The Periodic Table of Digital Authority™ rests on a simple, increasingly well-evidenced premise: in AI-driven discovery, ranking position is no longer the thing to measure. What matters is whether AI systems understand a brand, trust it, cite it, and surface it — and how consistently. Below is the independent research that supports that premise.
This is a working evidence base, not a literature review for its own sake. Each study below is summarised in our own words, linked to its original source, and annotated with how it relates to the framework. Where a source complicates or qualifies the thesis, that is noted too — the goal is an honest map of what the research actually shows.
The central thesis: AI visibility is driven by entity recognition, trusted citations, corroboration across sources, and appearance frequency — not traditional ranking position alone.
600 volunteers ran 12 identical prompts through ChatGPT, Claude, and Google's AI nearly 3,000 times. The same prompt returned the same list of brands less than 1% of the time, and the same list in the same order closer to 1-in-1,000. Fishkin's conclusion: any tool claiming a fixed "ranking position in AI" is unreliable. Crucially, the core was not random — top brands in each category appeared in 55–77% of responses regardless of phrasing, while rank within that set was effectively random.
Read the SparkToro research →Across 75,000 brands, branded web mentions showed the strongest correlation with appearing in AI Overviews (Spearman 0.664) — more than three times the correlation of backlinks (0.218). The three most correlated factors were all off-site brand signals: web mentions, branded anchors, and branded search volume. Brands in the top quartile for web mentions earned up to ten times more AI Overview mentions than the next quartile. The usual caveat applies and is worth stating plainly: correlation is not causation.
Read the Ahrefs study →A follow-up across multiple engines found the signals are not uniform. YouTube mentions showed the strongest correlation with AI visibility overall (~0.737); branded web mentions remained strong (0.66–0.71); but ChatGPT correlated only weakly with classic authority metrics like branded search volume and Domain Rating, while Google's AI Mode leaned most heavily on branded authority signals. In short: each engine sources and weights differently.
Read the multi-engine study →Google's position is that optimising for generative AI search is, in its framing, still SEO: AI Overviews are grounded in the normal search index via retrieval-augmented generation, and content must be crawlable, indexable, and demonstrate experience, expertise, authoritativeness and trustworthiness (E-E-A-T). Notably, Google also states that some tactics promoted elsewhere — llms.txt files, content "chunking," and structured data on its own — do not guarantee placement. We include this in full, including the parts that qualify other industry advice, because an honest evidence base reports the primary source as it stands.
Read Google's guidance →A large-scale longitudinal study issued 55,393 trending queries across 19 categories over 40 days. Among the findings: AI Overview activation averaged 13.7% but rose to 64.7% for question-form queries; AIO-cited domains were more credible than the co-displayed first-page results — yet nearly 30% of cited domains did not appear in those first-page results at all, indicating a source-selection mechanism distinct from Google's ranking algorithm. Decomposing answers into 98,020 atomic claims, 11% were unsupported by the cited pages.
Read the paper on arXiv →Read together, these sources converge on a consistent picture. Ranking position is an unreliable proxy for AI visibility (Fishkin). What predicts citation is corroboration — being mentioned, across the web, by sources the system trusts (Ahrefs). The signals differ by engine (Ahrefs, multi-engine). The platforms themselves confirm crawlability and demonstrable authority are prerequisites (Google). And citation selection is empirically a separate system from ranking, with a third of cited sources absent from the visible results (Xu et al.).
That is the case for measuring digital authority directly, across four dimensions — Understand, Trust, Cite, Surface — rather than inferring it from rank. It is what the Periodic Table of Digital Authority™ is built to do.
Sources are summarised in our own words and linked to their originals; all findings belong to their respective authors and publishers. Correlation studies indicate association, not causation, as their authors note. This evidence base is maintained and updated as new research is published.
Compiled by Douglas Lord, creator of the Periodic Table of Digital Authority™ and founder of Digital Dominator and AUTHORITY44.