Corrections archive

Peer-review pass. Tightened the epistemic boundary to the entry's own 'name the behavior, not the cause' rule: the definition now leads with the version mismatch; cache/retrieval layers are consistent-with, not asserted. Corrected the over-claim that true prevalence is likely higher (it is unmeasured; the observed count is only a lower bound on detectable events), and replaced 'expect weeks' with the actual 20-day observation and no published refresh SLA. Added two verified anchors: RESONEO's ChatGPT retrieval-stack study (reading cache serving copies 90+ days old) and OpenAI's offline-web-search docs.

Perplexity now cites this entry, surfacing it via the AI citation metrics hub page for the definition query; it joins the engines already citing it. The answer reproduced the coverage framing (your citations over all citations in the tracked prompt set) alongside a wall of competing definitions, so the term is broadly cited but the definitional space is crowded.

Microsoft Copilot now cites this entry (second-listed reference), reproducing the framing that passages must stay understandable when extracted independently and be self-contained, clear, sourced, and single-claim. Adds Copilot to the engines citing the term.

Clarified that a passing score proves single-fact retrieval, not merely that the model accepts long input, and not robust long-context understanding. Bounded RULER's 'half at 32K' finding to the 17 models it evaluated and flagged it as a 2024 snapshot whose exact fraction moves as models improve. Added the harder-benchmark progression with primary sources: NoLiMa (ICML 2025), which removes the lexical-overlap shortcut, and Sequential-NIAH (EMNLP 2025), which requires ordered needles. Framed the entry around vanilla NIAH within a larger test family, and added a context-rot cross-reference for the length axis.

Perplexity and Gemini now cite this entry for the definition query, reproducing the three-way distinction that citation, mention, and link are separate signals (grounding, naming, and a clickable URL). Two of five tested engines now cite it directly. ChatGPT and Claude cited third-party sources instead this time, so the term draws different sources on different engines.

Claude now cites this entry, reproducing the five-axis diagnostic framework (edge or location, cache state, path pattern, user agent plus referrer, and non-scraper user agent) and the confirmed-versus-candidate classification. Four of five tested engines now cite it directly; ChatGPT, Perplexity, and Gemini already did, each reproducing the same five axes. Microsoft Copilot did not cite it this time.

Perplexity now cites this entry, joining ChatGPT; two of five tested engines cite it directly. The answer reproduced the grounding-metadata framing (grounding chunks, grounding supports, and web search queries) and the distinction from Google AI Overviews and AI Mode. Claude, Copilot, and Gemini itself did not cite it this time.

ChatGPT now cites this entry directly for the definition query, listing it among its sources and reproducing its caveat that AI citation metrics are practitioner-created rather than standardized by peer-reviewed research; the answer closed by recommending per-engine, per-query tracking over a single blended visibility score. Two of five tested engines now cite it directly (ChatGPT and Perplexity).

ChatGPT and Perplexity both now cite this entry directly; with Claude, Copilot, and Gemini already citing it, all five tested engines have now cited this entry at least once. ChatGPT listed two of this glossary's pages as its top sources and reproduced the lock-down-the-denominator guidance; Perplexity cited the 40-of-100 worked example inline.

Microsoft Copilot now cites this entry (second in its references, after Brave's own documentation), reproducing the surfaces framing (AI Answers, Ask Brave, Featured Snippets, AI-powered descriptions) and the independent-index point that Brave citations reflect its own crawl and ranking decisions. Three of five tested engines now cite it directly.

Claude now cites this entry alongside the primary artifacts (the arXiv paper, GitHub, Hugging Face, OpenReview), including the comparison that the retrieval-ranking baseline outperformed the best C-SEO method by 7.6 times. Three of five tested engines now cite it directly, and ChatGPT cites it partially.

Microsoft Copilot now cites this entry first in its references, reproducing the six decisions a protocol locks down (query design, cadence, engine coverage, recording schema, disambiguation rules, signal thresholds) and the three failure modes of unstructured testing. With Copilot added, all five tested engines have now cited this entry at least once.

ChatGPT now cites this entry directly, where it previously appeared only in the folded sources list: the answer led with this page, and the rate-of-change versus snapshot framing plus the momentum-indicator caveat (velocity measures observed output, not the mechanism causing it) were cited inline. Four of five tested engines now cite it directly.

Microsoft Copilot now cites this entry first in its references, reproducing the three citation surfaces (WebSearch inline citations, DeepSearch reasoning traces, and structured citation fields in the xAI API) along with the native X integration and variable citation slot distinctives. Three of five tested engines now cite it directly.

Claude now cites this entry, opening its answer with a direct no and reproducing the Aggarwal et al. 2023 negative result (the only tested method to fall below baseline, at roughly 8 percent worse). Two of five tested engines now cite it directly.

Claude now cites this entry, reproducing the two-tier citation model: licensed publishers receive linked citations under the December 2025 content deals while other sources may appear unlinked, and social posts function as first-class sources alongside web pages. Two of five tested engines now cite it directly.

Microsoft Copilot now cites this entry first in its references, attributing the headline evidence row to it (position-adjusted word count 27.2 versus a 19.3 baseline, about a 41 percent relative gain, the largest measured lift among the nine methods tested). Four of five tested engines now cite it directly.

ChatGPT now cites this entry directly, where it previously appeared only in the folded sources list; it joins Perplexity and Microsoft Copilot, so three of five tested engines cite it directly. Both ChatGPT and Perplexity reproduced the retrieval, grounding, and attribution evaluation framing. Claude and Gemini did not cite it this time.

First confirmed AI-search citation for this entry: Microsoft Copilot lists it in its references for the definition query, alongside the 40 to 70 word extractable-answer framing. One of five tested engines now cites it; ChatGPT, Perplexity, Claude, and Gemini did not cite it this time.

Claude now cites this entry directly, reproducing the precision and recall asymmetry and the Liu, Zhang, and Liang (EMNLP 2023) benchmark (74.5 percent precision, 51.5 percent recall). With Claude added, all five tested engines have now cited this entry at least once across our probe rounds. This round Copilot placed it first in its references and ChatGPT cited it among its sources.

Perplexity now cites this entry directly (first in its sources), reproducing our framing that grounding reduces but does not eliminate hallucinations, since a grounded model can still misread a source or cite one that does not fully support the claim. It joins ChatGPT, so two of five tested engines cite it directly. Claude, Copilot, and Gemini did not cite it this time.

Microsoft Copilot now cites this entry, placing our page at the top of its sources and anchoring to the Aggarwal et al. 2023 result (PAWC 24.7 versus a 19.3 baseline, about 28 percent relative gain on citation visibility). Three of five tested engines now cite it directly, joining ChatGPT and Claude; Perplexity and Gemini did not cite it this time.

First confirmed AI-search citation for this entry: Perplexity cited it for the definition query, placing our page first and naming it as the key source, and reproduced the developer-facing provenance framing (inline reference links, search-grounded answers, attribution panels in the IDE). One of five tested engines now cites it directly; ChatGPT, Claude, Copilot, and Gemini did not.

