One-question keyword research.

Most keyword tools dump 5000 rows on you. This recipe uses your assistant to do the clustering for you so you get a small, ranked list of clusters with the search intent for each. From there, picking the next post is a 30-second decision.

Before you start

A topic in mind

How to run it

Pick the topic
Pick one topic broad enough to span 30 keywords but narrow enough to stay on-brand. Examples: "local SEO for plumbers", "AI marketing for Shopify stores".
Ask for clusters
Ask your assistant to find 30 keywords for that topic, group them into 5 to 7 clusters, and label each cluster with the search intent (informational, transactional, navigational).
Add difficulty
Ask the assistant to add a rough difficulty score for each cluster based on the keywords inside.
Pick the next post
Sort by intent transactional first, then by difficulty ascending. The top cluster is your next post topic.

Copy this prompt

Topic: [topic]. Find 30 keywords. Group into 5 to 7 clusters with the search intent and difficulty for each. Sort clusters by intent transactional first, then difficulty ascending.

Paste this into Claude, Cursor, or ChatGPT after you have connected the Seoable MCP server. New here? Read the MCP setup guide.

Try this recipe

Connect Seoable to your assistant in one command, then run this recipe today.

MCP setup guide →Start a free audit

More recipes

100 blog posts in your voice.

Run a $99 scan, then ask Claude to publish them through MCP. Done in a coffee break.

Find your AI-search blind spots.

See which AI engines cite you and which ignore you, then fix the gaps with one prompt.

Refresh your zombie posts.

List low-traffic posts in your assistant, regenerate the underperformers in your tone.

← Back to all recipes