keyword-research

Overview of keyword-research skill

What the keyword-research skill does

The keyword-research skill helps an AI agent turn a vague SEO goal into a structured keyword plan: target terms, intent classification, prioritization, and topic-cluster recommendations. It is built for users who need more than “give me keyword ideas” and want a repeatable workflow for deciding what to publish, target, or deprioritize.

Who this skill is best for

This keyword-research skill is a strong fit for content marketers, SEO leads, founders, agencies, and product teams that need content opportunities tied to business goals. It is especially useful when you already know your market but need the agent to organize demand, intent, and opportunity into an actionable list.

The real job to be done

Most users are not actually trying to “find keywords.” They are trying to answer:

• what topics can drive qualified traffic
• which keywords are realistic to target
• how to group terms into clusters instead of isolated blog posts
• what to publish first

That is where keyword-research is more useful than a generic brainstorming prompt.

What makes this keyword-research skill different

The repository includes practical support files that improve output quality:

• references/keyword-intent-taxonomy.md for consistent intent mapping
• references/keyword-prioritization-framework.md for scoring and triage
• references/topic-cluster-templates.md for turning keywords into content architecture
• references/example-report.md for deliverable shape and level of detail

This makes the skill more install-worthy if you want a research workflow, not just a list of terms.

When not to install it

Skip this keyword-research skill if you only need a few brainstorming ideas or if you expect the skill to fetch live search metrics by itself without tool access. The skill can structure and reason well, but real search volume, difficulty, and SERP conditions still depend on external SEO data sources or your own supplied metrics.

How to Use keyword-research skill

Install context and compatibility

The repository declares compatibility with Claude Code ≥1.0, skills.sh marketplace, ClawHub marketplace, and the Vercel Labs skills ecosystem. No system packages are required. Optional MCP network access helps if you want the agent to pull data from SEO tools.

If you use a marketplace-style installer, the baseline command is:

npx skills add aaron-he-zhu/seo-geo-claude-skills --skill keyword-research

Read these files first

For a fast evaluation path, read:

1. SKILL.md
2. references/example-report.md
3. references/keyword-intent-taxonomy.md
4. references/keyword-prioritization-framework.md
5. references/topic-cluster-templates.md

That file order tells you how the skill thinks, what output to expect, and how to judge whether the keyword-research usage matches your workflow.

What input the skill needs to work well

The quality of keyword-research output depends heavily on input completeness. Provide:

• website or company description
• product or service categories
• target audience
• geography and language
• business goal such as leads, trials, signups, or affiliate clicks
• known competitors
• current domain strength or ranking reality
• whether you want quick wins, cluster planning, or a full roadmap

Without this context, the skill may produce plausible but weak keyword sets.

Turn a rough goal into a strong keyword-research prompt

Weak prompt:

• “Find keywords for my SaaS”

Better prompt:

• “Run keyword-research for a B2B invoicing SaaS for US freelancers. Prioritize low-to-medium difficulty keywords with commercial or high-intent informational intent. Group results into topic clusters, show quick wins vs longer-term targets, and suggest content formats. Assume our domain is new and we need signup-driven traffic.”

The stronger version improves prioritization because it gives the agent business relevance, market scope, and ranking constraints.

Best workflow for first-time keyword-research usage

A practical workflow:

1. Define business goal and audience.
2. Provide seed topics or product categories.
3. Ask for intent classification and cluster expansion.
4. Ask the agent to prioritize by difficulty, relevance, and conversion potential.
5. Convert shortlisted keywords into a content plan.

This sequence matches the repository’s support materials better than asking for one giant keyword dump.

Use the example report as your output contract

references/example-report.md is valuable because it shows the expected deliverable shape: executive summary, top opportunities, quick wins, growth terms, and prioritized recommendations. If you want consistent outputs across projects, tell the agent to follow that report structure.

How to handle live metrics and tool data

This keyword-research skill appears designed to work with optional SEO tool integrations, but not to guarantee them. In practice, use one of these modes:

• supply exported keyword data from Ahrefs, Semrush, or Google Keyword Planner
• let the agent reason qualitatively when no tool data is available
• ask for assumptions to be explicitly labeled when metrics are estimated

If adoption depends on metric accuracy, verify the tool path before relying on the skill operationally.

Practical prompt pattern for better clustering

If your main goal is content planning, ask for:

• primary keyword
• secondary variants
• intent
• pillar vs cluster role
• suggested content format
• business relevance
• priority score

That aligns well with the repository’s prioritization framework and topic-cluster templates.

