← back to guides
    seobeginner15 minutes
    April 23, 2026

    Run a Complete AI SEO Audit With Claude and Your Search Console Data

    A prompt-by-prompt walkthrough of running a full SEO audit using Claude and your real Search Console data. Covers keyword performance, page analysis, content gaps, CTR optimization, trends, and GEO readiness.

    TL;DR

    This guide walks you through six AI SEO audit prompts that you paste directly into Claude with your Search Console connected. No spreadsheets, no SEO tool subscriptions, no manual data pulls. Claude queries your real performance data and delivers analysis that would cost $2,000+ from a consultant.

    What you need: Cogny Solo ($9/mo, 7-day free trial, no credit card required) with Search Console connected. That gives Claude direct access to up to 16 months of your search performance data.

    Time required: 15 minutes for the full audit | Difficulty: Beginner | Sections: 6 audit modules, each with an exact prompt

    Before You Start

    If you have not connected Search Console to Claude yet, follow the Search Console MCP integration guide first. Setup takes about 2 minutes.

    Once connected, Claude can query your search performance data directly through the Cogny MCP server at app.cogny.com/mcp. Every prompt in this guide assumes that connection is live.

    You can run these prompts in Claude Desktop, Claude Code, Cursor, or any MCP-compatible client. The prompts work the same way regardless of which client you use.

    1. Keyword Performance Audit

    What this does: Identifies ranking drops, keyword cannibalization, and quick-win keywords sitting in positions 11-20 that are close to page one.

    The Prompt

    Pull my Search Console data for the last 3 months. I need three things:

1. **Ranking drops**: Keywords where average position has worsened by 3+ positions compared to the prior 3-month period. Sort by lost clicks.

2. **Cannibalization**: Keywords where 2 or more pages are ranking for the same query. Show the query, all ranking URLs, and their respective positions and clicks.

3. **Quick wins**: Keywords in positions 11-20 with at least 500 impressions/month. Sort by impressions descending. These are my best candidates for moving to page one.

Format each section as a table. At the end, give me a prioritized action list — what should I fix first based on traffic impact.

    What Good Output Looks Like

    Claude should return three distinct tables. The ranking drops table highlights your biggest traffic losses. The cannibalization table reveals pages competing against each other. The quick wins table shows exactly where a small ranking improvement delivers real traffic.

    What to Do With the Results

    • Ranking drops: Check if dropped pages have been updated recently. If not, they likely need content refreshes or additional internal links.
    • Cannibalization: Pick one page to be the primary target for each cannibalized keyword. Redirect, consolidate, or differentiate the competing pages.
    • Quick wins: These are your highest-ROI SEO tasks. Even a 2-3 position improvement on a keyword at position 12 can double its clicks. Prioritize content updates, internal linking, and on-page optimization for these pages.

    2. Page-Level Analysis

    What this does: Finds underperforming pages that have strong impressions but weak clicks, and pages that used to perform well but have declined.

    The Prompt

    Analyze my top 50 pages by impressions over the last 3 months from Search Console.

For each page, show: URL, total clicks, total impressions, average CTR, and average position.

Then flag two groups:

1. **High impression, low CTR**: Pages with 1,000+ impressions but CTR below 2%. These have visibility but are not getting clicked.

2. **Declining pages**: Compare the last 3 months to the prior 3 months. Flag any page that lost more than 20% of its clicks.

For each flagged page, suggest a specific reason it might be underperforming (thin content, poor title, wrong intent, etc.) and one concrete action to fix it.

    What Good Output Looks Like

    A full table of your top 50 pages with the two flagged groups clearly separated. Each flagged page should have a specific, actionable recommendation — not generic advice.

    What to Do With the Results

    • High impression, low CTR: These pages rank well but their titles and meta descriptions are not compelling enough. Rewrite them. See Section 4 for a dedicated CTR optimization prompt.
    • Declining pages: Investigate whether the decline is due to algorithm changes, content staleness, or new competitors. Content refreshes with updated data and better structure typically recover lost traffic within 4-8 weeks.

