Use Search Console data with Claude skills to decide which SEO pages to create, improve or reject

A site with hundreds of pages risks duplicate targets, thin templated pages and unexplained traffic swings if keywords are picked or changed without checking what already ranks.

Keywords come from Google Keyword Planner and are checked by a person for fit and intent. A Search Console MCP connection pulls query and page data into the process, and each decision is recorded with the query, current URL, evidence, intent and a decision to improve, create or reject. Claude SEO-audit and keyword skills check for duplicate targets, canonicals, indexability, thin or templated content and unsupported claims. Traffic drops are diagnosed by comparing equivalent date ranges, then reverting only when evidence supports it. Before and after each deploy the live HTML is checked for title, canonical, robots, H1 and sitemap entry.

The site reported 166K Google clicks from 1.39M impressions over three months, an 11.9% click-through rate and 7.5 average position. The author says AI did not create permanent growth and traffic declined after a July peak.

The post gives no control group or baseline separating AI-assisted work from other work, so the clicks cannot be credited to AI. Costs, model versions, prompts and the cause of the decline are not stated, and the page does not show a year. The author runs the site as a product and says some ideas were reverted without saying which.

A site connected to Google Search Console, a Search Console MCP connection, and Claude with skills or a coding agent. Costs are not stated.

• DEV Community (Mofidul Islam) ↗ Firsthand write-up, publication date unknown

Source published: unknown. Last reviewed here: October 11, 2026. Spot a mistake? Tell us.