How to use keyword-research for a new site

For new or low-authority domains, tell the agent to bias toward:

• long-tail terms
• lower-difficulty opportunities
• narrow problem statements
• comparison and use-case content
• cluster pages that support one realistic pillar

Otherwise the output may over-index on obvious head terms you are unlikely to rank for soon.

How to use keyword-research for an existing content library

If you already have content, ask the skill to map:

• existing URLs to keyword intent
• missing subtopics in a cluster
• cannibalization risks
• consolidation candidates
• refresh opportunities

This is often a better use of keyword-research than net-new ideation because it ties recommendations to assets you can improve immediately.

Common constraints to know before installing

The biggest adoption blockers are not installation issues but expectation gaps:

• the skill is not a substitute for live SERP validation
• generic inputs lead to generic keyword sets
• priority scoring is only as good as the business context provided
• multilingual support in triggers exists, but your prompt still needs clear market and language scope

keyword-research skill FAQ

Is this keyword-research skill better than a normal AI prompt

Usually yes, if you need structure. A normal prompt can brainstorm terms, but this skill adds intent taxonomy, prioritization logic, cluster planning, and an example reporting format. Those reduce guesswork and make outputs easier to act on.

Does keyword-research include real search volume data

Not by itself. The repository signals optional SEO tool integrations, but you should assume live metrics require external tool access or user-supplied data. If you need defensible numbers, pair the skill with your keyword exports.

Is the keyword-research skill beginner-friendly

Yes, with one condition: beginners should provide clear business context and use the example report format. The underlying concepts are accessible, but better results come when you tell the skill what “success” means for your site.

When should I not use keyword-research

Do not use it as your only decision system for high-stakes SEO bets. It is strongest for research framing, prioritization, and content planning. You still need manual review for SERP reality, brand fit, and final editorial choices.

Can this keyword-research skill help with topic clusters

Yes. That is one of its stronger practical uses because the repository includes references/topic-cluster-templates.md. If your goal is topical authority rather than one-off posts, this skill is more valuable than simple keyword brainstorming.

Is it suitable for agencies and teams

Yes. The example-report pattern makes outputs easier to standardize across clients or internal stakeholders. Agencies can use it to create first-pass research briefs, then layer in live tool data and competitive review.

How to Improve keyword-research skill

Give sharper business constraints

The fastest way to improve keyword-research output is to be explicit about what matters most:

• lead generation vs awareness
• local vs national reach
• new domain vs established domain
• product-led vs editorial-led conversion
• short-term wins vs long-term authority

These constraints change what “good keywords” means.

Provide seed terms that reflect your market

Do not start with only broad nouns like “software” or “marketing.” Give 5 to 15 seed terms tied to real pain points, use cases, buyer language, and product categories. That helps the skill expand in the right semantic neighborhood.

Ask for assumptions to be labeled

A common failure mode is confidence without data. Improve trust by asking the skill to separate:

• confirmed data
• inferred estimates
• strategic assumptions
• items needing external validation

This is especially important when using keyword-research without connected SEO tools.

Force prioritization, not just ideation

Many weak outputs happen because users ask for “100 keywords” instead of decisions. Ask the agent to rank keywords by:

• business relevance
• realistic difficulty
• intent quality
• content gap value
• cluster contribution

That makes the output usable for publishing decisions.

Iterate on intent mismatches

If the first draft mixes educational, commercial, and navigational terms too loosely, tell the skill to reclassify using references/keyword-intent-taxonomy.md and prune low-fit intent categories. This usually improves both content targeting and conversion alignment.

Improve cluster quality after the first pass

After the initial keyword-research run, ask:

• which pillar pages deserve clusters
• which subtopics are redundant
• which pages should target comparison intent
• which supporting pieces can internally link to money pages

This step turns a keyword list into a working content architecture.

Sanity-check the prioritization model

The repository includes a clear prioritization framework, but your business may weight factors differently. If conversions matter more than traffic, tell the skill to increase the importance of business relevance and intent match over pure volume.

Use examples to shape the output you want

If the first report is too abstract, point the agent to references/example-report.md and ask it to match that level of specificity. Referencing a concrete example usually improves format consistency and usefulness faster than broad feedback.

Re-run keyword-research after real-world feedback

The best keyword-research guide is iterative: publish a few pieces, observe ranking and conversion behavior, then ask the skill to refine clusters and priorities using actual results. Once performance data exists, the skill becomes more useful because it can reason from evidence instead of assumptions.