    3. Content Gap Analysis

    What this does: Finds search queries where you are getting impressions but have no dedicated page targeting that topic. These are content opportunities your site is already partially ranking for.

    The Prompt

    Pull all queries from Search Console for the last 3 months where I have at least 100 impressions.

Group these queries by topic cluster. Then identify clusters where:
- I'm getting impressions across multiple related queries
- But no single page on my site ranks in the top 10 for the primary query in that cluster

These are my content gaps — topics where Google already associates my site with the subject, but I don't have a dedicated page to capture the traffic.

For each gap, provide:
- The topic cluster name
- Top 5 queries in the cluster with their impressions and positions
- A suggested page title and URL slug for new content
- Estimated monthly traffic opportunity based on current impressions and expected CTR at position 5

Rank the gaps by traffic opportunity.

    What Good Output Looks Like

    Claude should identify 5-15 content gaps depending on your site size. Each gap should have a clear topic cluster with real query data backing it up. The traffic estimates should be conservative — based on your actual impression data, not keyword tool projections.

    What to Do With the Results

    This is your content calendar for the next quarter. Start with the highest-opportunity gaps. When you create content for these topics, you already have a head start because Google is already showing your site for related queries. Use the specific queries Claude found as H2s and subheadings in your new content.

    4. CTR Optimization

    What this does: Identifies pages ranking well but getting fewer clicks than expected for their position, then generates improved title tags and meta descriptions.

    The Prompt

    Pull my Search Console data for the last 3 months. For each page, calculate the expected CTR based on its average position using these benchmarks:

- Position 1: 28-32% CTR
- Position 2: 15-18% CTR
- Position 3: 10-12% CTR
- Positions 4-5: 6-8% CTR
- Positions 6-10: 2-4% CTR

Flag every page where actual CTR is more than 30% below the expected CTR for its position.

For each flagged page:
1. Show the current top 3 queries driving impressions to that page
2. Analyze why the current title might not be compelling for those queries
3. Write 2 alternative title tags (under 60 characters) optimized for click-through
4. Write 1 meta description (under 155 characters) that includes a clear value proposition and a reason to click

Prioritize the list by potential click gain (impressions x CTR gap).

    What Good Output Looks Like

    A prioritized list of pages with underperforming CTR. Each entry should have specific title and meta description rewrites that address the actual queries driving traffic to that page. The rewrites should feel like something a human copywriter would produce — not generic SEO templates.

    What to Do With the Results

    CTR optimization is one of the fastest ways to increase organic traffic without changing rankings. Implement the title and meta description changes starting with the highest-impression pages. Track CTR changes in Search Console over the following 2-4 weeks. A well-optimized title tag can increase CTR by 20-50% at the same position.

    5. Trend Analysis

    What this does: Surfaces month-over-month and quarter-over-quarter trends across your search performance so you can spot momentum shifts early.

    The Prompt

    Analyze my Search Console data and create a trend report covering:

1. **Monthly summary (last 6 months)**: Total clicks, impressions, average CTR, and average position for each month. Calculate month-over-month change as percentages.

2. **Quarter-over-quarter comparison**: Compare the most recent full quarter to the one before it. Break down the change by:
   - Brand queries vs non-brand queries
   - Top 10 pages by click change (biggest gains and biggest losses)
   - Device split (mobile vs desktop) and any shifts in the ratio

3. **Seasonal patterns**: Look at the full date range available and flag any queries or pages that show clear seasonal patterns. For each, note when the peak and trough typically occur.

4. **Momentum keywords**: Queries where position has improved steadily for 3+ consecutive months. These are your rising stars — content that is gaining authority.

Summarize with 3 key takeaways and recommended actions.

    What Good Output Looks Like

    A structured report with clear data tables for each section. The monthly summary should read like a dashboard. The quarter-over-quarter breakdown should clearly separate brand from non-brand performance — this is critical because brand traffic can mask organic search problems. The momentum keywords section should highlight genuine upward trends, not random fluctuations.

