Ad Campaign Analyzer

Ad Campaign Analyzer

Usage

Use when reviewing a running campaign to decide what to kill, keep, or scale. Works for daily 15-minute ad reviews, weekly creative refresh planning, and monthly performance trend reviews.

Process

Step 1: Gather Inputs

Ask the user for:

1. Campaign data — one of:

• CSV or table with columns: ad/ad set name, impressions, clicks, conversions, spend, CPA
• Pasted text from ads manager
• Structured list of metrics per ad

1. Target CPA — the maximum they're willing to pay per acquisition
2. AOV (Average Order Value) — what they earn per conversion on the front end
3. Product/pricing info — what they sell, offer details, known conversion benchmarks
4. Daily budget per ad set (optional) — for scaling calculations
5. Days running (optional) — for statistical significance judgment
6. Historical data (optional) — from previous review for trend comparison

Step 2: Parse Campaign Data

Normalize the input into a consistent table structure:

| Ad / Ad Set | Impressions | Clicks | CTR | Conversions | Spend | CPA | Days Running | |-------------|------------|--------|-----|-------------|-------|-----|-------------|

Calculate any missing derived metrics:

• CTR = clicks / impressions × 100
• CPA = spend / conversions (∞ if 0 conversions)
• Conversion rate = conversions / clicks × 100

Step 3: Grade Each Ad — Red / Yellow / Green

🔴 RED = STOP

Kill this ad. It's burning money.

Criteria (any one triggers Red):

• Spent 1.5–2x target CPA with zero conversions
• CPA is 2x+ target CPA with statistically significant spend
• Consistently worsening metrics over multiple days with no improvement signs
• CTR below 0.5% after 1,000+ impressions (the creative isn't connecting)

Action: Turn off immediately. Redirect budget to greens.

🟡 YELLOW = LEAVE ALONE

Don't touch it. It needs more data or is borderline.

Criteria:

• CPA is close to target (within 0.5–1.5x) but not enough data to be confident
• Fewer than 1,000 impressions or fewer than 20 clicks — too early to judge
• Spend is under 1x target CPA — hasn't had a fair chance yet
• Metrics are mixed (good CTR but low conversion, or vice versa)

Action: Do nothing. Check again tomorrow. Resist the urge to tweak.

🟢 GREEN = SCALE

This ad is working. Give it more budget.

Criteria:

• CPA is consistently at or below target CPA
• Has statistically significant data (generally 10+ conversions)
• Metrics are stable or improving over time
• CTR is healthy for the targeting type

Action: Scale using the 20% Rule — increase daily budget by 20% every 48 hours.

Step 4: Benchmark Comparison

Compare each ad's metrics against industry benchmarks:

CTR Benchmarks (by targeting type): | Targeting | Expected CTR | |-----------|-------------| | Broad / run-of-network | 1–3% | | Interest-based targeting | 2–4% | | Lookalike / community-targeted | 3–5% |

CPA Targets by Offer Price: | Offer Price Range | Expected CPA Range | |-------------------|-------------------| | $7–27 (low ticket) | $20–40 | | $37–97 (mid ticket) | $40–120 | | $97+ (high ticket) | Varies — must model LTV |

Step 4b: Funnel Debugging — Find the Leak

Before changing creative, check whether the ad is actually the problem. Work from the surface inward:

1. Creative — Is the CTR acceptable? Low CTR = the ad isn't connecting. Test new hooks, visuals, or headlines.
2. Landing page — CTR is fine but conversions are flat? The LP is the problem.
3. Messaging — LP structure looks okay but still no conversions? The fundamental message may not be resonating.
4. Product and pricing — Different angles all fail? The offer itself may be the issue.
5. Market — Everything above looks solid but the right people still aren't converting? The segment may be wrong.

Work from #1 → #5 in order. Most founders jump to #3 or #4 when the actual problem is #1 or #2.

Data thresholds — don't debug on noise:

• CPA?

For each green ad, calculate:

• Profit per conversion = AOV − CPA
• ROAS = AOV / CPA (must be > 1.0 to be profitable)
• Break-even CPA = AOV (you make $0 at this point)
• Margin at current CPA = (AOV − CPA) / AOV × 100

Step 6b: LTV:CAC Health Check

If LTV data is available, assess the pricing-level health of the campaign:

LTV:CAC Ratio Benchmarks:

| Pricing Function | Average LTV:CAC | |-----------------|----------------| | No pricing function | 1.68 | | Yearly pricing review | 3.23 | | Continuous optimization | 11.09 |

Interpret the ratio:

• 5:1 — Excellent. May be under-investing in acquisition.

