---
name: meta-ads-analyzer
description: "Expert-level analysis and diagnosis for Meta Ads (Facebook/Instagram) campaigns. Interpret performance data, identify root causes of issues via Breakdown Effect analysis, and generate actionable recommendations. Use when analyzing Meta Ads, Facebook Ads, or Instagram advertising data."
---

# Meta Ads Analysis & Diagnosis

Expert framework for Meta Ads campaign analysis with focus on the Breakdown Effect.

## Core Principle

**The Breakdown Effect:** When breaking down campaign data by dimension (placement, age, gender, device), the sum of segment CPAs weighted by their spend does NOT equal the campaign-level CPA due to Meta's marginal cost optimization.

**Critical rule:** NEVER recommend pausing segments based solely on higher average CPA. Higher average cost does NOT mean poor performance — it often reflects the system capturing low marginal cost opportunities earlier.

## Analysis Workflow

### Step 1: Identify Evaluation Level

| Campaign Setup | Correct Evaluation Level |
|:---|:---|
| Advantage+ Campaign Budget (CBO) | **Campaign Level** |
| Automatic Placements (without CBO) | **Ad Set Level** |
| Multiple Ads in single Ad Set | **Ad Set Level** |

### Step 2: Analyze with Meta-Specific Lens

1. **Marginal Efficiency Analysis** — Infer marginal CPA trends from time-series data
2. **Ad Relevance Diagnostics** — Quality, Engagement, Conversion Rate Rankings
3. **Learning Phase Status** — ~50 results needed to exit learning phase

### Step 3: Synthesize Through Breakdown Effect Lens

Interpret all findings through marginal vs average cost dynamics.

## Key Concepts

### Marginal vs Average CPA

- **Average CPA** = Total Spend / Total Results
- **Marginal CPA** = Cost of the NEXT result
- System optimizes for lowest marginal CPA, not lowest average

**Example:** Placement A shows $10 avg CPA vs Placement B's $15. But Placement A's marginal CPA is rising sharply — system correctly shifts budget to Placement B to get more total results.

### Ad Auction Mechanics

Total Value = Advertiser Bid x Estimated Action Rate + Ad Quality

Winners are determined by Total Value, not just bid amount.

### Learning Phase

- Ad sets need ~50 optimization events to exit
- Performance during learning is unstable
- Avoid edits that reset learning (budget >20%, audience changes, creative swaps)

### Pacing

Meta distributes budget evenly across the day using pacing algorithms. Early underspend or overspend are normal system behavior.

## Mandatory Report Rules

1. **NEVER** recommend pausing segments based solely on higher average CPA
2. **ALWAYS** justify recommendations with data evidence + system mechanics
3. **USE** qualified language: "Estimated Reach of ~1,000" not "You reached 1,000"
4. **DISAMBIGUATE** clicks: "Clicks (all)" vs "Link Clicks" — always specify
5. **CHECK** `get_recommendations` API first if Meta API access available
6. Every insight must include data evidence and explanation

## Data Import

Users provide data via:
- CSV export from Meta Ads Manager
- Excel file with campaign data
- Manual data entry
- Meta Marketing API (if access token available)

### CSV Import Pattern

```python
import pandas as pd

# Read Meta Ads export
df = pd.read_csv("meta_ads_export.csv")

# Key columns
# Campaign name, Ad Set Name, Ad Name
# Impressions, Reach, Frequency
# Link Clicks, CPC, CTR
# Results, Cost per Result, Amount Spent
# Relevance Score / Quality Ranking
```

## When to Use

**Trigger words:** "meta ads", "facebook ads", "instagram ads", "рекламный кабинет", "CPA analysis", "breakdown effect", "ad performance", "campaign optimization", "ROAS", "cost per result"

## Reference Documents

For deep analysis, read these bundled references:
- `references/breakdown_effect.md` — The Breakdown Effect with examples
- `references/core_concepts.md` — Ad Auction, Pacing, Learning Phase
- `references/analysis_checklist.md` — Step-by-step analysis template