ChatGPT now cites this entry: our reranking page appears in ChatGPT's source list for the definition query. One of five tested engines now cites it directly. As an established retrieval term, the citation sits alongside the canonical owners (Cloudflare, Inkeep) that lead the inline references rather than leading itself.

First confirmed AI-search citation for this entry: Microsoft Copilot cited it at the top of its sources for the definition query, reproducing the enforcement-versus-voluntary distinction (blocking as mandatory enforcement, unlike robots.txt or AIPREF signals). One of five tested engines now cites it directly; ChatGPT, Perplexity, Claude, and Gemini did not.

Perplexity now cites this entry for the definition query, joining ChatGPT; two of five tested engines cite it directly. Claude, Copilot, and Gemini did not cite it this time.

Gemini now cites this entry, joining Claude and Microsoft Copilot; three of five tested engines cite it directly. Gemini and Copilot both reproduced the per-engine attribution-rate figures (Perplexity around 29 percent; ChatGPT, Gemini, and Copilot around 7 to 8 percent; Google AI Overview 38 to 76 percent). ChatGPT and Perplexity did not cite it this time.

Microsoft Copilot now cites this entry, joining ChatGPT, Perplexity, and Gemini; four of five tested engines cite it directly. Multiple engines anchored to the Liu, Zhang, and Liang (EMNLP 2023) benchmark figures (74.5 percent precision, 51.5 percent recall). Claude cited a third-party source instead.

Broad citation gains for this entry: Perplexity, Claude, and Microsoft Copilot all now cite it directly, joining ChatGPT, and Gemini references it through the AI citation metrics overview. Four of five tested engines cite it directly and the fifth cites it partially. Copilot and ChatGPT used the term page itself, while Claude reached it through the citation-tracking page.

Perplexity and Gemini now cite this entry, joining Claude; three of five tested engines cite it directly and ChatGPT cites it partially. On Claude and Gemini the citation-velocity page itself was used, while ChatGPT and Perplexity reached it through sibling metric pages (AI citation metrics, citation share). Copilot did not cite it this time.

First confirmed AI-search citations for this entry: Perplexity and Gemini both cite it for the definition query, with Gemini using our definition as the opening line and rendering the select, order, and format pipeline. Two of five tested engines now cite it directly; ChatGPT, Claude, and Copilot did not.

Perplexity and Microsoft Copilot now cite this entry, joining Gemini; three of five tested engines cite it directly. Each placed our page at or near the top of its sources and used our practitioner-shorthand framing (the retrieval-corpus backend, split into passage chunks, embeddings, lexical data, and provenance). ChatGPT and Claude cited primary or academic sources instead.

Microsoft Copilot now cites this entry, placing it second in its sources and reproducing our negative-result framing from the Aggarwal et al. study (keyword stuffing performed 8 to 10 percent worse than baseline, the only method to fall below it). One of five tested engines cites it directly; ChatGPT, Perplexity, Claude, and Gemini did not.

Perplexity now cites this entry, joining Claude; two of five tested engines cite it directly. Both placed our page first and reproduced our practitioner-shorthand framing and the Statistics Addition method from the Aggarwal et al. study, including our caveat against chasing a raw density ratio. ChatGPT, Copilot, and Gemini did not cite it this time.

First AI engine citation: Perplexity surfaced this entry as a primary source for 'What is AI search evaluation?', citing the pillar above the fold alongside the opening definition. First engine to cite the entry since publication; 1 of 5 tested engines now cites it. Perplexity reached this explainer directly rather than the underlying academic sources.

First AI engine citations: ChatGPT, Perplexity, and Claude all surfaced this entry for 'What is AIPREF (AI usage preferences)?', each drawing on our preferences-versus-authentication distinction (AIPREF declares usage preferences; it does not authenticate the requester, which is the separate Web Bot Auth effort). From 0 to 3 of 5 tested engines citing the entry.

First AI engine citation: Perplexity cited this entry for 'What is authority signals?', drawing on our citations-and-backlinks breakdown alongside established SEO glossaries. 1 of 5 tested engines now cites it.

Second AI engine citation: Perplexity surfaced this entry as its top source for 'What is Black-hat C-SEO?', listing it first in the answer's source list and drawing on our prompt-injection-and-deception framing. Joins Gemini; 2 of 5 tested engines now cite it.

Copilot citation confirmed for 'What is Citation Footprint?', reproducing our coined framing (a cumulative, monotonic count of distinct cited URLs measuring coverage rather than intensity, distinct from citation share and citation velocity). Fourth engine to cite this coined entry; 4 of 5 tested engines now cite it.

First AI engine citations: ChatGPT and Perplexity both surfaced this entry for 'What is Citation hallucination?', with ChatGPT ranking it the top source and Perplexity drawing on our core distinction (a hallucinated citation points to a source that does not exist, separate from misquoting a real source or answering with none). 2 of 5 tested engines now cite it.

Gemini citation confirmed for 'What is cite-ability in AI search?', with our entry dominating the source panel and Gemini reproducing the four-trait framing (context-free quote, answer-first layout, factual density, echo-and-expand). Fourth engine to cite this practitioner-coined entry; 4 of 5 tested engines now cite it.

First AI engine citations: Perplexity and Gemini both surfaced this entry as their top source for 'What is Deep research mode?', each reproducing our framing that deep research mode is an escalation of agentic retrieval applying query fan-out at much larger scale. 2 of 5 tested engines now cite it.

ChatGPT citation confirmed for 'What is Freshness signals?', surfacing this entry as the top source and drawing on our metadata-signals breakdown (datePublished, dateModified, Last-Modified headers, version history). Joins Perplexity; 2 of 5 tested engines now cite it.

Perplexity citation confirmed for 'What is GEO content methods?', surfacing this pillar as its top source and reproducing our negative-result framing (most headline content tweaks were weak or null levers in Aggarwal 2023, and retrievability matters more than stylistic rewrites). Joins Gemini; 2 of 5 tested engines now cite it.

Perplexity citation confirmed for 'What is Position-Adjusted Word Count?', surfacing this entry as its top source, ranked above the original GEO paper it explains. Joins Claude; 2 of 5 tested engines now cite it.

First AI engine citation: Perplexity cited this entry as its second source for 'What is Retrievability?' (after Wikipedia), drawing on our Azzopardi and Vinay 2008 grounding and the framing that retrievability is an upstream lever on-page content tweaks cannot fix. 1 of 5 tested engines now cites it.

Same-day peer-review pass: corrected the Chatbot Arena attribution (Arena ranks models on crowdsourced human votes; the LLM judge scores MT-Bench, and the study used Arena's human data to validate the judge), scoped the over-80% human-agreement figure to the original study's settings, and added that the GEO benchmark's subjective-impression metric is itself G-Eval/GPT-scored, so verbosity bias may explain part of some content tactics' measured lift. Softened 'introduced' to 'named and validated' and 'cancel position bias' to 'detect and reduce'.

