Agent-Managed Paid Acquisition
A policy-bounded acquisition workspace where campaigns, audiences, creatives, test cells, simulations, and budget decisions operate against explicit CAC/LTV guardrails.
The problem
Paid growth teams can now generate more creative and audience variants than they can responsibly govern. Without policy, automation simply creates faster spend movement and noisier optimization.
The thesis
Acquisition agents become useful when their action space is constrained by economics: target CAC, target LTV, confidence thresholds, budget-shift limits, cooldowns, and audit logs.
System flow
Acquisition flow diagram
High-level architecture for agent-managed paid acquisition.
Case-study artifacts
Test cells
Campaign variants are evaluated as measurable performance cells.
Guardrail loop
Budget shifts are bounded by caps, confidence, and target economics.
Economics
Insights compare CPA, ROAS, and LTV:CAC against campaign targets.
Decision frame 1
Input
Campaign objective, budget, channels, policy, creative variants, and audience templates.
Decision frame 2
Decision
Score test cells, identify winners/losers, and apply policy-bound budget movement.
Decision frame 3
Output
Insights panel, simulation distribution, budget timeline, override controls, and audit logs.
Evidence links
Shows architecture, readiness counts, and operating sequence.
Lists campaigns, states, cells, policy, and budget actions.
Shows editable audience templates and predicted CPC/CAC assumptions.
Shows scenario presets and Monte Carlo revenue distribution.
Shows economics, creative/audience trends, and budget timeline.
Guardrails
Operators can lock or revert budget overrides with audit records.
Campaign guardrails cap how much budget moves per iteration.
CAC and LTV thresholds inform scoring and scale/pause decisions.
System architecture
1. Campaign workspace
Stores objective, channel mix, budget, state, policy thresholds, and approval constraints.
2. Audience + creative library
Maintains editable targeting templates and creative variants with predicted CPC/CAC assumptions.
3. Test-cell engine
Combines creative, audience, and channel into measurable cells with spend, conversion, CAC, and ROAS.
4. Policy-aware iteration
Scores cells, proposes pauses or reallocations, and records every budget action in the audit trail.
KPI callouts
Economic policy
Campaign decisions are evaluated against target CAC, target LTV, and LTV:CAC guardrails.
Budget control
Budget movement is constrained by max-shift policy and operator override controls.
Simulation lab
Revenue scenarios use saved presets and Monte Carlo output for planning confidence.
Commercial framing
The commercial goal is not autonomous media buying for its own sake. It is faster learning under explicit unit-economics constraints, with enough auditability to trust the budget decisions.
What I built
I built the acquisition schema, campaign builder, audience and creative editors, test-cell controls, operator override flow, simulation panel, output views, connection scaffolding, and audit trail.