Topical map

Topical Mapper

First: Read the file ~/.claude/skills/seo-references/core.md before proceeding. Apply the methodology and voice to all output.

Also read: ~/.claude/skills/seo-references/data-pull-patterns.md for MCP tool parameter reference.

Step 1: Parse Domain

Extract the domain from the user's argument. Resolve it against the MCP Routing Map in core.md:

• Strip protocol, www, trailing slashes, lowercase
• Look up group_id, GSC property, GA4 property, Semrush domain
• If domain not in map, warn and ask if user wants to proceed with Semrush only

Step 2: Pull Live Data

Pull these data sources (follow data-pull-patterns.md for exact parameters):

Semrush (PRIMARY — mandatory):

1. Call mcp__semrush__execute_report — Domain Organic Search Keywords for this domain, database=us, display_limit=30. This reveals what the domain currently ranks for.
2. Identify the primary keyword (highest volume purchase-intent term). Call mcp__semrush__execute_report — Related Keywords for that primary keyword, display_limit=20. This finds untapped long-tails.

GSC (ENRICHMENT — optional): 3. Call mcp__gscServer__get_search_analytics with the resolved siteUrl, last 28 days, dimensions=[query, page], rowLimit=50. This shows what Google already associates with this domain and actual click data.

If any source fails, print the appropriate [SEO] warning from core.md and continue.

Step 3: Analyze & Generate

Using the pulled data + the methodology, generate the topical map:

Structure: 1 Hub → 3 Sub-Hubs → 3 purchase-intent pages per Sub-Hub = 13 pages total.

For each page, specify:

• Target keyword (from Semrush/GSC data — real keywords with real volume)
• URL slug (keyword in URL — e.g., /houston-maritime-injury-lawyer)
• H1 (keyword-optimized, purchase-intent)
• Above-the-fold CTA recommendation (call, form, booking — vertical-specific)
• Word count target (Hub: 800-1000, Sub-Hub: 600-800, purchase-intent: 400-500)
• Internal link targets (which other pages in the silo this page links to)
• Search volume + CPC (from Semrush data)
• Current ranking position (from GSC data, if available)

Prioritization: Order purchase-intent pages by: (search volume × CPC) descending, with preference for keywords where the domain has impressions but low position (ranking gaps = low-hanging fruit).

Localization: Every page MUST be localized to {{CITY}} with sub-neighborhoods where applicable (e.g. for Houston: Galleria, Heights, Montrose, Katy).

Format: Present as a clean hierarchy:

HUB: [keyword] — [url] — [volume] — [word count]
  SUB-HUB 1: [keyword] — [url] — [volume] — [word count]
    COMPACT: [keyword] — [url] — [volume] — [word count]
    COMPACT: [keyword] — [url] — [volume] — [word count]
    COMPACT: [keyword] — [url] — [volume] — [word count]
  SUB-HUB 2: ...
  SUB-HUB 3: ...

Then provide a detailed spec for each page.

Anti-patterns to avoid: Do NOT suggest "What is..." or "How to..." pages. Every page must target a searcher ready to convert.

Step 4: Output

Follow the Output Protocol from core.md:

1. Print the full topical map playbook to terminal
2. Extract structured summary
3. Save to Graphiti with name Topical Map — [domain]

Close with the ONE specific page to create TODAY that will have the highest impact.