Perplexity citation confirmed for the definition query. Perplexity surfaced the AI citation metrics overview inline and in its sources-used list, reproducing the per-metric breakdown. First tracked engine to cite this pillar.

Perplexity citation confirmed for the definition query; the entry surfaced inline and in Perplexity's sources. First tracked engine to cite this entry.

Perplexity citation confirmed for the definition query, with the entry surfaced as a top source in Perplexity's answer. Joins ChatGPT among the engines citing this entry.

Perplexity citation confirmed for the definition query, with the entry surfaced as the top source. Joins ChatGPT among the engines that have cited this entry.

Now cited by ChatGPT, Perplexity, and Gemini for the definition query, the first citations for this coined metric. Perplexity and Gemini surfaced the citation footprint page directly; ChatGPT reached it through the terms index. Each answer carried the coinage framing and the cumulative, breadth-over-intensity distinction.

ChatGPT and Claude citations confirmed for the definition query, joining Perplexity and Gemini, so four of the five tracked engines now cite this entry. Each reproduced the probe-versus-protocol distinction and the fixed-panel methodology.

Gemini citation confirmed for the definition query. The metric surfaced in Gemini's answer through the related AI citation metrics and citation footprint pages, using the share-of-voice framing.

Microsoft Copilot citation confirmed for the definition query, with the entry ranked the top source in Copilot's panel. Third tracked engine to cite this entry, joining ChatGPT and Perplexity.

Perplexity citation confirmed for the definition query; the entry surfaced in the Sources list of Perplexity's answer. Joins ChatGPT among the engines citing this entry.

Perplexity citation confirmed for the definition query, the first citation for this entry on any tracked engine. The entry ranked the top source in Perplexity's panel, carrying its two-tier (licensed-publisher versus general-web) framing into the answer.

Perplexity citation confirmed for the definition query, with the entry surfaced as a source in Perplexity's panel. Joins ChatGPT among the engines citing this entry.

Perplexity citation confirmed for the definition query, the first citation for this entry on any tracked engine, with the page ranked the top source.

ChatGPT and Microsoft Copilot citations confirmed for the definition query, each surfacing this entry as the top source. They join Perplexity, so three tracked engines now cite it. The page carried its passage-level retrieval framing into both answers.

Peer-review pass: softened the framing so the cutoff reads as a primary driver of retrieval and the citation surface, not the sole cause (retrieval also serves verification, long-tail knowledge, and user-requested sources). Reframed retrieval triggering as multi-factor (query type, uncertainty, product policy, user request), not a before/after-cutoff line. Surfaced per-engine differences into the body (Perplexity nearly always retrieves, ChatGPT conditionally, Claude answers stable knowledge without searching). Added an OpenAI model-docs anchor.

Now cited by all five tracked engines, the first GEO Glossary entry to reach every engine we track. Perplexity and Microsoft Copilot are the two newest to cite this metric, joining ChatGPT, Claude, and Gemini. In each case it surfaced through the AI citation metrics overview that defines it, carrying its linked-versus-unlinked distinction into the answer.

Now cited by Claude for the first time, bringing the count to two of five tracked engines. Claude assembled its answer from a cluster of related GEO Glossary method entries (Quotation Addition, Fluency Optimization, Authoritative Statement Strength, Statistical Density, and Definition-Lead Style) rather than from a single page, citing Quotation Addition first. It shows how a method built on several techniques can surface through the entries that define each one.

Now cited by Claude, the second of five tracked engines to cite the entry after ChatGPT. Claude drew the definition and the practical rules directly from this page, reproducing the answer-block-opening framing and the inverted-pyramid analogy it uses to explain leading with the definition.

Now cited by ChatGPT, joining Claude as the second of five tracked engines to cite the entry. ChatGPT used this page as its lead source and reproduced its Position-Adjusted Word Count framing for the method.

Now cited by ChatGPT, the first of five tracked engines to cite the entry. ChatGPT surfaced this page among its sources when answering what hallucination grounding is, alongside other AI reference glossaries.

Revalued the supporting Aggarwal PAWC figure to the paper's position-adjusted 'Overall' column (quotation addition 27.2 versus a 19.3 baseline); the earlier figure (27.8 vs 19.5) was the paper's plain Word Count sub-column.

Corrected the Aggarwal Table 1 figures: the values previously cited as PAWC (Authoritative 21.8 vs baseline 19.5, and the rest) were the paper's plain Word Count sub-column. Updated to the paper's actual position-adjusted Word Count (the 'Overall' column: Authoritative 21.3 vs baseline 19.3, a raw +10%), which is the metric the paper's headline gains are computed on. The load-bearing finding is unchanged: the paper characterizes Authoritative tone verbatim as 'no significant improvement', a null result regardless of the raw percentage.

Corrected the Aggarwal Table 1 figures: the values previously cited as PAWC (Cite Sources 24.9 vs baseline 19.5, and the rest) were the paper's plain Word Count sub-column. Updated to the paper's actual position-adjusted Word Count (the 'Overall' column: Cite Sources 24.6 vs baseline 19.3, about +27%; Quotation Addition 27.2 about +41%; Keyword Stuffing 17.7 about -8%), which is the metric the paper's headline gains are computed on. The named top-3 framing and the 4th-place standalone ranking are unchanged.

Revalued the supporting Aggarwal PAWC figures in the footnote to the paper's position-adjusted 'Overall' column (Cite Sources 24.6, Quotation Addition 27.2, baseline 19.3); the earlier figures (24.9 / 27.8 / 19.5) were the paper's plain Word Count sub-column, not its position-adjusted metric.

Revalued the supporting Aggarwal PAWC range to the paper's position-adjusted 'Overall' column (~27% to ~41% lift); the earlier range (~28% to ~43%) was derived from the paper's plain Word Count sub-column.

Corrected the Aggarwal Table 1 figures: the values previously cited as PAWC (Fluency Optimization 25.1 vs baseline 19.5, and the rest) were the paper's plain Word Count sub-column. Updated to the paper's actual position-adjusted Word Count (the 'Overall' column: Fluency Optimization 24.7 vs baseline 19.3, about +28%; Quotation Addition 27.2 about +41%; Keyword Stuffing 17.7 about -8%), which is the metric the paper's headline gains are computed on. Rankings unchanged (Fluency still 3rd by standalone score, still not in the paper's named top-3).

Corrected the Aggarwal Table 1 figures throughout the methods table and footnote: the values previously cited as PAWC (Quotation Addition 27.8 vs baseline 19.5, and the rest) were the paper's plain Word Count sub-column. Updated to the paper's actual position-adjusted Word Count (the 'Overall' column: Quotation Addition 27.2 vs baseline 19.3 about +41%, Keyword Stuffing 17.7 about -8%), which is the metric the paper's headline gains are computed on. The verdicts, the named top-3, and the null/negative findings are unchanged.

