Budget Optimization

Overview

Budget optimization finds where your money works hardest and recommends shifts from low-efficiency campaigns to high-performers. All budget changes require explicit approval.

The budget-optimizer skill analyzes spend efficiency across campaigns and adsets, comparing CPA, ROAS, and CPC to identify reallocation opportunities.

How It Works

Pull performance data

The optimizer fetches campaign and adset-level insights for your specified period (default: last 7 days).

Calculate efficiency metrics

CPA (Cost per Acquisition) — How much you pay per conversion
ROAS (Return on Ad Spend) — Revenue generated per dollar spent
CPC (Cost per Click) — How much each click costs

These are compared against benchmarks in ad-config.json.

Rank performers

Campaigns are ranked from most to least efficient. Top performers get flagged for budget increases, bottom performers for decreases.

Generate recommendations

The optimizer calculates suggested budget shifts with clear reasoning:

Where to pull budget from (losers)
Where to allocate it (winners)
How much to shift (percentage-based)

Wait for approval

Recommendations are presented with data. No budget changes occur until you explicitly approve.

Running the Optimizer

Ask the copilot directly

Tell me: "How should I shift my budget?"
# Or
Tell me: "Optimize my ad spend"
# Or
Tell me: "Where's my money working best?"

Via script (manual)

# Efficiency ranking
./scripts/budget-optimizer.sh efficiency --account act_123 --preset last_7d

# Budget shift recommendations
./scripts/budget-optimizer.sh recommend --account act_123

# Spend pacing check
./scripts/budget-optimizer.sh pacing --account act_123

Understanding Recommendations

The optimizer presents recommendations like this:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
💰 BUDGET OPTIMIZATION — Last 7 Days
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

EFFICIENCY RANKING:

1. Campaign "Product Launch" — $18.20 CPA, 4.2x ROAS, $450/day budget
   ✓ Top performer. Recommend +30% budget → $585/day

2. Campaign "Brand Awareness" — $24.50 CPA, 3.1x ROAS, $300/day budget
   → On target. No change.

3. Campaign "Retargeting" — $42.00 CPA, 1.2x ROAS, $200/day budget
   ⚠ Underperforming. Recommend -50% budget → $100/day

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
RECOMMENDED SHIFTS:

- Decrease "Retargeting" by $100/day
- Increase "Product Launch" by $100/day

Net change: $0 (reallocation only)

Target CPA: $25.00
Expected blended CPA after shift: $21.30 (15% improvement)

Proceed with these budget changes?

Budget changes take effect immediately. Meta will adjust delivery based on the new budget within the hour.

Adjusting Budgets Safely

Via Copilot (Recommended)

When you approve recommendations:

You: "Yes, make those changes"

Agent:
✓ Set "Retargeting" daily budget to $100 (was $200)
✓ Set "Product Launch" daily budget to $585 (was $450)

Budget shifts complete. Logged to learnings.md.

The agent executes using:

# Decrease underperformer
social marketing set-budget campaign CAMPAIGN_ID_1 --daily-budget 10000  # $100 in cents

# Increase winner
social marketing set-budget campaign CAMPAIGN_ID_2 --daily-budget 58500  # $585 in cents

Manual Commands

# Set adset daily budget
social marketing set-budget adset ADSET_ID --daily-budget 5000  # $50.00 in cents

# Set campaign daily budget
social marketing set-budget campaign CAMPAIGN_ID --daily-budget 10000  # $100.00 in cents

Budgets are specified in cents. $50.00 = 5000 cents.

Optimization Strategies

Strategy 1: Reallocation (Zero-Sum)

Keep total spend the same, shift budget from losers to winners.

Before: Campaign A ($200) + Campaign B ($200) = $400/day total
After:  Campaign A ($100) + Campaign B ($300) = $400/day total

Use when: You have a fixed budget and want better efficiency.

Strategy 2: Scale Winners

Increase budget on top performers without decreasing losers.

Before: Campaign A ($200) + Campaign B ($200) = $400/day total
After:  Campaign A ($200) + Campaign B ($350) = $550/day total

Use when: You have room to spend more and want to capitalize on momentum.

Strategy 3: Cut Losers

Decrease or pause underperformers without increasing others.

Before: Campaign A ($200) + Campaign B ($200) = $400/day total
After:  Campaign A ($100) + Campaign B ($200) = $300/day total

Use when: You’re overspending and need to reduce burn.

