Method: what is known about AI search and how to test a result
REFORGE 1.0 · method · sources checked 8 September 2026
An AI answer can link to a page, paraphrase its content or mention a brand without a link. These are different events, and the rules of REFORGE keep them apart. This page brings together the definitions, the research the rules rest on, and a way of testing a result that can also show a zero.
What does the research support?
Search remains part of preparing a page for AI answers. Google treats optimization for its AI search features as SEO: it requires no special file, special markup or fixed text chunk size. This describes Google; other assistants have rules of their own. Google Search Central.
Studies test different tasks. One experiment selects a source from a fixed set of documents; another includes retrieval, reranking and answer generation. The original GEO paper measured visibility in an experimental system with a fixed set of documents, and its result belongs to that system rather than to organic visibility. The later SAGEO study shows why document retrieval before answer generation is counted separately. GEO, KDD 2024, SAGEO, KDD 2026.
A July 2026 critical review of 45 studies, released as a preprint, compares methods and their limitations. It contains no universal citation formula — none for sentence length or any other single feature; accurate information, relevant material and clear writing remain the substantive tasks. Critical review.
How do the terms differ?
| Term | Possible meaning | What to specify |
|---|---|---|
| SEO | Work on a website’s presence in search | Pages, queries and search engines |
| GEO | Visibility in generated answers | Page citations, brand mentions or another outcome |
| AEO | Preparing content for direct answers | A search feature, voice response or generated answer |
| AI SEO | Using AI in SEO; sometimes visibility in AI search | The tool being used or the surface being measured |
| LLM SEO, LLMO | Market labels for presence in model responses | Whether web search is involved and what is observed |
| AIO, AISO, GSO | Abbreviations used differently by different authors | The full name and the source’s own definition |
| AI Overviews and AI Mode | Distinct Google Search features | The feature and mode actually tested |
These labels form no settled taxonomy. Ahrefs includes conventional direct-answer features under AEO and describes AI SEO as using AI in SEO work; AIO can also mean AI Optimization. The measurement is set by the surface and the event named before the work, rather than by the label of a service. AEO, AI SEO, Semrush’s use of AIO.
How can a result be tested?
Define the event first. A retrieved document, used material, exact quotation, source link, brand mention, visit and enquiry are counted separately. A link identifies a source; a verbatim copy of the words is a separate event. Google describes searching for material that supports a generated answer, without prescribing a sentence extraction format. Google Search Central.
A reproducible public experiment is built like this. A set of example questions and accessible documents is published in advance. Each run records the question, date, language, country, product and mode, the returned answer and the complete list of available source links; repeats are recorded separately. For overall visibility, a successfully completed query without an AI answer or target link is an observed zero; an access failure or unknown outcome is missing data, reported alongside the denominator. The share of links among received AI answers is a separate conditional metric. A result whose full source list is unavailable is marked as partial.
Page versions are compared on a stable question set with a suitable control: that is how the effect of an edit is separated from changes in the model, the search results or the audience. The conditions are declared in advance, and the conclusion applies to the named sample.
Platform reports also differ. Search Console’s new generative AI report shows impressions in AI Overviews and AI Mode when sufficient site data is available; Bing reports citations across its covered AI surfaces. Each report has its own denominator, so the figures are read separately, and an unavailable report value means missing data rather than a zero. Google, Bing.
Search access and model training are separate settings. OpenAI provides independent controls for OAI-SearchBot and GPTBot; Google’s generative search inclusion control concerns search. The decision on access stays with the site owner. OpenAI, Google.
What does the first STRUGA case show?
STRUGA is a REFORGE client and an affiliated site, so this case is the service’s own observation rather than an independent client experiment. The service page publishes the observation from STRUGA’s two storefronts, the current figures and the remeasurement date. REFORGE’s result and terms.
Revised pages and their open passports document what was checked and changed; they show the course of the work. A conclusion about the effect, with figures and a date, belongs here after the measurement window.
One caveat for the whole page: the public experiment under this protocol has not yet been conducted, and across STRUGA’s two storefronts an effect beyond noise has not yet been shown. Until those data exist, this page describes a method, not a result.
Read what connection includes: REFORGE.