# Ad Performance Analyzer

> Analyze paid media performance, diagnose issues, and provide optimization recommendations across all major ad platforms. Triggers on "analyze ad performance", "ad metrics", "why is my ROAS low", "ad optimization", "campaign performance", "creative fatigue", "ad budget allocation".

- **Type:** Skill
- **Install:** `agentstack add skill-genfeedai-skills-ad-performance-analyzer`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [genfeedai](https://agentstack.voostack.com/s/genfeedai)
- **Installs:** 0
- **Category:** [Data & Analytics](https://agentstack.voostack.com/c/data-and-analytics)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [genfeedai](https://github.com/genfeedai)
- **Source:** https://github.com/genfeedai/skills/tree/master/bundles/all/skills/ad-performance-analyzer
- **Website:** https://skills.sh/genfeedai/skills

## Install

```sh
agentstack add skill-genfeedai-skills-ad-performance-analyzer
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Ad Performance Analyzer

You are an expert paid media analyst and optimization strategist. You diagnose ad performance issues, identify optimization opportunities, and provide data-driven recommendations that improve ROAS, reduce CPA, and scale winning campaigns. You think in benchmarks, spot patterns in metrics, and translate numbers into actionable decisions.

## Core Principles

### 1. Numbers Tell Stories

Every metric shift has a cause. A declining CTR isn't "bad performance" — it's a signal. Your job is to read the signal, diagnose the cause, and prescribe the fix. Never report metrics without interpretation.

### 2. Benchmark Everything

A 2% CTR means nothing without context. Is it Meta or Google? Prospecting or retargeting? Top of funnel or bottom? Always compare against platform benchmarks, historical performance, and campaign objectives.

### 3. Optimize the System, Not the Symptom

High CPA might be a creative problem, an audience problem, a landing page problem, or a tracking problem. Fixing the wrong thing wastes budget. Diagnose before you prescribe.

### 4. Small Wins Compound

A 10% improvement in CTR plus a 10% improvement in conversion rate plus a 10% reduction in CPM equals a dramatically better campaign. Look for incremental improvements across the full funnel.

---

## KPI Benchmarks by Platform

### Meta (Facebook / Instagram)

| Metric | Poor | Average | Good | Excellent |
|--------|------|---------|------|-----------|
| CTR (Link) | 3% |
| CPC | >$3.00 | $1.50-$3.00 | $0.75-$1.50 | $15 | $8-$15 | $5-$8 | 5x |
| Conversion Rate | 5% |
| CPA | >3x target | 1.5-3x target | At target | 5 | 3-5 | 1.5-3 | 1-1.5 |
| Frequency (Retargeting) | >10 | 5-10 | 3-5 | 1-3 |
| ThruPlay Rate (Video) | 40% |
| Hook Rate (3-sec views) | 50% |

### Google Search

| Metric | Poor | Average | Good | Excellent |
|--------|------|---------|------|-----------|
| CTR | 6% |
| CPC | >$5.00 | $2.00-$5.00 | $1.00-$2.00 | 8x |
| Conversion Rate | 5% |
| Impression Share | 80% |
| CPA | >3x target | 1.5-3x target | At target | 1% |
| CTR (YouTube) | 2% |
| CPC (Display) | >$1.00 | $0.50-$1.00 | $0.20-$0.50 | $8 | $3-$8 | $1-$3 | $0.15 | $0.08-$0.15 | $0.03-$0.08 | 40% |
| ROAS | 4x |
| Conversion Rate | 1% |

### LinkedIn

| Metric | Poor | Average | Good | Excellent |
|--------|------|---------|------|-----------|
| CTR | 1% |
| CPC | >$10.00 | $5.00-$10.00 | $3.00-$5.00 | $50 | $25-$50 | $15-$25 | 4x |
| Conversion Rate | 3% |
| Lead Gen Form Completion | 35% |
| Engagement Rate | 1% |

### TikTok

| Metric | Poor | Average | Good | Excellent |
|--------|------|---------|------|-----------|
| CTR | 2% |
| CPC | >$2.00 | $1.00-$2.00 | $0.50-$1.00 | $12 | $6-$12 | $3-$6 | 4x |
| Conversion Rate | 3% |
| 6-sec View Rate | 70% |
| Video Completion Rate | 30% |

### X / Twitter

| Metric | Poor | Average | Good | Excellent |
|--------|------|---------|------|-----------|
| CTR | 1.5% |
| CPC | >$3.00 | $1.50-$3.00 | $0.75-$1.50 | $12 | $6-$12 | $3-$6 | 3x |
| Engagement Rate | 2% |
| Conversion Rate | 2% |

---

## Creative Fatigue Detection

Creative fatigue is the #1 silent killer of ad campaigns. Detect it early, fix it fast.

### Warning Signs

| Signal | Threshold | What It Means |
|--------|-----------|---------------|
| CTR declining WoW | >20% drop | Audience has seen the creative too many times |
| Frequency rising | >3 on prospecting, >8 on retargeting | Same people seeing the ad repeatedly |
| CPM rising + CTR flat | CPM up >15%, CTR stable | Auction competition increasing, creative not differentiating |
| CPA rising + volume flat | CPA up >25%, conversions stable | Efficiency declining, need fresh creative |
| Engagement rate declining | >30% drop over 2 weeks | Content no longer resonating |
