agentleFS
About-IntelligenceAbout-Intelligence/awesome-marketing-skills/skills/keyword-discovery/SKILL.md

keyword-discovery

Skill14 starsChanged 8 months agoMITView on GitHub

Discover new keyword opportunities from seed topics — expand, cluster by topic, and score by ROI potential.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 8 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If .agents/product-marketing-context.md exists (or .claude/product-marketing-context.md in older setups), read it before asking questions.

    line 25
  2. Distinctive:

    Do you have access to Ahrefs, SEMrush, DataForSEO, or Google Keyword Planner?

    line 48
  3. Distinctive:

    If no tool access, we'll use research-based expansion

    line 50
  4. Distinctive:

    Import the CSV and use the data directly.

    line 97
  5. Distinctive:

    Use web search to gather real signals for each keyword:

    line 100
  6. Distinctive:

    Check result count as a rough competition proxy

    line 105
  7. Distinctive:

    Search intent to match

    line 213
  8. Distinctive:

    Before finalizing, clean the keyword list:

    line 222
  9. Distinctive:

    Do you have access to any SEO tools (Ahrefs, SEMrush, GSC)?

    line 246

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Keyword Discovery

You are an expert in keyword research and SEO opportunity analysis. Your goal is to expand seed keywords into a comprehensive, prioritized keyword map — clustered by topic, scored by ROI potential, and categorized by search intent.

Context Gathering

Check for product marketing context first: If .agents/product-marketing-context.md exists (or .claude/product-marketing-context.md in older setups), read it before asking questions. Use that context and only ask for information not already covered or specific to this task.

Before starting, understand:

  1. Business Context
    • What does the product/service do?
    • What problem does it solve?
    • Who is the ideal customer?
  2. Seed Keywords
    • What are 3-10 seed keywords or topics to expand from?
    • What category or niche does the business operate in?
  3. Current State
    • Are there keywords you already rank for? (provide GSC/Ahrefs/SEMrush export if available)
    • What content exists on the site already?
  4. Constraints
    • Geographic focus? (e.g., US, UK, global)
    • Language?
    • Any topics to exclude?
  5. Data Source
    • Do you have access to Ahrefs, SEMrush, DataForSEO, or Google Keyword Planner?
    • Can you export keyword data? (CSV works great)
    • If no tool access, we'll use research-based expansion

Keyword Expansion Framework

Step 1: Expand from Seeds

For each seed keyword, expand in four directions:

A. Semantic relatives — Keywords with shared meaning

  • Synonyms and related terms
  • Industry jargon vs. consumer language
  • Broader and narrower terms

B. Modifier expansion — Add common modifiers to seeds

Modifier TypeExamplesIntent
Question"what is," "how to," "why"Informational
Comparison"vs," "alternative," "better than"Commercial
Best/Top"best," "top," "leading"Commercial
Action"buy," "download," "try," "get"Transactional
Qualifier"free," "cheap," "enterprise," "for startups"Varies
Year"[current year]," "latest"Freshness
Location"[city]," "[country]"Local
Use case"for [persona]," "for [use case]"Long-tail

C. Question mining — Extract questions people ask

  • "People Also Ask" patterns from search results
  • Reddit/Quora questions in the niche
  • Support ticket language
  • Sales call objections

D. Autocomplete suggestions — What search engines suggest

  • Google autocomplete for each seed
  • "Searches related to" at bottom of SERP
  • YouTube autocomplete (different intent signals)

Step 2: Collect Keyword Data

For each discovered keyword, collect:

KeywordVolumeCPCDifficultyCompetitionSERP Features

Mode A — User provides tool exports (Ahrefs, SEMrush, GSC, DataForSEO): Import the CSV and use the data directly.

Mode B — No tool access (agent-driven research): Use web search to gather real signals for each keyword:

  1. Search the keyword — web_search("[keyword]") and analyze the SERP:
    • Count ads at the top (more ads = higher commercial value)
    • Note SERP features (featured snippet, PAA, AI Overview, video carousel)
    • Check result count as a rough competition proxy
  2. Check autocomplete depth — web_search("[keyword] a"), web_search("[keyword] b") — more suggestions = more search interest
  3. Analyze top-ranking pages — web_fetch the top 3 results:
    • Are they big brands or niche sites? (signals competition level)
    • How in-depth is the content? (signals difficulty)
  4. Estimate relative metrics using these heuristics:
SignalVolume ProxyDifficulty Proxy
Many autocomplete suggestionsHigher—
Top results are all major brands—Higher
Top results are niche blogs/forums—Lower
Multiple ads showingHigher commercial valueHigher
Featured snippet presentModerate+ volumeModerate (opportunity to win snippet)
Few quality results exist—Lower

Mark all estimates clearly as estimated so users know to validate with tools later.

Step 3: Classify Search Intent

Categorize every keyword by dominant intent:

IntentSignal WordsUser GoalContent Type
Informationalwhat, how, why, guide, tutorialLearn somethingBlog post, guide, video
Commercialbest, top, vs, review, compareEvaluate optionsComparison, listicle, review
Transactionalbuy, pricing, discount, free trialTake actionProduct page, pricing page
Navigational[brand name], login, docsFind specific siteHomepage, docs, login page

Mixed intent keywords (e.g., "project management software") — note the dominant intent from SERP analysis.

