vs. baseline
    negative_sentiment_pct: 0.15
  segment_keys: ["segment", "region", "product_area"]
outputs:
  warehouse_table: COMPANY_ANALYTICS.EXEC_INTEL.VOC_SUMMARIES
  bi_views:
    looker_view: exec_voc_brief
    powerbi_dataset: exec_voc_brief_ds
brief_slo:
  refresh_cadence: "4h"
  ready_by_local: "08:30"
  data_freshness_max_minutes: 60
escalation:
  rules:
    - name: enterprise_negative_surge
      match: {segment: "enterprise", sentiment: "negative"}
      when:
        theme_change_pct: ">=0.20"
      notify:
        email: ["chief.staff@company.com", "cpo@company.com", "cro@company.com"]
        cc: ["support.ops@company.com"]
      owner: "A. Kim"
      sla_hours: 24
      require_human_signoff: true
observability:
  prompt_logging: true
  decision_ledger_table: COMPANY_ANALYTICS.GOVERNANCE.DECISION_LOG
  model_version: "vpc_llm_v2.3"
  audit_trail_sink: "s3://company-audit/voc-logs/"
rbac:
  roles:
    exec_reader: ["VOC_SUMMARIES", "DECISION_LOG"]
    analyst_editor: ["VOC_SUMMARIES", "DECISION_LOG", "CLASSIFICATION_RULES"]
    reviewer: ["DECISION_LOG"]
quality_gates:
  min_precision: 0.85
  min_recall: 0.70
  acceptance_criteria:
    - condition: CONFIDENCE < 0.78
      action: "route_to_human_review"
    - condition: theme == "billing" and segment == "enterprise"
      action: "always_review"
```

{
  "title": "Voice-of-Customer Summaries: 30-Day Executive Intelligence",
  "published_date": "2025-11-18",
  "author": {
    "name": "Elena Vasquez",
    "role": "Chief Analytics Officer",
    "entity": "DeepSpeed AI"
  },
  "core_concept": "Executive Intelligence and Analytics",
  "key_takeaways": [
    "Ship a governed VoC executive brief in 30 days using your existing stack: Snowflake/BigQuery/Databricks + Salesforce + Looker/Power BI.",
    "Use one trusted taxonomy to combine tickets, call transcripts, and sales notes for fewer reconciliation loops and faster decisions.",
    "Governance is non-negotiable: prompt logging, RBAC, data residency, and a decision ledger for executive summaries.",
    "Anchor adoption with a weekly executive brief: what changed, why it changed, and what to do next."
  ],
  "faq": [
    {
      "question": "How do we prevent ‘AI hallucinations’ in executive summaries?",
      "answer": "Use explainable prompts, store confidence scores, and link every claim to source records in the decision ledger. Set human review thresholds for high-severity topics and enforce quality gates (precision/recall) before publishing."
    },
    {
      "question": "Can this run in our regions with strict residency?",
      "answer": "Yes. Processing and storage occur in-region (US/EU). RBAC and dataset allow-lists keep data in the right geography, with audit logs to prove it."
    },
    {
      "question": "Do we need new tools to stand this up?",
      "answer": "No. We rely on Snowflake/BigQuery/Databricks for data, Salesforce for frontline context, and Looker or Power BI for the brief. The rest is orchestration and governance you can run in your VPC."
    }
  ],
  "business_impact_evidence": {
    "organization_profile": "B2B SaaS, 1,200 employees, North America + EMEA, multi-product portfolio",
    "before_state": "Analysts spent ~22 hours/week reconciling VoC slides across Support, Product, and RevOps. No shared taxonomy; exec debates stalled decisions.",
    "after_state": "Daily executive VoC brief in Power BI with anomaly alerts by segment; decision ledger with evidence links and reviewer sign-offs.",
    "metrics": [
      "40% analyst hours returned from VoC assembly and reconciliation",
      "Decision time on emergent issues reduced from days to same-day",
      "Theme detection coverage >90% across top six categories",
      "Baseline-to-action window cut from 5 days to <24 hours"
    ],
    "governance": "Legal/Security approved because the rollout included prompt logging, RBAC by role and region, data residency controls, human-in-the-loop on high-severity narratives, and models never trained on client data."
  },
  "summary": "Your execs want a single, trusted Voice-of-Customer brief. Here’s a 30‑day plan to ship VoC summaries from tickets, calls, and sales notes—governed and audit-ready."
}