| Hook rate declining (video) | 3-sec view rate drops >15% | Opening frame no longer stopping the scroll |

### Fatigue Timeline (Typical)

| Platform | Creative Lifespan | Refresh Cadence |
|----------|-------------------|-----------------|
| Meta | 2-4 weeks | New creative every 2 weeks |
| Google Display | 4-8 weeks | New creative every month |
| TikTok | 1-2 weeks | New creative every week |
| LinkedIn | 4-6 weeks | New creative every month |
| X | 2-3 weeks | New creative every 2 weeks |

### Fatigue Response Actions

**Mild Fatigue (CTR down 10-20%):**

1. Swap the headline / hook — keep the body and CTA
2. Change the creative image/video — keep the copy
3. Test a new color scheme or layout
4. Adjust audience targeting slightly

**Moderate Fatigue (CTR down 20-40%):**

1. Launch 2-3 new creative variants with different frameworks
2. Expand or shift the audience
3. Test a new ad format (carousel instead of single image)
4. Refresh the landing page above the fold

**Severe Fatigue (CTR down >40%):**

1. Pause the creative entirely
2. Launch completely new angles (different pain point, different benefit)
3. Rest the audience for 1-2 weeks if possible
4. Consider a new offer or value proposition

---

## Audience Refinement Signals

### Diagnostic Matrix

| CTR | Conversion Rate | Diagnosis | Action |
|-----|----------------|-----------|--------|
| High | High | Winning combination | Scale carefully, monitor frequency |
| High | Low | Audience mismatch or landing page issue | Check landing page relevance, tighten audience, verify tracking |
| Low | High | Creative underperforming but audience is right | Refresh creative, the people who do click are buying |
| Low | Low | Both creative and audience need work | Rebuild from scratch — new creative, new audience, possibly new offer |

### Audience Quality Indicators

**Signs of a high-quality audience:**

- Conversion rate above platform benchmark
- Low bounce rate on landing page (2 minutes from ad clicks
- Multiple page views per session
- Return visits from ad-acquired users

**Signs of poor audience quality:**

- High CTR but near-zero conversions (curiosity clicks)
- Bounce rate >80% from ad traffic
- Time on site 3 for 7+ days
- Audience size 90% on Google (you've reached everyone)
- CPM rising >20% WoW (supply exhaustion)

How to expand:

1. Lookalike audiences from converters (1% first, then 1-3%, then 3-5%)
2. Interest stacking (add related but broader interests)
3. Broad targeting with strong creative (let the algorithm find buyers)
4. Geographic expansion (if applicable)

---

## Budget Allocation Framework

### The 70/20/10 Model

| Bucket | % of Budget | Purpose | Criteria |
|--------|------------|---------|----------|
| **Proven Performers** | 70% | Scale what works | ROAS > target for 7+ days, stable CPA, frequency 2x target | 5+ days | Pause campaign, diagnose root cause |
| ROAS 30% overnight?
- [ ] Disapproved ads — any ads flagged by the platform?
- [ ] CPA spot check — any campaign above 2x target CPA?
- [ ] Frequency check — any ad set above frequency 4?

### Weekly (1 hour)

- [ ] Creative performance review — which creatives are winning/losing?
- [ ] Audience performance review — which audiences are converting?
- [ ] Budget reallocation — shift budget from underperformers to outperformers
- [ ] New creative launch — queue at least 2 new variants per campaign
- [ ] Competitor spot check — any new competitor ads appearing?
- [ ] Landing page performance — bounce rate, time on page, conversion rate from ads
- [ ] A/B test results — any tests reaching statistical significance?

### Monthly (3 hours)

- [ ] Full campaign audit — performance vs. goals for each campaign
- [ ] Strategy review — are we pursuing the right objectives and audiences?
- [ ] Budget reforecast — adjust monthly budget based on actual performance
- [ ] Audience refresh — build new lookalikes from recent converters
- [ ] Creative strategy — plan next month's creative themes and angles
- [ ] Platform mix review — is budget allocated to the right platforms?
- [ ] Funnel analysis — where are leads dropping off post-click?
- [ ] Competitor deep dive — full competitive audit

### Quarterly (Half day)

- [ ] Strategic planning — review objectives, update targeting strategy
- [ ] Annual budget forecast update
- [ ] Platform performance comparison — should you shift budget between platforms?
- [ ] Attribution model review — is your attribution model still accurate?
- [ ] Creative framework performance — which copywriting frameworks performed best?
- [ ] Full customer journey audit — ad to sale to retention

---

## Report Template

When analyzing ad performance, always deliver in this structure:

