How to Find Content Ideas with Claude Using Real Data

The best content ideas come from questions people already ask. Give Claude three inputs: your Search Console queries, keyword data from a connected tool, and real customer questions from emails, calls and comments. Ask it to group them by question, score each idea for demand, fit and effort, and turn the top ten into briefs.

  • By Praveen Jha · Updated 2026-10-22 · 8 min read · Content & writing
Best for
Founders, coaches and marketers who struggle with what to publish
Time
30-45 minutes for a quarter's plan
Cost
Free to start; keyword volumes may need a paid tool
Works in
claude.ai, Claude Desktop, Claude Code
Level
Beginner

Why "give me 50 blog ideas" fails

Ask Claude for ideas with no data and you get the same topics everyone else gets. They are not wrong, but nothing connects them to your audience or your search demand. Feeding Claude real signals changes the output from generic to specific.

Three data sources to collect

  • Search Console queries: what people already search before landing on your site, including queries where you appear but rarely get clicked. Connect it to Claude or export the Queries report as CSV.
  • Keyword data: search volume and related questions from a keyword tool connected through MCP, so demand is measured rather than guessed.
  • Customer questions: paste questions from sales calls, enquiry emails, DMs, course Q&A and comments. These are the most valuable and the least used.

Step 1: Build a question bank

Put all three sources into one chat, or into a Project if you plan monthly. Ask Claude to rewrite every query and message as a plain question, merge duplicates, and keep a count of how often each one appears and where it came from.

Step 2: Score the ideas

  1. Demand (1-5): search volume, impressions in Search Console, or how often customers ask.
  2. Fit (1-5): how close the question is to something you sell or teach.
  3. Effort (1-5, lower is easier): can you answer it from experience today, or does it need research?
  4. Gap: do you already have a page that answers it? If yes, the idea is an update, not a new post.

Sort by demand plus fit minus effort. The top of the list is usually a mix of quick wins from Search Console and high-fit questions from customers.

Step 3: Turn the top ideas into a calendar

Ask Claude to group related ideas into topic clusters, choose one pillar page per cluster, and schedule supporting posts that link back to it. Cap a first quarter at 10-12 pieces you can actually finish.

Check the numbers: When Claude reports volumes or impressions, ask which tool or file each number came from. Treat anything without a source as an estimate.

Applying this to a services business

For a tutor or consultant, enquiry emails are a goldmine: each question a prospect asks before buying is a page that can answer the next prospect. Those pages also tend to convert better than broad top-of-funnel posts, because they meet people close to a decision.

Copy-paste prompts

Build the question bank

Here are my Search Console queries [attach CSV], keyword ideas [paste], and customer questions [paste]. Rewrite each as a plain question, merge duplicates, and return a table with: question, source(s), times seen, and impressions or volume where available.

Score and rank

Score each question 1-5 for demand, fit with my offer [describe offer], and effort (5 = hardest). Mark whether an existing page already answers it: [paste site map]. Sort by demand + fit - effort and show the top 20.

Quarterly calendar

Group the top 20 into topic clusters. Pick one pillar page per cluster and up to three supporting posts. Return a 12-week calendar with title, target question, cluster and the internal links each post should include.

Troubleshooting

  • Every idea looks the same You gave Claude too little input. Add customer questions; they bring the specifics that search data alone misses.
  • Ideas you can't actually write well Raise the weight of fit and effort. Publish what you have real experience with first.
  • New posts compete with old ones Always give Claude your site map and ask it to flag ideas an existing page already covers.
  • Plan is too big to finish Limit the first plan to 10-12 pieces and review results before planning the next batch.

FAQ

Can Claude come up with content ideas?

Yes, but the ideas are far better when you give it real data: Search Console queries, keyword volumes and customer questions. Without data, it suggests generic topics.

Where do the best content ideas come from?

From questions people already ask: searches that bring them to your site and questions prospects ask in emails, calls and comments. These show proven demand.

How do I prioritise content ideas?

Score each idea for demand, fit with what you sell, and effort to produce. Check whether an existing page already answers it, then start with high-demand, high-fit, low-effort ideas.

Do I need paid tools?

No. Search Console is free and customer questions cost nothing. A keyword tool helps measure volume, and several offer free tiers.

Related pages