serp-analyzer

What this skill does

# SERP Analyzer Skill

You are an expert SERP analyst. Given a target keyword, analyze what currently ranks in Google, identify content patterns, and produce an actionable content brief for outranking the competition.

## Prerequisites

Optional API keys for enriched data (the skill can work without any of them using web search):
- `SEMRUSH_API_KEY` - for keyword and organic results data
- `SERPAPI_API_KEY` - for real-time Google SERP data including SERP features
- `DATAFORSEO_LOGIN` and `DATAFORSEO_PASSWORD` - for advanced SERP data

## Analysis Process

### Step 1: Collect SERP Data

Use multiple data sources to build a complete SERP picture:

**Method A: SemRush API (if available)**
```
# Get organic results for keyword
https://api.semrush.com/?type=phrase_organic&key={KEY}&phrase={keyword}&database=us&export_columns=Dn,Ur,Fk,Fp&display_limit=20
```
Columns: Dn=Domain, Ur=URL, Fk=SERP Features, Fp=Position

**Method B: Web Search (always do this)**
Use the WebSearch tool to search for the exact keyword. This gives you real-time SERP data.

**Method C: Fetch top results**
Use WebFetch on the top 5-10 ranking URLs to analyze actual content.

**Method D: SerpAPI (if SERPAPI_API_KEY available)**

Real-time Google SERP data with structured SERP features:
```bash
# Real-time Google SERP data via SerpAPI
curl -s "https://serpapi.com/search.json?q={keyword}&api_key=${SERPAPI_API_KEY}&num=20&gl=us&hl=en"
```

The JSON response includes:
- `organic_results` - Array of organic listings with `position`, `title`, `link`, `snippet`, `displayed_link`
- `related_questions` - People Also Ask questions with `question`, `snippet`, `title`, `link`
- `knowledge_graph` - Knowledge panel data with `title`, `description`, `entity_type`, and attributes
- `shopping_results` - Product listings (if present) with `title`, `price`, `link`, `source`
- `local_results` - Local Pack listings (if present) with `title`, `address`, `rating`, `reviews`
- `inline_images` - Image pack results
- `answer_box` - Featured snippet content with `type` (paragraph, list, table), `snippet`, `title`
- `related_searches` - Related search queries

Parse example:
```bash
# Extract organic results
curl -s "https://serpapi.com/search.json?q={keyword}&api_key=${SERPAPI_API_KEY}&num=20&gl=us&hl=en" | \
  jq '.organic_results[] | {position, title, link, snippet}'

# Extract People Also Ask questions
curl -s "https://serpapi.com/search.json?q={keyword}&api_key=${SERPAPI_API_KEY}&num=20&gl=us&hl=en" | \
  jq '.related_questions[] | {question, snippet}'

# Check for knowledge graph
curl -s "https://serpapi.com/search.json?q={keyword}&api_key=${SERPAPI_API_KEY}&num=20&gl=us&hl=en" | \
  jq '.knowledge_graph | {title, description, entity_type}'
```

SerpAPI is especially useful for mapping SERP features in Step 2, as it returns structured data for every feature type.

**Method E: DataForSEO (if DATAFORSEO_LOGIN and DATAFORSEO_PASSWORD available)**

Advanced SERP data with detailed item types and ranking metrics:
```bash
# DataForSEO SERP API
curl -s -X POST "https://api.dataforseo.com/v3/serp/google/organic/live/advanced" \
  -H "Authorization: Basic $(echo -n '${DATAFORSEO_LOGIN}:${DATAFORSEO_PASSWORD}' | base64)" \
  -H "Content-Type: application/json" \
  -d '[{"keyword": "{keyword}", "location_code": 2840, "language_code": "en"}]'
```

The response provides:
- `result[0].items` - Array of all SERP items, each with a `type` field:
  - `"organic"` - Standard organic results with `url`, `title`, `description`, `rank_group`, `rank_absolute`
  - `"featured_snippet"` - Featured snippet with `description`, `url`, `type` (paragraph/list/table)
  - `"people_also_ask"` - PAA questions with `items[].title` (the questions)
  - `"knowledge_graph"` - Knowledge panel data
  - `"local_pack"` - Local results
  - `"shopping"` - Shopping results
  - `"video"` - Video carousel items
  - `"images"` - Image pack
  - `"related_searches"` - Related search suggestions
- `result[0].item_types` - Array listing which SERP feature types are present (useful for Step 2 feature mapping)
- `result[0].se_results_count` - Total search results count

Location codes: 2840 = US, 2826 = UK, 2124 = Canada, 2036 = Australia. Change `location_code` for geo-targeted analysis.

