opsmarshal
Workflow 15 — RevOps Infrastructure

GTM Analytics Copilot

Natural-language questions over the revenue warehouse, answered only through governed metric definitions and with the SQL always shown — so “what's our CAC payback by segment” returns the number finance would sign, not a fluent guess.

At a glance

The GTM Analytics Copilot answers natural-language questions over the revenue warehouse only through governed metric definitions, and always shows the SQL, so the answer is a number finance would sign rather than a fluent guess. It is one of fifteen workflows in the Enterprise GTM Agent Stack, an Opsmarshal series on GTM stack architecture.

Owned by
RevOps + Analytics Engineering
Writes back to
Definition repo, Query + refusal log, Audit log
Held to
≥ 95% — Answer accuracy
Gate
Permission → Evidence → Approval → Audit
Read-only by designDefinitions change via PR onlyAnswers: SQL + definition cited
Business Pain

Why this exists

The status quo

Every GTM team runs on numbers that disagree: marketing's CAC, finance's CAC, and the board deck's CAC are three numbers. Self-serve BI made querying easy and consistency worse; LLM-over-SQL without governance makes confident wrongness instant.

What this workflow changes

The copilot answers only through the semantic layer. Ambiguous questions get a clarifying choice — “pipeline-influenced or closed-won attribution?” — instead of a silent assumption. Every answer shows its SQL and cites its definition version.

Architecture

Reference architecture

Inputs
Revenue warehousethe governed tables, access-scoped
Semantic layermetric definitions, PR-governed, versioned
Question streamand its refusal log — the coverage roadmap
Agent Layer
Intent Parsermaps questions to definitions; detects ambiguity
Query BuilderSQL through the semantic layer only
Answer Composernumber + SQL + definition citation, lineage on export
Governance
Gate: permission → evidence → approval → auditread-only against the warehouse · out-of-definition questions are refused with “define it first” · schema drift pauses the copilot automatically
Systems of Record
Definition repoPR flow, versioned, the single source of metric truth
Query + refusal logwhat was asked, answered, declined
Audit loganswer lineage for anything exported
DecisionChoiceRationale
Refusal is a featureUngoverned questions get “define it first,” not improvisationThe failure mode of analytics copilots is confident answers to ungoverned questions. Refusing cheaply beats retracting expensively.
Read-only, PR-governed definitionsThe copilot can't create metrics; humans version themMetric sprawl is how the three-CAC problem started. Definitions change through review, and the copilot always reflects the current merged truth.
Live Demo

Run a scenario through the gate

Three curated scenarios — one that passes, one that flags for a human, one that the gate blocks. Watch the verdict, the evidence rows, and the audit log.

Analytics Copilot

Governance audit log

[ready] agent idle — pick a scenario
Operator Control Panel

Someone has to run this thing

Agents don't remove operators — they change what operators watch. Console spec: who owns it, what they check, what pages them, and how they pull the plug.

GTM Analytics Copilot — Ops Console

Owner: RevOps + Analytics Engineering

Daily monitors

  • Query error rate and latency
  • Refusal log — what people ask that isn't governed is the roadmap

Weekly reviews

  • Answer spot-audit — sampled answers vs. hand-written SQL
  • Definition coverage growth — refusal rate should trend down

Alert thresholds → who gets paged

  • Semantic-layer / warehouse schema drift → sev-1: copilot auto-pauses
  • Answer latency degradation → Analytics Eng

Manual controls

  • Definition repo — PR flow with review, like code
  • Per-table access scoping
  • Kill switch — full stop; a wrong number is worse than no number
Eval Metrics

What this workflow is held to

≥ 95%
Answer accuracy
on the weekly audit sample vs. hand-written SQL
Refusal rate
trending down as definition coverage grows
Minutes
Time-to-answer
vs. the analyst-queue baseline it replaces
Failure Modes

How it breaks, and what catches it

Silent assumption

An ambiguous question gets one plausible interpretation of several, and the wrong number ships to a board deck. Mitigation: Ambiguity detection forces a clarifying choice — assumptions are never silent, ever.

Semantic drift

The warehouse schema changes under the layer; queries succeed and mean something different. Mitigation: Contract tests between layer and warehouse run on every deploy; drift pauses the copilot automatically.

Authority creep

Teams cite copilot answers in board materials with no audit trail. Mitigation: Every answer carries definition version + timestamp; exports include full lineage by default.

Opsmarshal builds workflows like this one inside client stacks — see GTM stack architecture and AI enablement and agentic workflows, or book a call to walk through your own.

opsmarshal

Revenue Operations and AI Agentic Workflows for B2B companies in transformation. We build the machine. You run the company.

Services

Company

© 2026 Opsmarshal. All rights reserved. Privacy  ·  Terms