Corrected the Aggarwal Table 1 figures: the values previously cited as PAWC (Keyword Stuffing 17.8 vs baseline 19.5, and the rest) were the paper's plain Word Count sub-column. Updated to the paper's actual position-adjusted Word Count (the 'Overall' column: Keyword Stuffing 17.7 vs baseline 19.3, mathematically about -8%), which is the metric the paper's headline gains are computed on. The negative-result finding is unchanged: Keyword Stuffing is still the only method scoring below baseline, and the paper's verbatim 'little to no performance improvement' framing stands.

Revalued the Aggarwal per-method figures in the footnote to the paper's actual position-adjusted PAWC values (the 'Overall' column: Quotation Addition 27.2 vs baseline 19.3, about +41%). The earlier figures (27.8 vs 19.5) were the paper's plain Word Count sub-column, not its position-adjusted metric.

Revalued the Aggarwal per-method figures in the footnote to the paper's actual position-adjusted PAWC values (the 'Overall' column: Quotation Addition 27.2 vs baseline 19.3, about +41%). The earlier figures (27.8 vs 19.5) were the paper's plain Word Count sub-column, not its position-adjusted metric.

Corrected the Aggarwal Table 1 figures: the values previously given as PAWC (baseline 19.5, Quotation Addition 27.8, and so on) were the paper's plain Word Count sub-column. Updated to the paper's actual position-adjusted Word Count (the 'Overall' column: baseline 19.3, Quotation Addition 27.2 at about +41%, Keyword Stuffing 17.7 at about -8%), which is the metric the paper's headline gains are computed on. Also fixed the metric symbol to Imp_pwc and clarified that PAWC is computed over the sentences that cite a source, so it measures attributed word share within an answer rather than citation frequency or rank.

Corrected the Aggarwal Table 1 figures: the values previously cited as PAWC (Quotation Addition 27.8 vs baseline 19.5, and the rest) were the paper's plain Word Count sub-column. Updated to the paper's actual position-adjusted Word Count (the 'Overall' column: Quotation Addition 27.2 vs baseline 19.3, about +41%; Keyword Stuffing 17.7, about -8%), which is the metric the paper's headline gains are computed on. The named top-3 framing and rankings are unchanged (Fluency Optimization is still 3rd by standalone score).

Corrected the Aggarwal Table 1 figures: the values previously cited as PAWC (Statistics Addition 25.9 vs baseline 19.5, and the rest) were the paper's plain Word Count sub-column. Updated to the paper's actual position-adjusted Word Count (the 'Overall' column: Statistics Addition 25.2 vs baseline 19.3, about +31%; Quotation Addition 27.2 about +41%; Keyword Stuffing 17.7 about -8%), which is the metric the paper's headline gains are computed on. Rankings and the named top-3 framing are unchanged (Statistics Addition still 2nd by standalone score).

Revalued the supporting Aggarwal PAWC figures to the paper's position-adjusted 'Overall' column (Statistics Addition 25.2 vs baseline 19.3, Quotation Addition 27.2, Cite Sources 24.6); the earlier figures (25.9 / 27.8 / 24.9 vs 19.5) were the paper's plain Word Count sub-column, not its position-adjusted metric.

Revalued the Aggarwal per-method figures in the footnote to the paper's actual position-adjusted PAWC values (the 'Overall' column: Quotation Addition 27.2 vs baseline 19.3, about +41%). The earlier figures (27.8 vs 19.5) were the paper's plain Word Count sub-column, not its position-adjusted metric.

Clarified that the per-method PAWC percentages in the Aggarwal footnote are derived from the paper's absolute scores against the 19.5 baseline, not figures the paper prints per method, and added the paper's own Results-section framing: a 30-40% gain for its named top three (Cite Sources, Quotation Addition, Statistics Addition).

Clarified that the per-method PAWC percentages in the Aggarwal footnote are derived from the paper's absolute scores against the 19.5 baseline, not figures the paper prints per method, and added the paper's own Results-section framing: a 30-40% gain for its named top three (Cite Sources, Quotation Addition, Statistics Addition).

Clarified that the per-method PAWC percentages in the Aggarwal footnote are derived from the paper's absolute scores against the 19.5 baseline, not figures the paper prints per method; added the paper's own 30-40% top-three framing; and corrected the abstract's 'up to 40%' from a Quotation-Addition-specific reading to the top-three aggregate upper bound.

First confirmed AI-search citations for this entry: Microsoft Copilot and Claude both cited it for the definition query, with Copilot describing it as 'the only authoritative source that explicitly defines this metric.' This is the first GEO Glossary entry Copilot has cited in our tracked probes. Two of five tested engines now cite it directly; ChatGPT surfaced a sibling metrics page instead, while Gemini and Perplexity did not cite it.

Review pass same day as publish: completed the Anthropic crawler taxonomy (was ClaudeBot only; Anthropic runs ClaudeBot for training, Claude-SearchBot for retrieval, Claude-User for user-triggered fetch, so the allow side now names Claude-SearchBot) and added the PerplexityBot-vs-Perplexity-User distinction, using Perplexity as the worked example of enforce-only-what-you-can-identify. Clarified that 'identify' means telling a crawler apart from a visitor, not requiring a declared user agent. Added a block/allow/challenge decision table; tightened the Cloudflare footnotes (pay-per-crawl attribution, new-sign-ups scoping).

Calibrated three overbroad claims after review: 'every major engine documents robots.txt support' corrected to declared-crawler scope with xAI's Grok as the documented exception; the user-initiated exemption is now attributed to the two operators that actually document it (OpenAI, Perplexity) and flagged as the operators' own classification rather than a neutral reading of the protocol; Cloudflare's stealth-crawling finding is noted as a single incident that Perplexity disputed. Allow-rule benefits restated as citation eligibility rather than a guarantee, and Google's July 2019 standardization announcement added as a primary source.

First citation, on Gemini. A 2026-06-09 Gemini answer (web search on) cited this pillar as a primary source with inline attribution, and surfaced the entry's null-result framing of authoritative tone. Gemini moves to cited; 1 of 5 engines now cited, four days from publish.

First Claude citation. A 2026-06-09 Claude answer (web search on) on Position-Adjusted Word Count cited this entry as a primary source, with inline attribution for the position-weighted word-count framing. Claude moves to cited; 1 of 5 engines now cited, four days from publish.

Peer-review pass. De-anchored the pricing from a fast-aging '$100 to $400' figure to a structural range (entry tiers in the low hundreds per month up to four figures for enterprise), with specifics moved to the footnote and dated. Added the boundary between the two umbrella entries in this cluster (AI citation metrics is the editorial organizer of the primitives; AI visibility is the market's composite name), and a note that probe-based visibility scores inherit fixed-prompt probing's blind spots. Softened 'moves the primitives' to 'can improve over time,' and made the lightweight proxy per-engine and per-prompt-set.

Removed an unsourced figure from the shared Aggarwal footnote: a previously-asserted 'Average 31.4% in combinations' for Cite Sources was found unlocatable in the GEO paper on re-check (ar5iv full text + web search). The verbatim facts are unchanged: the paper names Cite Sources in its top-3 despite its 4th-place standalone score without stating why, and the only combination result it reports is Fluency plus Statistics outperforming any single method by more than 5.5%.

