AWRSHIFT research: AI content operations and site intelligence
-
Editor’s pick
What my MCP server actually did vs what the docs promised
A field report from a live MCP connector: what the protocol documentation promised, what the server actually did, and the marker-string test that settles the difference.
Categories
Latest research
-
How to measure SEO edits without fooling yourself
Clicks are falling for everyone, declined pages rebound on their own, and your dashboard takes the credit. One day of adversarial design review: five measurement ideas died, and a better one survived.

-
What an AI Site Manager Found Reading All 884 Pages of One Site
An AI agent read a real 884-page site end to end and produced 203 evidence-backed proposals. Then its own refutation pass cut the biggest number nearly in half. A case study in…

-
llms.txt Adoption Data: Server-Log Analysis (12-Day Study)
We logged every request to our site for 12 days to answer one question: does anything actually read llms.txt? The bots came about a thousand times. The answer is zero.

-
My Agent Isn’t Autonomous. Neither Was the One That Broke Into Hugging Face.
The first runaway agent did not rebel: its humans disabled the brakes and then stopped watching. A close read of the Hugging Face incident, and the miniature twin of every one of…

-
The Scooter and the Sports Car
Everything I ship runs on one agent system I started half a year ago. Chat is a scooter; a real agent system is a sports car. The anchor of the series: the…

-
Fixing Content Decay: Advanced SEO Audit Services for Content-Heavy Sites
How to catch content decay before Search Console shows a collapse: the position-improvement paradox, n-gram overlap signals, and a monthly refresh-or-prune triage loop.

-
My Agent Isn’t Autonomous. That’s the Point.
What I actually do with my agent, why it is not autonomous, and the minimal rules that let a non-technical person lead a real agent system.

-
Refresh, Rewrite, or Redirect: A Content Audit Decision Matrix with Real Thresholds
A decision matrix with real thresholds for content triage: which impression, CTR, and link signals send a page to refresh, rewrite, redirect, or prune, and in what order.

-
Query Fan-out: Why Optimizing Passages Beats Optimizing Keywords in AI Search
How AI search engines split one query into sub-intent queries, why chasing the exact fan-out strings fails, and how to write self-contained passages that survive retrieval and earn citations.

-
Does Google Penalize AI Content? The Engineering-Grade Guide for US Content Leads
Does Google penalize AI content? Discover Google’s actual stance, the risk of scaled content abuse, and the exact quality gates needed to ensure SEO success.