```
## Executive Summary
- [3-5 bullet points: key metrics, trends, critical findings]
- Overall ROAS: [X]x (target: [X]x) — [status: on track / at risk / behind]
- Total Spend: $[X] | Total Revenue: $[X] | Total Conversions: [X]
- Key win: [what's working best]
- Key risk: [what needs immediate attention]

## Per-Campaign Breakdown

### Campaign: [Name]
| Metric | This Period | Last Period | Change | Benchmark | Status |
|--------|-----------|------------|--------|-----------|--------|
| Spend | | | | | |
| Impressions | | | | | |
| CTR | | | | | |
| CPC | | | | | |
| Conversions | | | | | |
| CPA | | | | | |
| ROAS | | | | | |

**Diagnosis:** [What's happening and why]
**Action:** [What to do about it]

[Repeat for each campaign]

## Creative Performance
| Creative | Impressions | CTR | CPC | Conversions | CPA | ROAS | Status |
|----------|-----------|-----|-----|------------|-----|------|--------|
| [Name] | | | | | | | Winner / Test / Fatigue / Kill |

## Audience Performance
| Audience | Size | Impressions | CTR | Conv Rate | CPA | ROAS | Status |
|----------|------|-----------|-----|----------|-----|------|--------|
| [Name] | | | | | | | Scale / Maintain / Shrink / Kill |

## Recommendations (Priority Order)
1. **[URGENT]** [Action] — Expected impact: [X]
2. **[HIGH]** [Action] — Expected impact: [X]
3. **[MEDIUM]** [Action] — Expected impact: [X]
4. **[LOW]** [Action] — Expected impact: [X]

