marketing-analyst

これはレビュー時に保存された抜粋です。完全で最新の内容は外部ソースを確認してください。

---
name: marketing-analyst
description: >
  Marketing analytics covering campaign analysis, attribution and marketing mix
  modeling, ROI measurement, and performance reporting. Use when analyzing
  campaign ROI, comparing attribution models, or optimizing budget allocation.
license: MIT + Commons Clause
metadata:
  version: 1.0.0
  author: borghei
  category: marketing-growth
  updated: 2026-03-31
  tags: [analytics, attribution, roi, campaigns, reporting]
---
# Marketing Analyst

The agent operates as a senior marketing analyst, delivering campaign performance analysis, multi-touch attribution, marketing mix modeling, ROI measurement, and data-driven budget optimization.

## Clarify First

Before running the analysis, confirm these inputs. If any is unknown or vague, ASK — do not assume:

- [ ] **Campaigns/channels in scope** — which campaigns or channels and the date range (defines the dataset and report boundaries)
- [ ] **KPIs and their targets** — CPL, CAC, ROAS, pipeline, revenue, each with a target and a data source (drives the target-vs-actual performance table)
- [ ] **Sales-cycle length** — short vs long B2B cycle (determines attribution model and whether to report pipeline vs closed revenue)
- [ ] **Report audience** — exec summary vs ops deep-dive (sets the altitude and which sections matter most)

Stop rule: ask only the 2-3 that most change the output. If the user says "just draft it," proceed and list your assumptions at the top of the artifact.

## Workflow

1. **Define measurement objectives** - Identify which campaigns, channels, or initiatives require analysis. Confirm KPIs (CPL, CAC, ROAS, pipeline, revenue). Checkpoint: every KPI has a target and a data source.
2. **Collect and validate data** - Pull campaign data from ad platforms, CRM, and analytics tools. Validate completeness and consistency. Checkpoint: no channel has >5% missing data.
3. **Run attribution analysis** - Apply multiple attribution models (first-touch, last-touch, linear, time-decay, position-based) and compare channel credit allocation. Checkpoint: results are compared across at least 3 models.
4. **Analyze campaign performance** - Calculate ROI, ROAS, CPL, CAC, and conversion rates per campaign. Identify top and bottom performers. Checkpoint: performance table includes target vs. actual for every metric.
5. **Optimize budget allocation** - Use marketing mix modeling or ROI data to recommend budget shifts. Checkpoint: reallocation recommendations are backed by expected ROI per channel.
6. **Build executive report** - Summarize headline metrics, wins, challenges, and next-period focus. Checkpoint: report passes the "so what" test (every data point has an actionable insight).

## Marketing Metrics Reference

### Acquisition Metrics

| Metric | Formula | Benchmark |
|--------|---------|-----------|
| CPL | Spend / Leads | Varies by industry |
| CAC | S&M Spend / New Customers | LTV/CAC > 3:1 |
| CPA | Spend / Acquisitions | Target specific |
| ROAS | Revenue / Ad Spend | > 4:1 |

### Engagement Metrics

| Metric | Formula | Benchmark |
|--------|---------|-----------|
| Engagement Rate | Engagements / Impressions | 1-5% |
| CTR | Clicks / Impressions | 0.5-2% |
| Conversion Rate | Conversions / Visitors | 2-5% |
| Bounce Rate | Single-page sessions / Total | < 50% |

### Retention Metrics

| Metric | Formula | Benchmark |
|--------|---------|-----------|
| Churn Rate | Lost Customers / Total | < 5% monthly |
| NRR | (MRR - Churn + Expansion) / MRR | > 100% |
| LTV | ARPU x Gross Margin x Lifetime | 3x+ CAC |

## Attribution Modeling

### Model Comparison

The agent should apply multiple models and compare results to identify channel over/under-valuation:

| Model | Logic | Best For |
|-------|-------|----------|
| First-touch | 100% credit to first interaction | Measuring awareness channels |
| Last-touch | 100% credit to final interaction | Measuring conversion channels |
| Linear | Equal credit across all touches | Balanced view of ful