AgentSkillsCN

ad-spend-optimizer

利用绩效数据、归因模型与ROI分析,优化付费广告预算在各渠道间的分配

SKILL.md
--- frontmatter
name: ad-spend-optimizer
description: Optimize paid advertising budget allocation across channels using performance data, attribution models, and ROI analysis
license: MIT
metadata:
  author: ClawFu
  version: 1.0.0
  mcp-server: "@clawfu/mcp-skills"

Ad Spend Optimizer

Systematically optimize paid advertising budget allocation across channels based on performance data, attribution analysis, and ROI targets.

When to Use This Skill

  • Quarterly budget planning
  • Channel mix optimization
  • Performance troubleshooting
  • Scaling paid acquisition
  • ROI analysis and reporting

Methodology Foundation

Based on marginal ROI optimization and portfolio theory for marketing, combining:

  • Channel performance analysis
  • Attribution modeling
  • Diminishing returns curves
  • Test and scale frameworks

What Claude Does vs What You Decide

Claude DoesYou Decide
Analyzes channel performanceBudget constraints
Calculates ROI by channelRisk tolerance
Recommends allocation shiftsTesting budgets
Identifies optimization opportunitiesBusiness priorities
Creates performance dashboardsPlatform selection

Instructions

Step 1: Audit Current Performance

Key Metrics by Channel:

MetricDefinitionTarget
ROASRevenue / Ad Spend>3:1
CACCost to Acquire Customer<LTV/3
CPACost per AcquisitionVaries
CTRClicks / ImpressionsBenchmark
Conv RateConversions / ClicksBenchmark

Step 2: Attribution Analysis

Attribution Models:

ModelLogicBest For
Last Click100% to final touchpointDirect response
First Click100% to first touchpointAwareness campaigns
LinearEqual across all touchpointsBalanced view
Time DecayMore to recent touchpointsShorter cycles
Position40% first, 40% last, 20% middleBalanced with emphasis
Data-DrivenML-based attributionSophisticated analysis

Step 3: Identify Optimization Opportunities

Optimization Levers:

LeverImpactEffort
Budget reallocationHighLow
Audience targetingHighMedium
Creative refreshMediumMedium
Bid optimizationMediumLow
Landing pageHighHigh
New channel testVariableMedium

Step 4: Calculate Optimal Allocation

Marginal ROI Analysis:

  • Where is next $1 best spent?
  • Identify diminishing returns
  • Find underinvested channels
  • Balance efficiency vs. scale

Step 5: Implement and Monitor

Test Framework:

  • 70% proven channels
  • 20% optimization tests
  • 10% new channel experiments

Examples

Example 1: Full Budget Optimization

Input:

code
Optimize this ad spend:
- Monthly budget: $100,000
- Channels: Google Ads (50K), Meta (30K), LinkedIn (15K), Other (5K)
- Goals: B2B SaaS, $200 target CAC, need 500 leads/month

Output:

code
## Ad Spend Optimization Analysis

### Current Performance Audit

| Channel | Spend | Leads | CAC | ROAS | CPL |
|---------|-------|-------|-----|------|-----|
| Google Ads | $50,000 | 180 | $278 | 2.1x | $278 |
| Meta | $30,000 | 150 | $200 | 2.8x | $200 |
| LinkedIn | $15,000 | 45 | $333 | 1.5x | $333 |
| Other | $5,000 | 20 | $250 | 2.0x | $250 |
| **Total** | **$100,000** | **395** | **$253** | **2.2x** | **$253** |

**Current Status:**
- Missing target: 395 leads vs 500 goal (79%)
- Above CAC target: $253 vs $200 goal

---

### Channel Analysis

#### Google Ads ($50K)

| Breakdown | Spend | Leads | CAC | Assessment |
|-----------|-------|-------|-----|------------|
| Brand | $10,000 | 80 | $125 | ✅ Efficient |
| Non-Brand | $25,000 | 70 | $357 | ⚠️ High CAC |
| Display | $15,000 | 30 | $500 | ❌ Cut |