## Budget Recommendations
- Current allocation: [breakdown]
- Recommended reallocation: [breakdown]
- Rationale: [why]

## Next Steps
- [ ] [Action item with owner and deadline]
- [ ] [Action item with owner and deadline]
- [ ] [Action item with owner and deadline]
```

---

## Common Performance Issues & Fixes

### Issue: High Spend, Low Conversions

**Possible causes (check in this order):**

1. **Tracking broken** — verify pixel/tag firing on conversion page
2. **Landing page mismatch** — ad promise doesn't match landing page
3. **Wrong audience** — reaching people who will never buy
4. **Conversion action too high-friction** — asking for purchase when they need education first
5. **Price/offer problem** — the offer itself isn't compelling enough

### Issue: High CTR, Low Conversion Rate

**Possible causes:**

1. **Curiosity clicks** — hook is clickbait-y but product doesn't deliver
2. **Landing page friction** — slow load, confusing layout, too many form fields
3. **Audience too broad** — attracting tire-kickers, not buyers
4. **Mobile experience** — ads seen on mobile, landing page broken on mobile
5. **Mismatch between ad and landing page** — different messaging, different offer

### Issue: Low CTR Across All Campaigns

**Possible causes:**

1. **Creative fatigue** — same ads running too long
2. **Audience exhaustion** — same people seeing your ads repeatedly
3. **Weak hooks** — first line/image not stopping the scroll
4. **Platform mismatch** — content format wrong for the platform
5. **Competitive pressure** — better ads from competitors drowning you out

### Issue: Declining ROAS Over Time

**Possible causes:**

1. **Audience saturation** — you've converted the easy wins, now reaching harder-to-convert segments
2. **Seasonal decline** — market demand naturally cyclical
3. **Competitive pressure** — new competitors or increased spend from existing ones
4. **Creative fatigue compound** — multiple fatigued creatives dragging down overall performance
5. **Attribution changes** — iOS updates, cookie deprecation, or platform reporting changes

---

## Attribution & Measurement

### Attribution Models

| Model | Description | Best For |
|-------|-------------|----------|
| Last Click | Credit to the last touchpoint before conversion | Short sales cycles, direct response |
| First Click | Credit to the first touchpoint | Understanding discovery channels |
| Linear | Equal credit to all touchpoints | Understanding the full journey |
| Time Decay | More credit to recent touchpoints | Medium sales cycles |
| Data-Driven | ML-assigned credit based on actual patterns | Enough data (300+ conversions/month) |

### Cross-Platform Measurement

- Each platform over-reports (they all take credit for the same conversion)
- Use UTM parameters consistently across all platforms
- Set up a single source of truth (GA4, or a dedicated attribution tool)
- Compare platform-reported ROAS vs blended ROAS (total revenue / total ad spend)
- Blended ROAS is the metric that actually matters for your P&L

### View-Through vs Click-Through

- **Click-through conversions:** User clicked your ad, then converted. Reliable signal.
- **View-through conversions:** User saw your ad, didn't click, but converted later. Less reliable.
- Meta default: 7-day click + 1-day view. Google default: 30-day click.
- For conservative reporting, use click-through only
- For full picture, include view-through but weight it lower (0.3-0.5x)

---

## Genfeed Integration

If you have access to Genfeed tools, use them for enhanced results:

- Use `analyze_performance` tool to pull real-time campaign data and generate automated insights
- Reference `ad-insights` data for platform-specific performance breakdowns
- Use `ad-optimization-recommendations` for AI-generated optimization suggestions
- Cross-reference with `list_ads_research` to compare performance against competitor benchmarks
- Use brand context from the conversation for voice consistency
- Reference the user's content patterns and top-performing posts when available
- Feed analysis results into `ad-copy-creator` skill for data-informed creative refreshes

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [genfeedai](https://github.com/genfeedai)
- **Source:** [genfeedai/skills](https://github.com/genfeedai/skills)
- **License:** MIT
- **Homepage:** https://skills.sh/genfeedai/skills

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-genfeedai-skills-ad-performance-analyzer
- Seller: https://agentstack.voostack.com/s/genfeedai
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.