The Semantic Indexing Probe Protocol is a method for testing whether a work’s presence in general search corresponds to its availability in AI-mediated answer systems. It separates layers that are often treated as one: crawl and index status, semantic parsing, entity recognition, retrieval by a summarizer, and actual injection into an answer.
Testing proceeds through seven phases. General-index phases establish whether a canonical URL can be found and semantically parsed. Summarizer phases ask whether the work appears in generated answers and whether that appearance can be confirmed across systems. Delta analysis compares the layers, while pattern mapping records changes over time.
The protocol’s four query tiers create controls around the target. Direct queries should retrieve the object; conceptual queries test project terminology; adjacent queries test plausible but non-obligatory relevance; and control queries test inappropriate surfacing. This design distinguishes absence from over-injection and direct sourcing from secondary discussion.
Results are preserved as evidence rather than inferred from a single answer. The protocol records exact prompts, dates, output text, cited sources, screenshots, and platform differences, making it suitable for repeated longitudinal mapping of a retrieval basin.