Short answer

Which research artefacts does toolindex publish?

toolindex separates published evidence datasets, methodological research notes, a public research methodology and predefined study protocols. The research notes now cover measurement methodology, mentions versus citations, technical GEO foundations and content for AI search. The planned enterprise AI experiment remains explicitly without results until measurements are actually run.

Facts from vendor documentation. Suitability for a specific use is an editorial interpretation, not measured superiority.

Research · Decision context

Which research types does toolindex distinguish?

toolindex treats source review, structured datasets and empirical measurements as different evidence types. This prevents a documented vendor statement from being presented like an independent test and a study plan from being presented like a result.

  • Evidence dataset: structured, source-linked statements within a defined research scope.
  • Methodology: rules for sources, interpretation, corrections and limitations.
  • Study protocol: predefined questions, metrics and exclusion rules without a predetermined result.

Research · Decision context

Which research notes have already been published?

Four methodological research notes are published: GEO measurement methodology; mention, recommendation and citation; technical GEO covering crawling, rendering and bot delivery; and content for AI search with a clear anti-spam rule for prompt variants. All four separate documented sources, editorial interpretation and unproven causality. [90][91][92][94]

  • GEO measurement methodology: prompt sets, engines, repetitions and remeasurement.
  • Mention, recommendation and citation: separate measurement objects and source analysis.

Research · Decision context

Which dataset is already published?

The 2026 GEO agency evidence snapshot maps four currently reviewed agencies against five identical criteria. This produces 20 comparison cells. Each cell points to the underlying public sources and keeps limitations next to the statement.

Research · Decision context

Which empirical study has not yet been run?

For enterprise AI, toolindex has a predefined ChatGPT study protocol with 20 discovery questions and 30 planned repetitions per question. That represents 600 planned answer slots per fixed measurement configuration. Until the measurement is run, toolindex publishes no mention, recommendation or citation results from it.

Research · Decision context

When does a protocol become a result?

An empirical result may only be published after the measurement configuration, time window, error handling and evaluation rules are documented and the answers have actually been collected. Until then, the page remains labelled as a study protocol and outside the indexable result set.

Research · Decision context

How can research data be checked or corrected?

Sources and review dates are shown next to the relevant statements. Reports of outdated or unclear information can be submitted through the documented correction process. A submission is not accepted automatically and is checked again against suitable sources before publication.

Evidence

Sources and research limits

4 sources used on this page. Checked means that a public source was read. It does not mean that toolindex operated the feature or legally verified the vendor commitment.

  1. 90
    Google Search Central: Google's Guide to Optimizing for Generative AI Features on Google Search

    · Vendor source

    Reviewed area: Official guidance: SEO fundamentals remain relevant; create unique non-commodity people-first content; avoid separate pages for every query or fan-out variation; crawlable technical structure; Google Search ignores llms.txt and other special AI text files; generative AI visibility can be measured in Search Console.

  2. 91
    Google Search Central: AI Features and Your Website

    · Vendor source

    Reviewed area: Official guidance for AI Overviews and AI Mode: no additional technical requirements beyond Search eligibility; crawling, internal links and textual content remain foundational; no special AI text file or schema is required.

  3. 92
    Google Search Central: Understand JavaScript SEO Basics

    · Vendor source

    Reviewed area: Explains crawl-render-index processing; server-side or pre-rendering remains useful because it is faster for users and crawlers and not all bots can run JavaScript.

  4. 94
    OpenAI: Publishers and Developers - FAQ

    · Vendor source

    Reviewed area: Official publisher guidance: public websites can appear in ChatGPT Search; OAI-SearchBot should not be blocked if content should be discoverable, cited and linked in ChatGPT Search.