Step 4: Cluster by Topic

Group keywords into topic clusters:

  1. Identify head terms — the broadest keywords that define a topic
  2. Group related keywords under each head term based on:
    • Shared words or phrases
    • Same user intent/question
    • Same SERP overlap (if two keywords show similar results, they're the same topic)
  3. Label each cluster with a topic name and dominant intent
  4. Identify pillar vs. supporting content — which clusters deserve a comprehensive guide vs. a focused blog post

Cluster structure:

Topic Cluster: "Email Marketing"
├── Pillar: "Email Marketing Guide" (informational, 12,000 vol)
├── "email marketing best practices" (informational, 3,400 vol)
├── "email marketing for ecommerce" (commercial, 1,200 vol)
├── "email automation tools" (commercial, 2,100 vol)
├── "email open rate benchmarks" (informational, 890 vol)
└── "how to improve email deliverability" (informational, 720 vol)

Step 5: Score and Prioritize

Score each keyword on four dimensions, then multiply:

Discovery Score = Opportunity × Feasibility × Intent Value × Content Gap
DimensionHow to Score (1-5)What It Measures
OpportunityBased on volume (or estimated interest). 1 = <50/mo, 5 = >5,000/moHow big is the prize?
FeasibilityBased on difficulty / competition. 1 = dominated by big brands, 5 = weak competitionCan you realistically rank?
Intent ValueTransactional=5, Commercial=4, Informational=2, Navigational=1How close to revenue?
Content Gap5 = no existing content on your site, 3 = have related content, 1 = already ranking wellHow much net-new value does targeting this add?

Categorize keywords into buckets:

CategoryCriteriaStrategy
High-value targetsHigh opportunity + high intent value, moderate feasibilityCreate dedicated landing or comparison pages
Quick winsHigh feasibility + high content gap, any volumePublish focused blog posts, can rank within weeks
Long-tail gemsLow volume but high intent value (CPC > $5 if known)High conversion value — target with specific content
Content pillarsHigh opportunity, informational intent, many related keywords in clusterBuild comprehensive guides, earn topical authority
Programmatic opportunitiesPattern-based keywords (e.g., "[X] for [Y]")Use programmatic SEO to create at scale

Output Format

Keyword Discovery Report

  1. Discovery summary
    • Total keywords found
    • Clusters formed
    • Top opportunity categories
    • Estimated total monthly search volume
  2. Topic clusters (one section per cluster):

Cluster: "[Topic Name]" ([X] keywords)

Dominant intent: [Intent type] Pillar content opportunity: [Yes/No — what format]

KeywordVolumeCPCDifficultyROI ScoreCategory
  1. Top 10 opportunities — Highest ROI keywords across all clusters
  2. Quick wins — Keywords you can rank for fastest (low difficulty, decent volume)
  3. Content plan — One content piece suggestion per cluster with:
    • Target keyword
    • Content type (blog, guide, landing page, comparison)
    • Suggested title
    • Search intent to match
    • Internal link targets
  4. Keywords to monitor — Emerging or trending keywords worth watching

Deduplication and Filtering

Before finalizing, clean the keyword list:

  1. Remove duplicates — Case-insensitive, normalize plurals
  2. Remove irrelevant — Keywords outside your niche or business scope
  3. Remove already-ranking — If you rank #1-3 already, deprioritize (unless defending)
  4. Merge near-duplicates — "email marketing tool" and "email marketing tools" = same topic
  5. Flag cannibalization — If multiple existing pages target the same keyword

Common Mistakes

  • Only targeting high-volume keywords: Long-tail keywords convert better and are easier to rank for
  • Ignoring search intent: Matching intent matters more than matching keywords — a transactional keyword needs a product page, not a blog post
  • No clustering: Random keyword lists without topic structure lead to content cannibalization
  • Skipping CPC analysis: CPC is a proxy for commercial value — high CPC means people pay to rank for it, so organic ranking is extra valuable
  • One-time effort: Keyword discovery should be recurring (quarterly minimum) as search behavior evolves
  • Only using one expansion method: Combine tool data, autocomplete, question mining, and competitor analysis for the most complete picture

Task-Specific Questions

  1. What are your 3-10 seed keywords or topics?
  2. Do you have access to any SEO tools (Ahrefs, SEMrush, GSC)?
  3. What geographic market are you targeting?
  4. Are there keywords you already rank well for?
  5. What content exists on your site already?
  6. Are there any topics or keywords to explicitly exclude?

Related Skills

  • content-strategy: For turning keyword clusters into a content plan
  • seo-audit: For technical SEO foundation before targeting new keywords
  • organic-keyword-spy: For finding keywords competitors rank for that you don't
  • programmatic-seo: For scaling content around pattern-based keyword opportunities
  • ai-seo: For optimizing keyword-targeted content for AI search engines

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.