The Generative Engine Optimization (GEO) landscape is undergoing a paradigm shift: aggregated data shows ChatGPT is heavily integrating the site: operator into its search queries. Until recently, the share of fanout containing this operator hovered between 0.3% and 0.5%; with the GPT-5.6 rollout in early August, this percentage jumped to 16-17%. This technical evolution reshapes how web content is queried by language models, shifting focus from simple indexing to source structuring.

Add AlexTech.ai asPreferred Source on Google

The quantitative leap in site: operator usage

Promptwatch monitoring, a tool dedicated to tracking chatbot responses, highlights a clear discontinuity. After a brief dip between August 3 and 5, consistent with pre-launch experimentation, operator adoption exploded from August 8 onward. Although the data covers only queries with automated tracking enabled, the trend suggests a significant architectural change in ChatGPT's internal search tool. Simon Willison hypothesizes that the new interface follows a search(query, recency, domains) structure, prioritizing explicit domain control over generic operator use.

Implications for AI visibility strategy

This technical change fits into an already tense context: as reported by AlexTech, brands are already manipulating AI answers through fake posts to boost their presence. Systematic use of site: might paradoxically reduce reliance on high-noise sources like Reddit — a trend confirmed by Promptwatch data showing a decline in the platform's use in searches — but strengthens the need for publishers to optimize their specific domains. The unified search introduced by OpenAI had already marked the beginning of this transition toward more structured information access.

Transparency as an analysis limit

OpenAI's lack of public documentation on system prompt changes makes it difficult to validate internal mechanisms with certainty. Willison notes that leaked prompt collections do not yet show specific updates discouraging Reddit use, leaving room for interpretations based on empirical observations. For IT professionals and business decision-makers, the signal is clear: visibility in generative search engines no longer depends solely on traditional ranking, but on the ability to be explicitly identified and queried by AI models.