SEO Analysis — MCP Marketers

Use this skill to run disciplined, multi-tool SEO analysis workflows using the MCP Marketers server's 32 tools. Do not call tools in isolation. Most analysis calls hit paid DataForSEO APIs, while setup/report/read helpers are free or cached — cost awareness and correct chaining are still critical.

Activation boundaries

Use this skill when

• Running a technical SEO audit on a website
• Discovering and analyzing keywords for a domain or topic
• Analyzing or comparing backlink profiles
• Investigating competitors and domain overlaps
• Analyzing SERP features and ranking landscapes
• Tracking keyword ranking positions over time
• Planning content strategy based on keyword gaps and trends
• Monitoring brand visibility in AI/LLM responses
• Analyzing local business presence and reviews
• Optimizing app store listings or e-commerce product SEO

Do not use this skill when

• Writing application code or building MCP servers — use build-mcp-use-server instead
• Reviewing a pull request — use review-pr instead
• Researching a coding bug or library choice — use run-research instead
• Planning work without running SEO tools — use plan-work instead
• Auditing a local HTML snapshot, file:// page, or private localhost preview with no public ingress. This skill requires a publicly reachable HTTP/HTTPS target or a different local audit workflow.

Runtime assumptions and preflight

This skill assumes the MCP Marketers tool surface is already connected, authenticated, and billable in the current runtime. Before Step 0, confirm the checks that apply to the selected workflow:

Check	What must be true before the first analysis call
Tool surface	The runtime exposes the workflow's starting tool and any mandatory follow-up helpers named in the first reference file. If the user wants a compiled deliverable, also confirm start-report-workflow and compile-report. If you choose an async mode such as mode: "task", also confirm manage-tasks.
Input type	Site crawl workflows need a publicly reachable HTTP/HTTPS target. Domain/business/app/product workflows need a valid identifier for that tool family (domain, place ID/query, appId, ASIN/product ID). Keyword/SERP/trends/AI-query workflows can run with keywords or queries only and do not require a public site unless later steps add one.
Billing	The current environment is allowed to make paid DataForSEO-backed calls for the workflow you plan to run.

Hard blockers: missing starting tool, unsupported input type, local-only crawl target, or no billing authority. Stop before the first paid analysis call when any hard blocker is present. If a later helper tool is missing, continue only when the workflow reference explicitly gives a fallback; otherwise stop and report the missing tool instead of improvising with unrelated tools.

Core operating rules

1. Resolve geography first. If the user mentions a non-US location, run resolve-geo before any analysis tool. Most tools silently default to US/English — wrong geo wastes paid API calls. resolve-geo itself is a free setup step.
2. Start report workflows for deliverables. Run start-report-workflow before analysis if the user wants a compiled report. All tool outputs auto-capture. Compile at the end with compile-report. These report helpers are free orchestration steps.
3. Follow nextSteps without abandoning the base workflow. Tool nextSteps refine the current workflow; they do not replace the Workflow Selector. Finish the core sequence first, then take high-priority nextSteps that stay inside the same workflow branch.
4. Batch keywords. One call with 500 keywords is far cheaper than 500 calls with 1 keyword. Use analyze-keywords and research-keywords with arrays.
5. Use task mode for SERP. analyze-serp(mode: "task") saves 50-80% over mode: "live". Use manage-tasks to poll results.
6. Set explicit limits. Defaults are generous (100-1000). Lower limit when you only need top results.
7. Handle large result sets. When a tool returns >50 rows and creates a dataset handle, use read-result-set to paginate/filter/aggregate. Use export-result-set for downloads.
8. Never orphan a tool call. A single audit-site or research-keywords without follow-up tools is incomplete analysis. Chain to the next step.
9. Confirm before expensive operations. Ask the user before running audit-site with maxPages > 200 or enableJs: true, analyze-serp in live mode on >5 keywords, or analyze-ai-presence(dataType: "llm_response").
10. Track costs. Analysis responses include cost data. Report estimated spend after multi-tool workflows and distinguish free orchestration calls from paid analysis calls.

Workflow selector

Before starting, identify the user's goal in this table. Read the first reference file listed, then follow the tool sequence.