First confirmed ChatGPT citation: ChatGPT surfaced this entry as a GEO Glossary source for the 'what is Brave Search AI citation' definition query. The chatgpt status moves from untested to cited.

First confirmed ChatGPT citation: ChatGPT surfaced this entry among its sources (in the folded 'more sources' tray) for the 'what is ChatGPT search citation' definition query. The chatgpt status moves from untested to cited.

First confirmed Perplexity and Gemini citations: both surfaced this entry as a top source for the 'what is citation probe protocol' query, with Perplexity ranking it first and Gemini's source panel listing several GEO Glossary entries for it. The perplexity status moves from not-cited to cited and gemini from untested to cited.

Removed an unsourced figure surfaced while building the GEO content methods pillar: a previously-asserted 'Average 31.4% in combinations' for Cite Sources (offered as the reason it appears in the paper's named top-3 over Fluency) was found unlocatable in the GEO paper on re-check via the ar5iv full text and a web search. The verbatim facts stand: the paper names Cite Sources in its top-3 despite its 4th-place standalone score, without stating why; the only combination result it reports is Fluency plus Statistics outperforming any single method by more than 5.5%.

Completed the engine roster after peer review: added Microsoft's Researcher in Microsoft 365 Copilot (March 2025, distinct from the Think Deeper toggle) and Anthropic's Claude Research (April 2025), so all five tracked engines are covered, and re-scoped the December 2024 to February 2025 window as the first wave with Microsoft and Anthropic following that spring. Softened over-general claims (broadly similar rather than near-identical; products differ; Gemini Deep Research is not stated to use query fan-out, only Google's Deep Search is). Added a note that deep research is the most expensive, least-probed citation surface.

Added the first causal-isolation evidence on how much AEO actually moves traffic: a 2026 single-domain natural experiment (Glasp.co) found ChatGPT referrals grew 5.7x overall but only about 1.82x after removing platform-level growth (untreated pages on the same domain grew 3.5x on tailwind alone), so headline AEO multiples overstate the causal effect. Framed as a single-site result, not a generalizable estimate.

Added the grounded-AI-search side: a Tow Center study of eight grounded tools found wrong citations in over 60% of responses, with over half of Gemini and Grok 3 responses citing fabricated or broken URLs. Introduced the hybrid failure mode it surfaces (the source is real but the link is fabricated, broken, or points to a syndicated copy), distinct from a wholly non-existent source. Softened claims that grounded retrieval reduces the problem (it does not remove broken-link or misattribution failures), broadened the definition to cover fake cases and quotations, and corrected the Mata sanction to the two lawyers and their firm.

Corrected an attribution error: the 'cited from beyond the top 10' pattern is measured for the standalone assistants (ChatGPT, Gemini, Copilot, Perplexity), not for Google's AI Overview, whose citations come disproportionately from the top 10 (about 38% in Ahrefs' 2026 update, down from 76% in 2025). Turned that contrast into a finding (AI Overview hugs organic ranking while independent assistants nearly decouple), softened the AI Overview claim to a 'correlate', noted that Ahrefs' own opening sentence mis-attributes its 12% headline, and added the 15,000-prompt dataset size.

Claude citation confirmed for the definition query ('What is cite-ability in AI search?'). The entry surfaced as a cited source in Claude's web-search answer, alongside several third-party glossaries. Third engine to cite this entry (after ChatGPT and Perplexity); 3 of 5 tested engines now cite it. Notable because cite-ability is a practitioner-coined term with no upstream primary paper, so Claude cites this explanation directly rather than reaching past it to a canonical source.

Added a side-by-side disambiguation of LLMO against GEO, AEO, and AIO (the AI-search acronyms readers most confuse), plus two FAQ entries on whether LLMO and GEO differ in practice; the umbrella framing is hedged as contested to match the GEO and AIO entries. Also fixed two sourcing errors: the data-selection citation pointed to arXiv:2402.16827 (a pretraining-scoped survey) while the text described instruction tuning, now repointed to arXiv:2402.05123; and the LangChain default chunk size was off ~4x (RecursiveCharacterTextSplitter defaults to 4000 characters, not ~250 tokens).

Peer-review pass. Fixed the attachment draft's title ('Associating AI Usage Preferences with Content in HTTP', not 'Attaching...') and reframed the maturity reasoning: both drafts are adopted, still-active working-group documents, and the attachment draft's lapsed latest revision is the routine six-month Internet-Draft auto-expiry, not abandonment (it carries an Aug 2026 milestone). The watch-don't-deploy conclusion stands; the reasoning is now pre-standardization, not 'expired.' Also softened framing (effort-to-standardize vs finished standard; complementary to rather than successor to robots.txt).

Peer-review pass. Corrected the Status framing that treated contextual chunking (Anthropic) as the single 2024-2026 response to context loss; it is now presented as one of several approaches, alongside late chunking (Jina AI: embed the whole document, then split) and semantic chunking, with the late-chunking paper added as a non-vendor-single primary source. Tightened over-absolute claims (chunk as the retrieved unit, every production RAG system) to reflect that not all retrieval works by text chunking. Added a boundary note distinguishing chunking (splitting that produces units) from sub-document retrieval (selecting among the units).

Verified the venue against the ACL Anthology record (Transactions of the ACL, Vol. 12, 2024, pp. 157-173) and disambiguated the source from the other Liu et al. 2023 paper in the glossary, 'Evaluating Verifiability in Generative Search Engines' (arXiv:2304.09848), which shares the same first and senior author but is a different work; both are now cited by short title rather than a bare author-year. Also tightened the framing of front-loading as one evidence-backed mechanism among several, and clarified that BM25 and dense-similarity scoring do not weight within-body position even where production systems layer on field or heading boosts.

Perplexity and Gemini citations confirmed for the definition query ('What are citation precision and citation recall in AI search engines?'), both surfacing 'Citation precision and recall - GEO Glossary' as a primary source. Combined with the same-day ChatGPT citation, this entry is now cited by three engines (ChatGPT + Perplexity + Gemini).

ChatGPT citation confirmed for the definition query ('What are citation precision and citation recall in AI search engines?'). The entry surfaced as a 'GEO Glossary' source in ChatGPT search, alongside the Liu et al. verifiability paper and other references. First confirmed citation for this entry on any engine.

ChatGPT citation confirmed for the definition query ('What is an inverted index in information retrieval?'). The entry surfaced as a 'GEO Glossary' source in ChatGPT search, alongside Wikipedia and GeeksforGeeks. First confirmed citation for this entry on any engine.

Perplexity citation confirmed for the definition query ('What is passage-level optimization for AI search?'). 'Passage-level optimization - GEO Glossary' surfaced as a primary source. First confirmed citation for this entry on any engine.

Perplexity citation confirmed for the definition query ('What is sub-document retrieval in RAG?'). 'Sub-document retrieval - GEO Glossary' surfaced as a primary source alongside RAG-technique sources. First confirmed citation for this entry on any engine.

