01 — Marketing Twin
Spend that proves itself.
A live model of demand generation — channels, audiences, creative, and spend — simulated against pipeline and revenue before the budget is committed.
North Star
Cost per qualified pipeline dollar
The stream
Automation → Capability → Value driver → Financial value
Automation
Always-on attribution and spend reallocation
Capability
Know the marginal return of every channel weekly
Value driver
Marketing efficiency
Financial value
Lower CAC, higher pipeline per dollar
Automation
Agentic brief, copy and variant generation
Capability
Ship 5× more tested creative at same headcount
Value driver
Speed to market
Financial value
Earlier revenue recognition per campaign
Automation
Audience and offer matching from the CDP
Capability
Right offer, right segment, governed consent
Value driver
Conversion quality
Financial value
Higher win rate on sourced pipeline
Value system
Value ≠ North Star ≠ KPI ≠ driver ≠ process metric.
The translation layer that connects a corporate objective to something an operating team can actually move on Monday.
Value
Profitable demand
Economic outcome
Contribution margin on sourced revenue
North Star
Cost per qualified pipeline dollar
Outcome KPIs
Driver metrics
Process metrics
Causal model
Value leakage
Where the value goes missing today.
Leakage is multiplicative. Every gate that stays leaky discounts everything built upstream of it.
Targeting
Spend against segments that never convert.
Wasted media
Hand-off
Qualified leads aged out before first touch.
Lost pipeline
Attribution
Last-click credit hides the real driver.
Misallocated budget
Consent
Audiences rebuilt after consent failures.
Rework and exposure
Digital twin
Simulate the interventions before anyone funds them.
Instead of implementing ten recommendations and learning the result in six months, the twin stacks them first.
Cost per qualified pipeline dollar
Guardrails
- Brand safety score maintained
- No special-category inference
- Share of voice not below floor
Constraints
- Quarterly budget cap
- Agency contract commitments
- Regional consent regimes
Automation
What runs without asking.
Budget reallocation loop
Weekly mix optimization against pipeline value, not clicks.
Attribution Agent
Creative production line
Brief → draft → brand check → variant set, human approved.
Content Ops Agent
Campaign QA + consent gate
Blocks launch when consent basis or claims fail policy.
Guardrail Agent
Campaign Planner Agent
Plans spend and sequencing against a pipeline target.
LLM reasoning + optimization solver
Attribution Agent
Maintains multi-touch and uplift models, explains shifts.
SLM classification + causal uplift
Content Ops Agent
Generates and routes creative through brand and legal checks.
LLM + RAG on brand corpus
Integrations
Where it plugs into the domain.
Vendor-agnostic by design. The twin reads and writes through whatever stack the domain already runs on.
Systems of record
- CRM (Salesforce, HubSpot, Dynamics)
- Marketing automation
- CDP
Channels
- Ad platform APIs
- Email / SMS
- Web & product analytics
Data & runtime
- Warehouse / lakehouse
- Metric layer
- MCP connectors
- Model gateway
Domain platforms we connect
CRM
Marketing automation
CDP & audiences
Advertising & channels
Analytics & content
Data & runtime
Plus anything else with an API — connectors are added per engagement, not sold as a platform lock-in.
Value
The levers, and what they move.
Cost
Kill spend on channels with no marginal pipeline effect.
CAC · Cost per qualified pipeline dollar
Growth
Shift budget to segments with proven uplift.
Sourced pipeline · Conversion rate
Speed
Compress brief-to-live cycle time.
Days from brief to campaign live
Risk
What could go wrong, and what stops it.
Every risk in this domain has a named containment in the runtime — not a slide.
Generated claims that cannot be substantiated
Claims policy engine + human approval before publish
High
Profiling without a lawful basis
Consent state enforced at audience build time
High
Attribution model drift after channel change
Holdout tests and uplift re-estimation each cycle
Medium
Compliance · Security
Built into the runtime, not bolted on.
EU AI Act
Limited risk — transparency obligations (generated content, personalization).
- Disclose AI-generated or AI-assisted content where required
- Log model, prompt and data lineage for every published asset
- Human review before any external publication
GDPR
- Lawful basis and consent state enforced at audience build time
- No special-category inference for targeting
- Data minimization: hashed identifiers only in the model layer
- DSAR and erasure propagated to the twin and vector store
Security
- SSO/SCIM with least privilege
- Tenant-isolated retrieval indexes
- Prompt-injection filtering on ingested web content
Controls & oversight
- Brand and claims policy engine
- Spend approval thresholds
- Full audit trail of agent actions
Agentic execution
Agents earn authority. They are not given it.
No agent in this twin controls anything it has not first proven in replay, evaluation, simulation and shadow.
01
Historical replay
Re-run the last 12 months. Would the agent have been right?
02
Offline evaluation
Scored against held-out outcomes, not opinion.
03
Digital twin
Simulated against the causal model under stress.
04
Shadow mode
Runs live, decides nothing. Divergence is logged.
05
Human recommendation
Proposes; a person executes and rates it.
06
Bounded pilot
One segment, capped exposure, hard rollback.
07
Human-supervised execution
Acts inside thresholds, humans approve exceptions.
08
Progressive autonomy
Authority widens only where evidence widened.
The unit of value
Don't buy transformation. Buy measurable movement.
This twin is contracted the way it is engineered: a baseline, a target, a guardrail, and an attribution method agreed in advance.
Baseline
Cost per qualified pipeline dollar $0.42
Target
$0.30 within two quarters
Guardrail
Sourced pipeline volume never below baseline
Proof / attribution
Geo and audience holdouts, reconciled to CRM revenue
Next twin
Sales Twin →