These tool sequences omit start-report-workflow and manage-tasks for brevity. Insert start-report-workflow before the first analysis call whenever the user wants a compiled deliverable. Insert manage-tasks whenever you deliberately choose an async/task mode and need to wait for the result before continuing.

User goal	Start with	Tool sequence	Key references (read first before starting)
Site health / technical audit	audit-site	resolve-geo -> audit-site(instant) -> audit-site(lighthouse) -> audit-site(redirect_chains, duplicate_tags) -> analyze-domain(overview) -> compile-report	references/workflows/site-audit.md
Find keyword opportunities	research-keywords	resolve-geo -> research-keywords(related) -> analyze-keywords(difficulty) -> analyze-keywords(intent) -> analyze-serp(organic) -> find-keyword-gaps -> compile-report	references/workflows/keyword-research.md
Competitor intelligence	analyze-competitors	resolve-geo -> analyze-competitors(domain) -> compare-domains(overview) -> find-keyword-gaps -> compare-backlinks -> find-link-opportunities -> compile-report	references/workflows/competitor-analysis.md
Backlink profile / link building	analyze-backlinks	resolve-geo -> analyze-backlinks(backlinks) -> analyze-backlinks(spam_scores, referring_domains, anchors) -> track-backlinks(growth_summary) -> find-link-opportunities -> compile-report	references/workflows/backlink-analysis.md
SERP features / ranking landscape	analyze-serp	resolve-geo -> analyze-serp(organic) -> analyze-serp(people_also_ask, related_searches) -> analyze-keywords(difficulty) -> compile-report	references/workflows/serp-analysis.md
Track ranking positions	track-rankings	resolve-geo -> track-rankings(current) -> track-rankings(changes) -> analyze-serp(organic, for movers) -> compile-report	references/workflows/rank-tracking.md
Content strategy / gaps	analyze-content	resolve-geo -> analyze-content(search) -> research-keywords(related) -> analyze-keywords(intent) -> find-keyword-gaps -> generate-content(subtopics) -> compile-report	references/workflows/content-optimization.md
AI/LLM brand visibility	analyze-ai-presence	analyze-ai-presence(mentions) -> analyze-ai-presence(top_domains) -> analyze-ai-presence(keyword_data) -> analyze-serp(ai_overviews) -> compile-report	references/workflows/ai-presence.md
Local business SEO	analyze-business	resolve-geo -> analyze-business(profile) -> analyze-business(reviews) -> analyze-serp(local_pack) -> research-keywords(related) -> compile-report	references/workflows/local-seo.md
App store optimization	analyze-app	resolve-geo -> analyze-app(info) -> analyze-app(keyword_rankings) -> analyze-app(competitors) -> analyze-app(reviews) -> compile-report	references/workflows/aso.md
E-commerce product SEO	analyze-products	resolve-geo -> analyze-products(search) -> analyze-products(product_info, reviews) -> platform-specific follow-ups -> research-keywords(related) -> compile-report	references/workflows/ecommerce-seo.md
Market trends / rising topics	analyze-market-trends	resolve-geo -> analyze-market-trends(rising_searches) -> research-trends(explore) -> research-keywords(related) -> analyze-keywords(difficulty) -> compile-report	references/workflows/trends-analysis.md
Full comprehensive report	Combine workflows	start-report-workflow -> site-audit sequence -> keyword sequence -> backlink sequence -> competitor sequence -> compile-report	Read all relevant workflow refs

Execution sequence

0) Route and orient

1. Confirm the MCP Marketers tool surface is available for the chosen workflow. Minimum surface: the starting tool, any mandatory helpers in the first workflow reference, compile-report plus start-report-workflow if the user wants a compiled deliverable, and manage-tasks if you plan to use async/task modes.
2. Confirm the chosen workflow's input is supported. Crawl-based site audits need a publicly reachable HTTP/HTTPS target. Domain/business/app/product workflows need the correct identifier format. Keyword/SERP/trends/AI-query workflows do not need a public site unless later steps introduce one.
3. Find the user's goal in the Workflow Selector table.
4. Read the first reference file listed for that workflow.
5. If the user specifies a non-US location, run resolve-geo first.
6. If the user wants a deliverable, run start-report-workflow first.

1) Run the primary tool

Call the starting tool from the Workflow Selector with appropriate parameters. Consult references/guides/parameter-guide.md for non-obvious parameter choices.