Monetization impact reminder: A 1% improvement in monetization yields a 12.7% increase in bottom-line revenue — roughly 4x the impact of acquisition and 2x the impact of retention. If LTV:CAC is weak, the fix may be pricing, not ads.

Step 6c: Attribution Notes

When analyzing multi-channel campaigns, note attribution limitations:

• Single-channel campaigns: last-click is sufficient.
• Multi-channel campaigns: flag that attribution is approximate.
• Don't over-complicate analytics. The goal is action (red/yellow/green), not perfect measurement.

Step 7: Generate Scaling Recommendations

For each green ad, provide specific scaling numbers:

The 20% Rule:

• Current daily budget → recommended new budget (current × 1.2)
• When to apply: 48 hours after last budget change
• Next check-in date

For the overall campaign:

• Total daily spend recommendation
• Budget reallocation from reds to greens
• When to add new creatives to the mix

Output Format

# Campaign Analysis

**Date:** [current date]
**Campaign:** [campaign name or description]
**Period:** [date range of data]
**Target CPA:** $[amount]
**AOV:** $[amount]

---

## Traffic Light Summary

| Grade | Count | % of Spend |
|-------|-------|-----------|
| 🔴 Red (Stop) | X | X% |
| 🟡 Yellow (Wait) | X | X% |
| 🟢 Green (Scale) | X | X% |

**Campaign Health:** [Healthy / Needs Attention / Critical] — [one sentence summary]

---

## Ad-Level Grades

| Ad / Ad Set | Grade | Spend | CPA | Target CPA | CTR | Conv. | Action |
|-------------|-------|-------|-----|-----------|-----|-------|--------|
| [name] | 🔴 | $X | $X | $X | X% | X | Stop — [reason] |
| [name] | 🟡 | $X | $X | $X | X% | X | Wait — [reason] |
| [name] | 🟢 | $X | $X | $X | X% | X | Scale to $X/day |

---

## Benchmark Comparison

| Metric | Your Average | Benchmark | Status |
|--------|-------------|-----------|--------|
| CTR | X% | X–X% | ✅ On track / ⚠️ Below / 🔥 Above |
| Conversion Rate | X% | X–X% | ✅ / ⚠️ / 🔥 |
| CPA | $X | $X–X | ✅ / ⚠️ / 🔥 |

---

## Creative Fatigue Alerts

[List any ads showing fatigue signals, or "No fatigue signals detected."]

---

## Profitability

| Ad / Ad Set | CPA | AOV | Profit/Conv. | ROAS | Margin |
|-------------|-----|-----|-------------|------|--------|
| [name] | $X | $X | $X | X.Xx | X% |

**Overall ROAS:** X.Xx
**Overall Profit/Conversion:** $X

---

## Action Items

### Immediate (Today)
- [ ] Stop: [list red ads]
- [ ] Scale: [list green ads with specific new budgets]

### This Week
- [ ] [Creative refresh, new tests, etc.]

### Review Cadence
- Next daily check: [tomorrow]
- Next weekly review: [date]
- Next monthly review: [date]

---

## Scaling Plan

| Ad | Current Budget | New Budget | Apply On | Next Increase |
|----|---------------|-----------|----------|---------------|
| [name] | $X/day | $X/day | [date] | $X/day on [date] |

**Total daily spend:** $X → $X (recommended)

Rules

• The grading method is deliberately binary. Red means stop, not "let's give it one more day." Kill losers fast and feed winners.
• Yellow is the discipline zone. Don't "optimize" yellows. Either there's enough data to judge or there isn't.
• The 20% rule exists because ad platforms optimize delivery around your budget. Jumping budgets overnight resets the algorithm's learning.
• CPA is the North Star, not CTR. A high-CTR ad that doesn't convert is worse than a low-CTR ad with great CPA.
• These benchmarks are starting points. After 2-4 weeks, your own data becomes the benchmark.
• If everything is red, the problem isn't the ads — it's the offer or the funnel.
• Creative fatigue is inevitable. Plan for it. Have your next batch of creatives ready before the current ones die.

Source & license

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

• Author: superamped
• Source: superamped/ai-marketing-skills
• License: MIT
• Homepage: https://superamped.com

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