**Findings:**
- Brand is highly efficient but limited scale
- Non-brand has high CAC but scalable
- Display not performing—recommend pause

**Recommendations:**
- Increase brand budget (limited by search volume)
- Optimize non-brand (better audiences, negatives)
- Pause display, reallocate to Meta

---

#### Meta ($30K)

| Breakdown | Spend | Leads | CAC | Assessment |
|-----------|-------|-------|-----|------------|
| Lookalike | $15,000 | 85 | $176 | ✅ Best performer |
| Interest | $10,000 | 45 | $222 | ✅ Good |
| Retargeting | $5,000 | 20 | $250 | ✅ Standard |

**Findings:**
- Lookalike audiences are star performers
- Interest targeting has room to scale
- Strong overall channel

**Recommendations:**
- Increase lookalike budget significantly
- Test new lookalike seeds
- Shift underperforming Google budget here

---

#### LinkedIn ($15K)

| Breakdown | Spend | Leads | CAC | Assessment |
|-----------|-------|-------|-----|------------|
| Sponsored Content | $10,000 | 35 | $286 | ⚠️ High |
| Lead Gen Forms | $5,000 | 10 | $500 | ❌ Very high |

**Findings:**
- Highest CAC channel
- But: LinkedIn leads often higher quality (enterprise)
- Lead Gen Forms underperforming

**Recommendations:**
- Reduce overall LinkedIn spend
- Shift to sponsored content only
- Test LinkedIn for enterprise segment specifically

---

#### Other ($5K)

**Breakdown:** Reddit, Quora, programmatic
**Performance:** Mixed, small sample sizes

**Recommendation:** Continue testing but don't scale yet

---

### Recommended Budget Reallocation

#### Before vs After

| Channel | Current | Proposed | Change |
|---------|---------|----------|--------|
| Google Ads | $50,000 | $35,000 | -$15,000 |
| Meta | $30,000 | $50,000 | +$20,000 |
| LinkedIn | $15,000 | $8,000 | -$7,000 |
| Testing | $5,000 | $7,000 | +$2,000 |
| **Total** | **$100,000** | **$100,000** | - |

---

#### Detailed Allocation

**Google Ads ($35K)**
| Campaign | Budget | Expected Leads | Expected CAC |
|----------|--------|----------------|--------------|
| Brand | $12,000 | 95 | $126 |
| Non-Brand (optimized) | $23,000 | 75 | $307 |
| Display | $0 | 0 | - |
| **Total** | **$35,000** | **170** | **$206** |

**Meta ($50K)**
| Campaign | Budget | Expected Leads | Expected CAC |
|----------|--------|----------------|--------------|
| Lookalike | $30,000 | 170 | $176 |
| Interest | $15,000 | 65 | $231 |
| Retargeting | $5,000 | 20 | $250 |
| **Total** | **$50,000** | **255** | **$196** |

**LinkedIn ($8K)**
| Campaign | Budget | Expected Leads | Expected CAC |
|----------|--------|----------------|--------------|
| Sponsored Content | $8,000 | 28 | $286 |
| Lead Gen | $0 | 0 | - |
| **Total** | **$8,000** | **28** | **$286** |

**Testing ($7K)**
| Test | Budget | Purpose |
|------|--------|---------|
| TikTok | $3,000 | New channel |
| YouTube | $2,500 | Video content |
| Reddit (scale) | $1,500 | Community |

---

### Projected Results

| Metric | Current | Proposed | Change |
|--------|---------|----------|--------|
| Total Leads | 395 | 473 | +20% |
| Blended CAC | $253 | $211 | -17% |
| Goal Achievement | 79% | 95% | +16 pts |
| ROAS | 2.2x | 2.6x | +18% |

