How does the CEAVERS Index measure AI search visibility?

CEAVERS issues a fixed bucket of prompts against six LLMs in five languages each quarter and records whether and how named brands appear in each response.

Which LLMs are covered?

ChatGPT (and ChatGPT Search), Claude, Gemini, Perplexity, Microsoft Copilot, and Apple Intelligence.

Is CEAVERS independent of vendors?

Yes. CEAVERS does not accept payment for evaluation and has no commercial affiliation with any vendor measured by the Index.

Research Methodology — v1.0

v1.0 — Published 2026-05-12. This page documents the structure of the CEAVERS methodology. Full text is under editorial review and will replace these section stubs before the first official Index release.

Methodology at a glance

Stub. Final version: scope, sample design, update cycle, and independence statement in tabular form.

What we measure

Stub. Final version: definitions of brand mention, citation, source attribution, and language-of-response.

Sample design

Stub. Final version: LLM coverage (ChatGPT, Claude, Gemini, Perplexity, Copilot, Apple Intelligence), prompt template construction, language coverage (EN, IT, ES, FR, PT), and query-bucket stratification.

Scoring approach

Stub. Final version: per-criterion weights, normalisation across language buckets.

Validation

Stub. Final version: inter-rater reliability, cross-LLM consistency checks, replication protocol.

Limitations

Stub. Final version: known sample biases, LLM-version drift, language coverage gaps.

Changelog

v1.0 (2026-05-12) — Initial publication.

How to cite

@techreport{ceavers_methodology_2026,
  title       = {Research Methodology v1.0},
  author      = {CEAVERS Editorial},
  institution = {Centre for European AI Visibility Evaluation and Research Standards},
  year        = {2026},
  url         = {https://ceavers.org/methodology/}
}