Approval Workflow Example

You: "Analyze spend efficiency"

Agent:
[Pulls last 7 days of campaign data]

Found 3 campaigns:
- "Product Launch": $18.20 CPA, 4.2x ROAS (Winner)
- "Brand Awareness": $24.50 CPA, 3.1x ROAS (On target)
- "Retargeting": $42.00 CPA, 1.2x ROAS (Underperformer)

Your target CPA: $25.00

Recommendation:
- Cut "Retargeting" budget by 50% ($200 → $100)
- Shift that $100/day to "Product Launch" ($450 → $550)

Expected result: Blended CPA drops from $26.80 to $21.30

Should I make these changes?

You: "Yes"

Agent:
✓ Updated "Retargeting" daily budget: $100
✓ Updated "Product Launch" daily budget: $550

Changes logged to workspace/brand/learnings.md.
I'll monitor performance and report back tomorrow.

Safety Guidelines

Never shift budget without approval. The optimizer will always:

Show current budget vs. recommended budget
Explain the data justification (CPA, ROAS, trend)
Estimate the expected outcome
Wait for explicit “yes” before executing

Before Approving Budget Shifts

Do the efficiency metrics justify the change?
Is the recommendation based on enough data? (minimum 7 days preferred)
Will this affect other campaigns or overall account pacing?
Are you prepared for the new spend level if scaling up?
Have you checked for external factors (seasonality, new creative, audience changes)?

Benchmarks and Thresholds

Default thresholds from ad-config.json:

Metric	Default	Meaning
Target CPA	$25.00	Aim to stay at or below this cost per acquisition
Target ROAS	3.0x	Aim for at least $3 revenue per$ 1 spent
Spend pace alert	±15%	Warn if daily spend is 15%+ over/under budget
Efficiency variance	±20%	Flag campaigns 20%+ above/below target

You can customize these in ad-config.json:

{
  "benchmarks": {
    "target_cpa": 25.00,
    "target_roas": 3.0,
    "spend_pace_threshold": 0.15,
    "efficiency_variance": 0.20
  }
}

Audit Trail

Every budget change is logged to workspace/brand/learnings.md:

## 2026-03-04 — Budget Optimization

**Changes Made:**

- Campaign "Retargeting" (ID: 120212345682)
  - Old budget: $200/day
  - New budget: $100/day
  - Reason: CPA $42.00 vs. target $25.00, ROAS 1.2x vs. target 3.0x
  - User approval: Confirmed after efficiency analysis

- Campaign "Product Launch" (ID: 120212345683)
  - Old budget: $450/day
  - New budget: $550/day
  - Reason: CPA $18.20 (28% below target), ROAS 4.2x (40% above target)
  - User approval: Confirmed after efficiency analysis

**Expected Impact:**
- Blended CPA: $26.80 → $21.30 (20% improvement)
- Total spend: $400/day (unchanged, reallocation only)

Next Steps

Pause and Resume — Stop bleeders, restart winners
Safety Model — Understand approval gates and audit logging

Get Started

Core Concepts

Skills

Configuration

Reports

Actions

Advanced

Overview

How It Works

Running the Optimizer

Ask the copilot directly

Via script (manual)

Understanding Recommendations

Adjusting Budgets Safely

Via Copilot (Recommended)

Manual Commands

Optimization Strategies

Strategy 1: Reallocation (Zero-Sum)

Strategy 2: Scale Winners

Strategy 3: Cut Losers

Approval Workflow Example

Safety Guidelines

Before Approving Budget Shifts

Benchmarks and Thresholds

Audit Trail

Next Steps

​Overview

​How It Works

​Running the Optimizer

​Ask the copilot directly

​Via script (manual)

​Understanding Recommendations

​Adjusting Budgets Safely

​Via Copilot (Recommended)

​Manual Commands

​Optimization Strategies

​Strategy 1: Reallocation (Zero-Sum)

​Strategy 2: Scale Winners

​Strategy 3: Cut Losers

​Approval Workflow Example

​Safety Guidelines

​Before Approving Budget Shifts

​Benchmarks and Thresholds

​Audit Trail

​Next Steps

Overview

How It Works

Running the Optimizer

Ask the copilot directly

Via script (manual)

Understanding Recommendations

Adjusting Budgets Safely

Via Copilot (Recommended)

Manual Commands

Optimization Strategies

Strategy 1: Reallocation (Zero-Sum)

Strategy 2: Scale Winners

Strategy 3: Cut Losers

Approval Workflow Example

Safety Guidelines

Before Approving Budget Shifts

Benchmarks and Thresholds

Audit Trail

Next Steps