---

### Implementation Roadmap

**Week 1:**
- [ ] Pause Google Display campaigns
- [ ] Reduce LinkedIn Lead Gen to $0
- [ ] Increase Meta Lookalike by $5K

**Week 2:**
- [ ] Complete Google to Meta shift
- [ ] Launch TikTok test campaign
- [ ] Set up new Lookalike audiences

**Week 3-4:**
- [ ] Monitor and optimize
- [ ] Scale what's working
- [ ] Report on progress

---

### Monitoring Dashboard

┌────────────────────────────────────────────────────────┐ │ AD SPEND DASHBOARD Week 1 │ ├────────────────────────────────────────────────────────┤ │ │ │ SPEND PACING │ │ Google: ████████░░░░ $8,200/$8,750 (94%) │ │ Meta: █████████░░░ $11,800/$12,500 (94%) │ │ LinkedIn:██████░░░░░░ $1,500/$2,000 (75%) │ │ Testing: █████░░░░░░░ $900/$1,750 (51%) │ │ │ │ LEADS THIS WEEK │ │ Target: 118 | Actual: 124 | Pacing: +5% │ │ │ │ CAC THIS WEEK │ │ Target: $211 | Actual: $195 | Status: ✅ │ │ │ │ ALERTS: │ │ ⚠️ LinkedIn CPC up 15% - monitor │ │ ✅ Meta Lookalike exceeding expectations │ │ │ └────────────────────────────────────────────────────────┘

code

Example 2: Quick Channel Assessment

Input:

code
Should we increase spend on this channel?
- Channel: Facebook
- Current spend: $10K/month
- CAC: $150
- Target CAC: $200
- Leads: 67
- Need: More leads

Output:

code
## Channel Scale Assessment: Facebook

### Current Performance

| Metric | Value | Status |
|--------|-------|--------|
| Spend | $10,000 | Baseline |
| Leads | 67 | - |
| CAC | $150 | ✅ Below target |
| Headroom | $50 | Room to scale |

### Scale Recommendation: YES, but carefully

**Why scale:**
- CAC ($150) is 25% below target ($200)
- Indicates efficiency headroom
- Leads are needed

**How to scale:**

| Scenario | Spend | Expected Leads | Expected CAC |
|----------|-------|----------------|--------------|
| Conservative | $15,000 | 90 | $167 |
| Moderate | $20,000 | 110 | $182 |
| Aggressive | $25,000 | 125 | $200 |

**Recommendation:** Start with moderate (+$10K)

### Scaling Checklist

- [ ] Expand Lookalike audiences
- [ ] Test new interest targets
- [ ] Increase frequency caps gradually
- [ ] Monitor CAC weekly
- [ ] Set alert at $185 CAC

### Warning Signs (Stop Scaling)

- CAC exceeds $200
- CTR drops >20%
- Frequency >3.0
- Negative ROI on increment

Skill Boundaries

What This Skill Does Well

  • Analyzing channel performance
  • Recommending budget shifts
  • Calculating ROI projections
  • Creating optimization frameworks

What This Skill Cannot Do

  • Access your ad accounts
  • Make real-time bid changes
  • Know your specific creative
  • Guarantee performance

Iteration Guide

Follow-up Prompts:

  • "Analyze [specific channel] performance"
  • "How should we test [new channel]?"
  • "Create a pacing dashboard for [budget]"
  • "What's causing [performance issue]?"

References

  • Google Ads Optimization Guide
  • Meta Business Suite Best Practices
  • LinkedIn Marketing Solutions
  • AdEspresso Budget Allocation

Related Skills

  • google-ads-expert - Google-specific
  • aarrr-metrics - Full funnel view
  • growth-loops - Sustainable growth

Skill Metadata

  • Domain: Acquisition
  • Complexity: Intermediate-Advanced
  • Mode: centaur
  • Time to Value: 2-3 hours per analysis
  • Prerequisites: Ad account access, performance data