    What to Do With the Results

    • Monthly trends: Use these to set baselines and track the impact of your SEO work over time. Run this prompt monthly to maintain a rolling view.
    • QoQ comparison: If non-brand traffic is declining while brand holds steady, you have a content or authority problem. If mobile CTR is dropping relative to desktop, check your Core Web Vitals.
    • Seasonal patterns: Plan content updates and promotional pushes 4-6 weeks before historical peaks.
    • Momentum keywords: Double down on these. Add internal links, expand the content, and build supporting pages around rising topics.

    6. GEO Readiness Check

    What this does: Evaluates which of your pages are structured to be cited by AI search engines like ChatGPT, Gemini, and Perplexity. GEO (Generative Engine Optimization) is increasingly important as AI-generated answers pull traffic from traditional search results.

    For more background on why GEO matters, read What Is GEO and Why It Matters in 2026.

    The Prompt

    Look at my top 30 pages by impressions from Search Console. For each page, assess its readiness for AI citation (GEO) based on these criteria:

1. **Query type**: Is the primary query informational, navigational, or transactional? Informational queries are most likely to trigger AI-generated answers.

2. **Position risk**: Pages ranking 3-10 for informational queries are at highest risk of losing traffic to AI answers. Flag these.

3. **Structured content signals**: Based on the queries and page URLs, assess whether each page likely has:
   - Clear, direct answers in the first paragraph
   - Structured data / schema markup potential
   - Listicles, tables, or step-by-step formats that AI can easily parse
   - Original data, statistics, or unique insights that AI models would want to cite

4. **Citation opportunity**: For pages that rank well for informational queries, rate the AI citation opportunity as High, Medium, or Low based on query intent and likely content format.

Group the results into three buckets:
- **Optimize now**: High-traffic informational pages at risk — need structural updates for AI citation
- **Monitor**: Medium risk pages to watch as AI search evolves
- **Low priority**: Transactional or navigational pages less affected by AI answers

For the "optimize now" bucket, give specific recommendations to improve each page's citability.

    What Good Output Looks Like

    A clear three-bucket classification of your top pages. The "optimize now" bucket should have specific, structural recommendations — things like adding summary paragraphs, implementing FAQ schema, restructuring content into scannable formats, or adding original data points that make the page worth citing.

    What to Do With the Results

    GEO readiness is about making your content the source that AI models reference. Focus on the "optimize now" pages first. Add concise, factual summary paragraphs at the top of informational pages. Implement structured data where applicable. If you want to track AI-driven traffic to your site, the GEO Conversion Report template in Cogny can monitor visits from ChatGPT, Gemini, Perplexity, and other AI referral sources.

    Running the Full Audit

    You can run all six prompts in a single Claude conversation. The MCP connection persists throughout the session, so Claude builds context as you go. By the time you reach the GEO readiness check, Claude already has a full picture of your search performance.

    The complete audit takes about 15 minutes. Here is the recommended order:

    1. Keyword Performance — establishes your baseline
    2. Page-Level Analysis — identifies which pages need work
    3. Content Gap Analysis — finds new opportunities
    4. CTR Optimization — quick wins on existing pages
    5. Trend Analysis — puts everything in context
    6. GEO Readiness — future-proofs your strategy

    After the first audit, re-run monthly to track progress. The trend analysis prompt is especially valuable when run consistently — it builds a performance narrative over time.

    What This Costs

    This entire audit runs on Cogny Solo at $9/month. That includes Search Console access through the managed MCP server, up to 16 months of historical data, and the ability to connect additional channels like LinkedIn Ads and Bing Webmaster Tools.

    The 7-day free trial requires no credit card. You can run this full audit during the trial and decide if the ongoing monitoring is worth it.

    For comparison, a consultant running a similar audit manually charges $1,500-3,000. Traditional SEO tools that provide some of this analysis cost $100-300/month and still require you to interpret the data yourself. With Claude and your actual Search Console data, you get the analysis and the interpretation in one step.

    Start your free trial at cogny.com and run your first AI SEO audit today.

    Next Steps

    get started

    Try Cogny free

    Connect your marketing platforms to Claude Code in under 2 minutes. No credit card required.