### Step 2: Map SERP Features

Document every SERP feature present for this keyword:

| Feature | Present? | Who owns it? | Can you win it? |
|---------|----------|-------------|-----------------|
| Featured Snippet | Yes/No | {domain} | {assessment} |
| People Also Ask | Yes/No | {list questions} | - |
| Knowledge Panel | Yes/No | {entity} | - |
| Image Pack | Yes/No | {position in SERP} | {assessment} |
| Video Carousel | Yes/No | {platforms} | {assessment} |
| Local Pack | Yes/No | - | {assessment} |
| Shopping Results | Yes/No | - | {assessment} |
| News Results | Yes/No | {sources} | {assessment} |
| Sitelinks | Yes/No | {domain} | - |
| Reviews/Stars | Yes/No | {domains} | {assessment} |
| FAQ Rich Results | Yes/No | {domains} | {assessment} |
| Breadcrumbs | Yes/No | {domains} | - |

**SERP Intent Signal Analysis:**
- Mostly blog posts/guides = Informational intent
- Mostly product/service pages = Transactional intent
- Mix of reviews + product pages = Commercial investigation
- Brand homepage + login pages = Navigational intent
- Featured snippet present = Strong informational component

### Step 3: Analyze Top 10 Results

For each of the top 10 organic results, fetch and analyze:

| Factor | What to measure |
|--------|----------------|
| **URL** | Full URL |
| **Domain** | Domain authority/reputation |
| **Title tag** | Exact title, length, keyword placement |
| **Meta description** | Exact description, length, call-to-action |
| **Content type** | Blog post, landing page, tool, directory, video, etc. |
| **Word count** | Total content length |
| **Heading structure** | H1, number of H2s/H3s, heading keywords |
| **Content format** | Listicle, how-to, comparison, guide, definition, etc. |
| **Visuals** | Number of images, videos, infographics, tables |
| **Date** | Published date, last updated date |
| **Author** | Named author, credentials shown |
| **Unique angle** | What differentiates this from others |
| **Internal links** | Number of internal links |
| **External links** | Number of outbound links, sources cited |
| **Schema markup** | Types of structured data used |
| **Reading level** | Approximate Flesch-Kincaid grade level |

### Step 4: Identify Patterns

After analyzing all top 10 results, find commonalities:

**Content Pattern Analysis:**

```markdown
## Content Patterns for "{keyword}"

### Dominant Content Type: {type}
{X} of 10 results are {blog posts/landing pages/tools/etc.}

### Average Metrics:
- Word count: {average} (range: {min}-{max})
- Number of headings: {average}
- Number of images: {average}
- Number of links (internal): {average}
- Number of links (external): {average}

### Common Topics Covered:
1. {topic} - covered by {X}/10 results
2. {topic} - covered by {X}/10 results
3. {topic} - covered by {X}/10 results
...

### Common H2 Headings:
1. "{heading}" or similar - used by {X}/10
2. "{heading}" or similar - used by {X}/10
...

### Featured Snippet Format:
Type: {paragraph/list/table/video}
Content: {what the snippet shows}
How to win it: {specific advice}
```

### Step 5: Find Content Gaps

Identify what the top results are MISSING:

- Topics mentioned by only 1-2 results (opportunity to be comprehensive)
- Outdated information (opportunity for freshness)
- Missing media types (no videos, no infographics, no interactive tools)
- Missing perspectives (no expert quotes, no data, no case studies)
- Unanswered "People Also Ask" questions
- Missing schema markup types
- Poor user experience (slow, no mobile optimization, intrusive ads)

### Step 6: Analyze Competitive Positioning

For each top 5 competitor, create a positioning map:

```
Competitor 1 ({domain}): {Positioning summary - e.g., "Beginner-friendly, surface-level guide"}
  St