ChatGPT citation confirmed for the definition query ('What is authoritative statement strength in AI search content optimization?'), surfacing as a 'GEO Glossary' source. This completes four-engine coverage: ChatGPT, Perplexity, Claude, and Gemini all cite this entry, the first term on the site confirmed cited by four distinct engines.

Gemini citation confirmed for the definition query ('What is authoritative statement strength in AI search content optimization?'). Gemini surfaced the entry as a 'GEO Glossary' source and reproduced its framing (the paper-verbatim null quote plus the +11.8% raw-vs-statistically-null distinction). Third engine to cite this entry (Perplexity + Claude + Gemini), making it the glossary's most broadly-cited term.

Same-day revisions. Method names switched to the paper's own terminology (Simple Language / Citations / Quotes / Statistics rather than Aggarwal's Easy-to-Understand / Cite Sources / Quotation Addition / Statistics Addition); Aggarwal labels kept in parentheses. Added a second Discussion quote where the paper notes Aggarwal's own PAWC data, on inspection, also points to decreasing scores. Added specific evaluation LLMs (gpt-4o-mini + claude-3-5-haiku) as concrete anchor for the more-production-realistic framing. Clarified the paper benchmarks nine C-SEO methods plus a separate baseline (ten items total).

Gemini citation confirmed for the definition query ('What is citation match rate in AI search?'). The entry surfaced as a 'GEO Glossary' source in Gemini's answer, cited inline in several places. Third engine to cite this entry (after ChatGPT and Claude). Notable: citation match rate is a practitioner-coined metric with no vendor-canonical primary paper, so Gemini cites this explanation rather than an upstream source.

Claude citation confirmed for two of this entry's tracked queries: the definition query ('What is authoritative statement strength in AI search content optimization?') and the Aggarwal vendor-research query ('How did Aggarwal 2023 rank Authoritative tone among content methods?'). The entry surfaced as a primary source in Claude's desktop web-search answers. Same-day cross-engine signal: Perplexity also confirmed cited today across the same probes plus the operational variant.

Perplexity citation confirmed for all three of this entry's tracked queries: the definition query ('What is authoritative statement strength in AI search content optimization?'), the operational query ('Does using authoritative tone increase ChatGPT citation rate?'), and the vendor-research query ('How did Aggarwal 2023 rank Authoritative tone among content methods?'). The entry surfaced as a primary source with inline citations in all three Perplexity answers. First multi-probe citation event for this entry, one day after the lede was rewritten to lead with the paper's verbatim 'no significant improvement' framing.

Claude citation confirmed for the definition query ('What is citation match rate in AI search?'). The entry surfaced as a primary cited source in Claude's desktop answer. Cross-engine consistency: ChatGPT first cited this entry 2026-05-17; Claude now also cites for the same probe.

Claude citation confirmed for the definition query ('What is citation share in AI search?'). The entry surfaced as a primary cited source in Claude's desktop answer. First confirmed Claude citation for this entry.

Perplexity citation confirmed for the Aggarwal vendor-research probe ('What did Aggarwal 2023's GEO paper say about Cite Sources as a content method?'). Cite Sources Optimization | GEO Glossary surfaced as one of 11 sources, with the sibling Quotation Addition entry also cited inline. Citation comes one day after the entry's lede was rewritten to lead with the paper's verbatim named top-3 framing.

Perplexity citation reconfirmed for the definition query ('What is cite-ability in AI search?'). Cite-ability | GEO Glossary continues to surface as a primary source ten days after first observed citation on 2026-05-20, with inline citations across the answer body.

Claude citation confirmed for the Aggarwal vendor-research query ('What did Aggarwal 2023's GEO paper say about Fluency Optimization as a content method?'). The entry surfaced as a primary cited source in Claude's desktop answer. First confirmed Claude citation for this entry.

PAWC labeling sweep. Critical reframing: lede previously called Fluency Optimization 'one of the four top-performing single methods', but the paper's verbatim named top-3 (Cite Sources, Quotation, Statistics) does NOT include Fluency despite its 3rd-place standalone Table 1 PAWC. The paper's strongest framing of Fluency is the §5.3 Fluency-plus-Statistics combination pair (+5.5% over any single method), not as a named top single-method intervention. Aggarwal footnote labels values as 'Table 1 main GEO-bench' + Table 5 Perplexity.ai per-engine caveat.

Corrected the Mintlify blog post URL and announcement date. The Sources block referenced 'mintlify.com/blog/llmstxt' (404) with the date November 14, 2024; the actual post is 'Simplifying docs for AI with /llms.txt' at 'www.mintlify.com/blog/simplifying-docs-with-llms-txt', published November 20, 2024. Body sentence about the Mintlify platform-wide rollout date updated to match. Caught by the new site-wide link-audit pass across all 226 frontmatter source URLs.

Claude citation confirmed for the Aggarwal vendor-research query ('What did Aggarwal 2023's GEO paper say about Quotation Addition as a content method?'). The entry surfaced as a primary cited source in Claude's desktop answer. First confirmed Claude citation for this entry.

Claude citation confirmed for the Aggarwal vendor-research query ('What did Aggarwal 2023's GEO paper say about Statistics Addition as a content method?'). The entry surfaced as a primary cited source in Claude's desktop answer; it also surfaced as the cited source for the parallel Cite Sources Aggarwal query, a cross-entry signal.

Mechanism + adoption fact-check (same-day after publish). §How it works now lists three required headers (Signature-Agent, Signature-Input, Signature); Signature-Agent (not Signature-Input) carries the bot's key directory URL that the verifier reads to fetch the public key. Status in 2026 updated to the agent-first / crawler-later split: Google's experimental Google-Agent signs at agent.bot.goog and OpenAI's ChatGPT agent signs at chatgpt.com, while Googlebot, GPTBot, and OAI-SearchBot remain unsigned. Added IETF draft-meunier-web-bot-auth-architecture (Cloudflare + Google co-authored) and AWS Bedrock AgentCore implementer footnotes.

Critical paper-misrepresentation fix. Entry previously framed the Authoritative result as a '+11.8% measurable modest lift'. But Aggarwal Section 4 verbatim: 'one would expect a more persuasive and authoritative tone can boost visibility. However, to the contrary we find no significant improvement.' The +11.8% PAWC is a raw number; the paper itself frames it as null and reports no p-values. Description, metaDescription, lede, Status, C-SEO Bench section, and FAQs rewritten to lead with the paper-verbatim null. Raw PAWC retained for transparency. Entry was misrepresenting peer-reviewed research; correction restores cluster discipline.

Same-day peer-review revision. Critical lede fix: asymmetry was reversed (precision 74.5% > recall 51.5% means engines were better at faithful citation than at citing every claim-bearing sentence). Critical fact fix: removed unverified Profound per-response density numbers; replaced with the study's actually-reported source-concentration patterns. Added Li & Sinnamon 2024 + Profound footnotes to Sources. Softened the precision-failure framing, precision-recall trade-off, and hallucination-grounding definition. NeevaAI dates made precise. 2x2 labels improved. Back-links added to attribution-rate, cite-ability, pillar.