2) Chain to follow-up tools

Follow the tool sequence from the Workflow Selector. At each step:

• Check the tool response's nextSteps for additional relevant tools
• Use read-result-set if the response created a dataset handle
• Adjust parameters based on results (e.g., narrow limit after initial broad scan)

Use this prioritization rule:

• First: complete the static tool sequence for the chosen workflow
• Second: take nextSteps marked or described as direct continuations of the current phase
• Third: defer optional branch suggestions until the core workflow is complete

Example nextSteps shape:

{
  "nextSteps": [
    {
      "tool": "audit-site",
      "params": { "dataType": "lighthouse", "pageUrl": "https://example.com/" },
      "reason": "Validate performance and CWV after the instant crawl found critical issues",
      "priority": "high"
    },
    {
      "tool": "audit-site",
      "params": { "dataType": "duplicate_tags", "taskId": "task_123" },
      "reason": "Inspect duplicate metadata from the same crawl",
      "priority": "medium"
    }
  ]
}

In that example, run the high-priority lighthouse step if it matches the current site-audit branch, then continue with other site-audit follow-ups. Do not jump to unrelated workflows just because they appear in nextSteps.

3) Interpret results with domain knowledge

Do not return raw tool output. Interpret findings using SEO domain expertise from references/guides/seo-domain-knowledge.md. Flag severity levels: critical issues that need immediate action vs. medium-priority optimizations vs. nice-to-haves.

4) Compile and deliver

Run compile-report to gather all tool outputs into a unified report. Summarize key findings, prioritized actions, and estimated API cost.

Tool quick-reference

When unsure which tool to use, consult references/guides/tool-selection-matrix.md for the full mapping of user questions to tools and follow-ups.

When optimizing for API cost, consult references/guides/cost-optimization.md for tier-based cost strategy and specific savings patterns.

Pitfalls

Pitfall	Fix
Forgetting geo resolution	Run resolve-geo before any location-sensitive tool
Using mode: live for bulk SERP	Use mode: task — saves 50-80%
Unbounded keyword lists	Set limit, minVolume, maxDifficulty
Running audit-site with default maxPages on huge sites	Set maxPages: 50 for quick, maxPages: 200 for standard
Comparing >5 domains at once	compare-domains and compare-backlinks cap at 5
Ignoring nextSteps	Use high-priority nextSteps that continue the current workflow branch after the base sequence is underway
Not starting a report workflow	Call start-report-workflow first if user wants a deliverable
Keyword research without intent check	Always follow with analyze-keywords(dataType: intent)
Confusing the two trends tools	analyze-market-trends = "what's trending?" (discovery); research-trends = "how is this keyword trending?" (tracking)
Running expensive tools without confirmation	Ask before audit-site(enableJs: true), analyze-serp(mode: live, bulk), or analyze-ai-presence(llm_response)

Reference routing

Workflows

File	Read when
references/workflows/site-audit.md	Technical SEO audit, site health, page speed, crawl issues
references/workflows/keyword-research.md	Keyword ideas, search volume, difficulty, intent analysis
references/workflows/competitor-analysis.md	Competitive intelligence, domain comparison, keyword gaps
references/workflows/backlink-analysis.md	Link profile analysis, link building, disavow preparation
references/workflows/serp-analysis.md	SERP features, ranking landscape, search results analysis
references/workflows/rank-tracking.md	Ranking position monitoring, change detection
references/workflows/content-optimization.md	Content strategy, content gaps, topic clusters
references/workflows/ai-presence.md	AI/LLM brand visibility, mentions, citations
references/workflows/local-seo.md	Local business SEO, reviews, Google Business
references/workflows/aso.md	App store optimization, mobile app analytics
references/workflows/ecommerce-seo.md	Product SEO, Amazon/Google Shopping optimization
references/workflows/trends-analysis.md	Trend discovery, rising searches, seasonal patterns

Guides

File	Read when
references/guides/tool-selection-matrix.md	Unsure which tool fits the task
references/guides/cost-optimization.md	Planning workflows with many API calls
references/guides/seo-domain-knowledge.md	Interpreting results, advising users, understanding SEO metrics
references/guides/parameter-guide.md	Setting non-obvious parameters, understanding defaults and limits