A site cleaned up 363 thin posts on May 23. The penalty arrived June 5. Algorithmic re-evaluation runs on its own clock — usually 14 to 21 days. Here is the editorial decision log that keeps you from misreading the lag.
A reproducible Python pipeline that embeds every page, scores semantic similarity, subtracts links you already have, and outputs a ranked internal-link review queue.
A working Python pipeline for automated search-engine URL submission with IndexNow and the Google Indexing API, plus an honest take on what each channel can and cannot do.
Word count is the most overrated SEO signal in 2026. Sites with 1,500+ word posts get demoted; sites with 800-word posts rank top three. Here is the quality rubric that replaces length as a signal.
A working Python + n8n pipeline that groups thousands of keywords into content topics by intent using embeddings, UMAP and HDBSCAN.
A 1,200-post site lost 99.6% of its traffic in a single day despite averaging 1,250+ words per post. The reason is not depth — it is signature similarity at scale. Here is what Google’s classifier is actually reading.
Build a Python pipeline that turns a target keyword into a data-grounded SEO content brief in under a minute: scrape the SERP, count what ranks, and let an LLM draft the outline from real evidence.
Stop writing a new script for every SEO question. Build a Model Context Protocol (MCP) server that exposes Google Search Console as tools an agent can drive live — with working Python code.
Build a pipeline that finds CTR-decayed pages in Search Console, rewrites their title tags and meta descriptions with an LLM, QA-gates the output, and ships the winners.
Choosing an embedding model for SEO isn’t a leaderboard question. A practitioner’s framework for OpenAI vs Voyage vs Cohere vs BGE, with a reproducible AUC benchmark, the dimension trade-off, and the cost math at scale.