5 Claude Prompts That Run Your SEO on Autopilot

Most SEO advice assumes you have hours to spare. You probably don't. This guide covers five specific Claude prompts that do the analysis, surface the wins, and draft the content, so you spend time acting instead of digging through spreadsheets.

The underlying approach is borrowed from a founder who documented growing from 0 to 116K organic clicks, with a significant portion of the workflow now running through Claude instead of manual work.

Why Claude over a dedicated SEO tool

Tools like Ahrefs and Semrush are built for SEO professionals. They're deep, expensive, and require you to know what question to ask before you open them.

Claude connected to your own Search Console and GA4 data is different. You're feeding it your actual traffic numbers, not a third-party crawl estimate, and asking it to reason over what's happening. The outputs aren't generic recommendations. They're specific to your site, your queries, your CTRs.

According to Ahrefs, 96.55% of pages get zero organic traffic from Google. The gap between ranked and not-ranked usually isn't content quality. It's that no one looked at what queries were almost ranking and pushed them over the edge. Claude can do that scan in under a minute.

Step 0: Connect Claude to your Search Console and GA4 data

You need a paid Claude plan (Pro, Max, or Team) to use custom connectors. The free tier won't support this.

Once you're on a paid plan:

1. Go to Settings → Connectors in Claude
2. Add this MCP server URL: https://api.babylovegrowth.ai/api/mcp
3. Sign in with Google and allow read access to your Search Console and GA4 properties

The connection is read-only. Claude can't publish anything or change your site settings.

If you'd rather not set up the connector, export your Search Console data as CSV (Performance → Export) and paste it directly into the chat. Every prompt below works either way.

Prompt 1: GSC quick wins

This prompt finds the rankings you're most likely to improve this quarter with the least effort.

Pull last 28 days from Search Console. List queries ranked 5-15 with 200+ impressions, sorted by position x impressions. For each: the ranking URL, current CTR, and the one on-page change to make.

Positions 5-15 are the sweet spot. You're already indexed and getting impressions, which means Google thinks your page is relevant. A better title tag, a clearer H1, or one added section can move a position-9 result to position-4, which roughly triples clicks at the same impression volume.

Moz data shows that position 1 captures around 27% of clicks while position 5 captures about 7%. Moving three spots on a query with 1,000 monthly impressions is worth more than writing an entirely new article chasing a keyword you currently don't rank for at all.

Claude will return a table with the specific URL and one change per query. Work through the top five results in a single afternoon.

Prompt 2: Query clustering

Publishing one thin page per keyword is a common early-stage mistake. You end up with ten pages fighting each other for the same query.

Pull my top queries from Search Console over the last 90 days and group them into topic clusters. For each cluster: the theme, the queries in it, total clicks and impressions, the page that ranks (or 'no page yet'), and whether it deserves its own hub page or should fold into an existing one. Sort clusters by total impressions.

The output tells you two things: where you have content overlap (consolidation opportunities) and where you have impression volume but no dedicated page (content gaps).

Sorting by total impressions means you're prioritizing by actual search demand, not guessing. If a cluster has 8,000 impressions across six queries and no dedicated page, that's your next article.

Prompt 3: Content drafting from gaps

This prompt chains off what the clustering prompt found.

Find 5 query clusters where we get impressions but no clicks and have no dedicated page. For the top cluster, draft the article: H2/H3 outline, entities to cover, and 5 internal links from existing pages.

The phrase 'impressions but no clicks' matters. It means Google is already surfacing your domain for these queries in some capacity. The search intent exists, the audience exists, you just don't have a page that fully satisfies the query.

Claude's draft won't be publish-ready. Treat it as a first-pass structure. The H2/H3 outline alone saves 30-45 minutes of planning. The internal link suggestions are based on your existing content, so they're relevant rather than generic.

A note on GEO (Generative Engine Optimization): the entity coverage Claude recommends here also improves how your content appears in AI search engines like ChatGPT and Perplexity. According to BrightEdge's 2024 research, AI-generated answers in search engines now appear for over 84% of queries in some categories. Writing content that covers the right entities, not just the right keywords, serves both Google and AI-powered search.

Prompt 4: Title tag checker

High impressions with low CTR means you're showing up but not getting the click. That's a messaging problem, not a ranking problem.

From my Search Console data over the last 90 days, list queries and pages with high impressions but a CTR well below what their average position should earn. For each: the query or page, impressions, current CTR, average position, the CTR I'd expect at that position, and a rewritten title tag plus meta description that better matches intent. Sort by the clicks I'm leaving on the table.

The 'sort by clicks I'm leaving on the table' instruction is the key part. Claude calculates the gap between expected CTR for your position and actual CTR, then ranks results by the raw click volume you'd gain by fixing each one.

Expected CTR benchmarks by position (based on Search Engine Journal analysis of 2024 click data):

• Position 1: ~27%
• Position 3: ~13%
• Position 5: ~7%
• Position 10: ~2.5%

A page sitting at position 3 with a 4% CTR has a title tag problem. Claude rewrites both the title and meta description targeting search intent, not just keyword inclusion.

Prompt 5: Weekly report

This is the one to schedule.

Read this week's Search Console + GA4 data. Report position, impression, and CTR moves vs last week, explain the 3 changes that mattered, and set 3 priorities for next week. Exec-readable, no fluff.

Set a recurring calendar block for Monday at 8am. Run this prompt, read the output, and you have your SEO priorities for the week before you open your inbox.

The 'exec-readable, no fluff' instruction matters. Without it, Claude will give you a verbose breakdown with caveats. With it, you get a short summary you can actually act on.

The report covers:

• What moved (positions, impressions, CTR)
• Why it moved (Claude's interpretation, not just raw numbers)
• What to do next week (three specific actions, not vague suggestions)

Common mistakes when using these prompts

Using too short a date range. Search Console data is noisy week-to-week. The quick wins prompt uses 28 days. The clustering and title tag prompts use 90 days. Don't shorten these windows to 7 days thinking you'll get fresher data. You'll get noisier results and worse recommendations.

Treating drafts as final. The content drafting prompt gives you structure and entity coverage. It doesn't replace your knowledge of your audience or your own voice. Edit the draft before publishing.

Ignoring the 'no page yet' clusters. When clustering returns a group with high impressions and no matching page, that's a direct signal from Google that search demand exists. These are high-probability content investments.

Running prompts without acting on the output. The title tag checker will give you rewritten titles. Copy them into your CMS the same session. If you save the output to revisit later, you probably won't.

What this setup doesn't do

This won't build backlinks for you. Off-page authority still requires deliberate outreach, PR, or content worth linking to.

It also doesn't fix technical issues. If you have crawl errors, slow Core Web Vitals, or broken internal links, Claude can help you analyze the data but it can't push fixes to your site.

Think of this as your weekly SEO analyst, not a full-stack SEO agency. It handles the data interpretation and content planning layer. The implementation is still yours.

Getting started in under 30 minutes

1. Connect the MCP server or export your last 90 days from Search Console as CSV
2. Run Prompt 1 (quick wins) and pick your top three pages to update today
3. Run Prompt 2 (query clustering) to see where your content has gaps
4. Schedule Prompt 5 for Monday 8am as a recurring reminder

The whole setup takes one session. The returns compound weekly as you consistently act on what the data shows.