Claude Keyword Research: Build and Review a Search Plan

Claude does not have access to live search volume, keyword difficulty, or trend data. What it does well is the analysis layer on top of that data: grouping a large keyword list by intent, identifying cannibalization risks, mapping keywords to the right page type, and drafting an editorial plan. That work takes hours done manually; Claude can do it in minutes when you provide the measured data.

The workflow is: gather real keyword data from a tool that has it, export it as a list, then brief Claude to analyze and organize it. Tools like fenbs keep the scope decisions and approved article tasks in one place so a new session picks up the plan rather than restarting the analysis.

What you need before briefing Claude

You need keyword data with at least volume estimates and the queries themselves. Export this from a keyword research tool: Google Search Console (for your existing traffic), Google Keyword Planner, Ahrefs, Semrush, or a similar service. Claude cannot retrieve this data itself.

Also prepare a clear scope brief: which topic or site section are you planning for, who is the audience, and what action should a visitor take. Claude's grouping and intent review will be much more useful with a defined scope than against a general list of thousands of queries.

Step 1: Group keywords by intent

1. Paste your keyword list (up to a few hundred at a time works well) into Claude.
2. Ask Claude to group them by search intent: informational (learning), navigational (finding a specific page), commercial (comparing options), and transactional (ready to act).
3. Within each intent group, ask Claude to identify clusters — sets of closely related queries that could be served by one page.
4. Review the clusters. Claude may combine queries you want to keep separate or split ones that belong together. Adjust the groupings based on your knowledge of the topic.

Step 2: Review for cannibalization

Ask Claude to flag clusters that are similar enough to risk cannibalization — two pages competing for the same query. This is useful when you are planning new content alongside existing pages. Give Claude a list of your current URLs and their target topics alongside the keyword clusters, and ask it to identify where a new page would compete with an existing one.

Claude's assessment here is a starting point. It can identify surface-level overlaps in the keyword language, but whether two pages actually cannibalize each other in your specific site depends on factors like domain authority, page structure, and how the queries behave in practice. Verify the clusters it flags before deciding to consolidate or leave them separate.

Step 3: Map to page types

Ask Claude to recommend a page type for each cluster: a blog post, a landing page, a comparison page, a glossary entry, a how-to guide. It makes this recommendation based on the likely intent behind the queries. Review each suggestion against your editorial standards and the competitive landscape you have observed — what page types are ranking well for these queries.

Step 4: Draft the editorial plan

With clusters mapped to page types, ask Claude to draft a prioritized editorial calendar. Tell it your publishing cadence, which topics are most commercially important, and whether there are any seasonal or timely priorities. Ask it to produce a list of working titles, target keywords, and content types in a table.

Review the plan against your capacity and business priorities. Claude does not know your team size, your publishing constraints, or which topics have internal knowledge behind them. Those filters are yours to apply. The editorial plan it drafts is a structured starting point, not a final schedule.

Checking intent accuracy before assigning a page type

Claude's intent groupings are based on the text of the queries, not on observed click behavior. A query that looks informational ("what is the best project management tool") may actually have strong commercial intent when you look at the pages that rank for it. Before assigning a page type to a cluster, cross-reference Claude's suggestion with a manual look at the top-ranking pages for the most important query in that cluster. If the ranking pages are comparison pages and Claude suggested a blog post, adjust the recommendation.

Using Claude for ongoing keyword review

Once you have a plan, Claude is useful for periodic reviews: paste new queries from Search Console that have emerged since the last review and ask it to slot them into the existing cluster structure or flag a new cluster. Ask it to identify which planned articles most closely match newly emerging queries.

Keep a record of the scope decisions you made — which topics are in, which are out, and why. That context makes the next review faster and keeps different sessions consistent. A decision made in March ("we are not targeting broad AI terms, only Claude-specific ones") is still useful context in September.

How to do this with fenbs

1. Create a free account and add a project for the keyword strategy.
2. Record the scope decisions as Decisions and rules: which topics are in, which are out, and the intent types you are prioritizing.
3. Add each planned article as a task, with the target keyword cluster and the page type.
4. Connect Claude from Integrations so it reads the scope and approved plan before suggesting new clusters or titles.
5. After each keyword review session, update the tasks and record any new scope decisions.