Module 3: Budget Reallocation Agent

Make data-driven budget decisions: where to cut, where to scale, and why

What This Agent Does

Analyze spend efficiency across campaigns
Identify underperforming budget allocation
Find scale opportunities with proven ROI
Calculate optimal reallocation amounts
Provide justification for every budget change

This removes emotion and politics from budget decisions.

Prerequisites

Completed Module 1 & 2
Performance data with spend + ROAS
Clear target ROAS or CPA thresholds

Step 1: Define Budget Reallocation Standard

Create: `workflows/budget_reallocation_standard.md`

Objective: Reallocate budget to maximize growth within acceptable risk.

When to Scale (Increase Budget)

ROAS above target by 20%+
Volume capacity available (not maxed out)
Sustained for 7+ days

When to Cut (Decrease Budget)

ROAS below target for 14+ days
CPA 30%+ above acceptable threshold
No improvement after creative refresh

Risk Tolerance

Max budget increase per campaign: 30% weekly
Max budget cut: 50% weekly (don't turn off immediately)
Minimum data window before decision: 7 days

Step 2: Prepare Budget Inputs

Export and place in `/inputs`:

bash

current_budget_allocation.csv (campaign, current daily budget)

bash

performance_by_campaign.csv (spend, revenue, ROAS, CPA)

bash

scale_capacity_notes.md (manual notes on ad set limits)

Ensure performance data includes at least 7 days of history.

Step 3: Create Reallocation Workflow

Analysis Steps

Rank campaigns by ROAS efficiency
Identify candidates for scale (meeting thresholds)
Identify candidates for cuts (below thresholds)
Calculate recommended reallocation amounts
Provide reasoning for each change

Rules

Never reallocate based on single-day spikes
Account for seasonality in decision logic
Check scale capacity before recommending increases

Output Format

Cut list with amounts
Scale list with amounts
Net budget change summary

Step 4: Run the Budget Agent

bash

claude "Execute workflows/budget_reallocation_workflow.md using latest budget data."

Step 5: Save Output

bash

outputs/budget_reallocation_YYYY_MM_DD.md

Step 6: Human Override

Verify scale candidates have room to grow
Check if cuts align with strategic priorities
Adjust timing based on cash flow

Step 7: Iterate Monthly

Review accuracy of ROAS predictions
Refine scale/cut thresholds
Update risk tolerance if needed

Common Failure Modes

Premature cuts

Cause: Thresholds too aggressive

Fix: Increase minimum data window to 14 days

Missing seasonal patterns

Cause: No historical context in analysis

Fix: Add year-over-year comparison

Over-concentration risk

Cause: Agent doesn't check portfolio balance

Fix: Add diversification rules to standard

Why This Works

Budget moves with data, not gut feel
Risk is quantified, not assumed
Decisions are documented and auditable
Emotions are removed from reallocation

This is how efficient growth scales.

← Previous: Module 2: Creative Performance Review Agent Next: Module 4: Pre-Launch Risk Assessment Agent →