Same-day revision. Critical fact correction: IndexNow launched October 18, 2021 as a Bing and Yandex collaboration only (prior text incorrectly listed Seznam and Naver as founding partners); Naver joined July 2023. Now consistent with the IndexNow Protocol entry. Hedged the ChatGPT search retrieval-pipeline framing to match the chatgpt-search-citation entry. Softened the DuckDuckGo Bing-grounded characterization. Clarified that Featured Snippets is extractive (predates generative AI) and that AI-powered descriptions cite the single linked result rather than synthesizing multiple sources.

Follow-up cluster-consistency fix. Writing the new DuckDuckGo AI citation entry surfaced that DuckDuckGo's Search Assist actually runs DuckDuckGo's own DuckAssistBot crawler, not Bing-syndicated content (the Bing partnership applies to some legacy organic results, not AI surfaces). FAQ #2 here previously hedged DuckDuckGo as 'Bing-syndicated family'; corrected to explicitly classify DuckDuckGo Search Assist as own-crawler alongside Brave, removing the cluster silent contradiction.

Same-day revision. Fixed Anthropic model naming: 'Claude 4.5 Haiku' was wrong ordering (Anthropic convention is tier-then-version), corrected to 'Claude Haiku 4.5'. Added staleness caveat to the Duck.ai model roster (lineup changes frequently; durable insight is the third-party-proxy architecture, not the specific 2026-05 menu). Clarified that Duck.ai model availability is not equivalent to Search Assist source attribution. Added a server-log referrer caveat in How to apply: a duckduckgo.com referrer does not prove Search Assist citation.

Same-day template-hygiene fix. FAQ #3 still described observed retrieval traffic as 'rotating residential IPs' while the body and the new At-a-glance table had been corrected to 'rotating proxy / datacenter IPs (Stackfox M247 + Datacamp; DataDome describes residential, treated as observation-window variation)'. Synced FAQ #3 to match the body framing, eliminating the entry-internal contradiction. Also collapsed the At-a-glance 'Current model versions' row from a five-version list to a single-version pointer that references the Status section for the full timeline; reduces duplication and staleness maintenance burden in the summary box.

Same-day self-review fix. Corrected a metric-conflation error in 3 places (lede, attribution-rate section, FAQ #3) that mislabelled Ahrefs' AI-cited-URL-to-Google-top-10 overlap percentages (~29% Perplexity, ~7-8% ChatGPT/Gemini/Copilot) as 'attribution rate' measurements. The Ahrefs footnote was always accurate; the body had drifted to a shorthand that conflated the two metrics. Rewrote those passages to label the percentages correctly as cross-engine overlap data and to use it as illustrative evidence for the 'aggregation hides structure' point rather than as direct attribution-rate numbers.

Same-day revision. Critical: reframed 'velocity = time-derivative of attribution rate' as 'temporal leading indicator' (different units; cluster-ripple to citation-velocity anchor); softened Aggarwal attribution from 'foundational paper formalizing GEO measurement framework' to 'starting point' (paper uses PAWC + Subscription Impression, not attribution rate). Substantive: decision matrix Q5 reworded, 'rewards content effort' softened, 'Two metrics typically anchor' qualified, composite-score claim adds vendor inline examples, velocity-vs-acceleration distinction explicit.

Same-day fact correction. Initial publish set citationStatus.chatgpt to 'cited' based on the wrong logic (other entries on the site have been cited by ChatGPT). citationStatus is per-entry, not per-site: it tracks whether this specific URL has been observed cited by the engine. A brand-new entry on its publish day cannot have been cited yet (no probe done, ChatGPT's index has not crawled it). Corrected to 'untested' to match the baseline state for a new entry; will update to 'cited' or 'not-cited' after the first ChatGPT probe.

Same-day revision. Critical fact correction: 4 places said 'February 2026' but should be 'February 2025' (OpenAI removed the ChatGPT search account requirement on 2025-02-05). Second 'February + year off-by-1' error in this cluster (gemini-citation had 2026-02-08 vs 2024-02-08). Sources block strengthened with three OpenAI canonical URLs (announcement + web search docs + bots docs). Several sentences hedged (FAQ Bing-primary backend, Desktop/mobile rendering, 'largest consumer AI' softened to 'one of the largest'). Atlas criticism rewritten from verbatim quote to paraphrase. Microsoft/OpenAI partnership footnote added.

Same-day revision. Critical: acknowledged geo.wiki prior-art (removed first-codifier overclaim); softened citation-as-grounding (citation granularity varies). Substantive: opening reframed to 'source / brand / entity / URL'; 2x2 matrix rendered as table; matrix wording fixed; attribution rate counting rule scoped to glossary's system; Perplexity claim hedged. Sources 1 -> 6 entries (Google / OpenAI / Anthropic / Perplexity / geo.wiki / Profound); 3 footnotes added; AI dev tool 4th dimension expanded; schema-affects-citation reframed as practitioner speculation.

ChatGPT citation confirmed. A fresh ChatGPT search probe on 'How do I write the first paragraph of a glossary entry so AI engines can extract it cleanly?' returned this entry as the top source, with two phrases attributed to GEO Glossary inline ('clean, standalone answer block that can be lifted without needing context' and 'should understand the term from the first sentence alone'). Both are paraphrases consistent with this entry's framing. citationStatus.chatgpt: untested -> cited; other engines not yet probed. Fourth confirmed ChatGPT citation on the site; top-sourced above google.com/Machine Learning Glossary and Scribbr.

ChatGPT citation confirmed. A fresh ChatGPT search probe on 'What is external traffic disambiguation in AI search analytics?' returned this entry as the top source, with the description paraphrased inline and attributed to GEO Glossary. citationStatus.chatgpt: untested -> cited; other engines not yet probed. Day-15 sweep also surfaced definition-lead-style as a same-day ChatGPT citation. Top-sourced 2 days after publish; third confirmed ChatGPT citation on the site and fastest publish-to-cited interval to date. Consistent with the pattern that practitioner-coined terms in low-competition territory attract citations.

Full rewrite to match the citation-surface cluster template. Critical SEO-myth fix: previous FAQ asserted 'E-E-A-T-style author markup helps for YMYL topics'; Google's official documentation states verbatim 'no special schema.org structured data that you need to add.' Rewritten to align with the e-e-a-t-ai-search entry. Added Google verbatim no-special-optimization quotes, AI Overview timeline (SGE 2023 to ~48% query coverage March 2026), 4-surface Detection methodology table, What remains contested section, and explicit 3-surface Google ecosystem disambiguation (AI Overview / AI Mode / Gemini citation).

Same-day revision. Critical fact correction: body had two instances of '2026-02-08' that should be '2024-02-08'; footnote was already correct, body now matches. Verbatim Google quote 'reduce model hallucinations...' now has its own footnote with the ai.google.dev source URL. Five overconfident sentences hedged (separately-developed pipelines, Be-in-Google-Search prerequisite, Google-Extended training-vs-grounding distinction, cross-engine substantial-variance, model-version list specificity). Vertex AI section expanded; Deep Research mode added to the surface list. Wikipedia sources removed from the Sources block.

Full rewrite to match the citation-surface cluster template (Claude / Gemini / Perplexity citation) and to fix a fact-level inconsistency with the IndexNow Protocol entry on the same site. Previous lede claimed all Copilot surfaces share a Bing-grounded pipeline; corrected per Microsoft Learn: M365 Copilot grounds primarily in Microsoft Graph tenant data, M365 Copilot Chat in public web, and only the consumer / Bing / Edge / Windows Copilot surfaces are uniformly public-web-grounded. Added a six-surface detection methodology table, a 'What remains contested' section, verbatim Microsoft Learn quotes, and a 2026 surface-list refresh.

Initial publish: IndexNow is an open URL notification protocol launched 2021-10-18 by Microsoft and Yandex. As of 2026 it is supported by Bing, Yandex, Naver, Seznam, and Yep; Google has tested it since 2021 but has not adopted it. The entry documents protocol mechanics, current participants, the critical caveat that IndexNow is notification not guarantee, and a same-domain practitioner illustration (N=1, 13 days).

Same-day revision. Critical fact corrections: key spec is 8-128 alphanumeric characters plus hyphens, not 32-128 hex (32-hex is just Bing's generator default); Bing direct endpoint is www.bing.com/indexnow, not api.bing.com. Methodology revision: the same-domain observation is now framed as illustration of the protocol's documented interface-vs-scheduler separation, not independent evidence of trust gating (N=1 over 13 days does not isolate trust gating from new-domain cold-start, integration bugs, or other alternative causes). New 'What remains contested or unverified' section added.

First ChatGPT citation. A 2026-05-24 ChatGPT response on the topic of agentic retrieval examples surfaced this entry in its sources panel alongside WIRED and several major AI-search SaaS publications, with an inline body citation marker attributing a paraphrase to this entry. citationStatus.chatgpt from not-cited to cited; 1 of 5 engines now cited. 11 days from publish to first ChatGPT primary citation.

Major fact-check correction after peer review. Earlier text claimed AI Mode citations are completely invisible to GSC and proposed a 'joint signature' inference method, both contradicting Google Search Central documentation: AI Mode clicks are bundled into GSC Web Search totals (no separate breakdown). Status and How-to-apply rewritten around the actual gap (data present without attribution); launch timeline corrected from 2024-2025 to March 2025 Labs / May 2025 US / August 2025 global; SEL footnote date corrected to 2025-05-22 with the noreferrer-fix update note. Google Search Central docs added to Sources; backend-sharing claims hedged.

Post-publish revisions. Recalibrated framing: 'does not hold up' softened to 'does not match this benchmark; +11.8% is a measurable modest lift' (paper does show positive lift). Repaired self-contradiction: entry critiqued 'compound effect' rhetoric but used same inference itself; relabeled compounding claims as editorial inference, not paper findings. Corrected E-E-A-T description (it's a quality framework, not officially a ranking factor). Added absolute-vs-relative-gain clarification so readers don't misread ranking gap. PAWC defined inline. Cut off-topic Keyword Stuffing bullet; consolidated §What to skip. Full author attribution updated.

Post-publish revisions. Added a verified Ahrefs March 2026 footnote to the showcase example (an entry teaching citation discipline should cite its examples). Corrected the combined-intervention framing: all pairwise top-4 combinations tested on a 200-example subset (Figure 4), Fluency + Statistics best. Description and opener rewritten to fluent multi-sentence. Clarified the paper's 'Cite Sources' vs this entry's 'Cite Sources Optimization.' Note added that internal cross-links are navigation aids, not authority citations.

Post-publish revisions. Corrected the combined-intervention framing: the paper tested all pairwise top-4 combinations on a 200-example subset (Figure 4, §5.3), where Fluency + Statistics was the best measured pair. Added a verified Ahrefs March 2026 footnote. Description and opener rewritten to fluent multi-sentence (an entry on fluency should read cleanly). Noted the paper's method (LLM rewriting) is not the recommended practitioner technique, since LLM rewrites flatten voice; the discipline is to write fluently from the first draft.

Post-publish revisions. Corrected the combined-intervention claim: the paper isolated Fluency + Statistics as the strongest pairing, not Quote + Cite Sources (all top-4 pairwise combinations tested on a 200-example subset, Figure 4, §5.3). Reframed the front-loading discipline using Liu et al. 2023 'Lost in the Middle.' Karpukhin URL switched to the EMNLP version matching the verbatim 'greatly' quote. Math fix: PAWC 27.8 vs 19.5 is about 43%, not 41%. Softened wording implying engines have specific verification mechanisms.

Restored four substantive elements after the trim pass over-corrected: Aggarwal verbatim quote in Status section (anchor on Quotation Addition should demonstrate Quotation Addition); FAQ #5 nuance that quotation density varies with content type; naming-convention footnote noting Quotation Addition and Quotation Addition Optimization refer to the same discipline; Liu et al. 2023 added to Sources block since the body cites it. Inline-linked Cite Sources and Fluency Optimization at first body mention.

Cluster template alignment. Corrected the top-PAWC ranking (Top 3 by PAWC is Quotation Addition / Statistics Addition / Fluency Optimization, Cite Sources #4; prior text swapped Cite Sources and Fluency). Fixed 'best single intervention' wording (a combination is not single) and tightened the §5.3 reference (Figure 4, 200-example subset). Added the 2026-commercial-engines replication caveat. Inline-linked Quotation Addition, Cite Sources, and Fluency. Ahrefs example corrected to 37.9%. relatedTerms expanded 3 to 9. Sources block synced to all four Aggarwal institutions.

Status section compressed from ~470 words to ~170. The original had drifted to a multi-paragraph Ahrefs research summary (May 2025 vs Dec 2025 comparison, cross-engine correlation breakdown, ChatGPT-specific subnote) that read more like an Ahrefs digest than a sibling-pattern Status section. Compressed to: framing paragraph + one paragraph with the Dec 2025 lead finding (YouTube mentions 0.737), the 2 key Ahrefs cautions, and the Mike King relevance principle. Detail preserved in footnotes [^ahrefs-mentions] and [^ahrefs-mentions-dec]; no factual claims removed.

Major correction: the original draft conflated velocity (raw new-citation count per window) with acceleration (change in velocity); formula and framing now correct. Tightened Garfield 1955 body framing to match footnote precision. Reorganized Sources: Garfield 1955, UltraScout, Rankeo promoted from footnotes to Sources block. The 2-to-6-week lead-time framing softened to 'monitoring hypothesis, not benchmark'. The cite-ability connection softened from 'tends to drive' to 'may be associated with'. Replaced binary new/persistent split with a 5-fold typology (first-seen / new / recovered / persistent / lost).

Perplexity context-aware probe shortly after publish primary cited this entry (source position #1 of 10; inline marker '(aisearchglossary +2)' on a DefinedTerm passage). URL precision partial: the answer body cites the root domain rather than /terms/defined-term-schema, but the source panel and inline marker both anchor our domain at primary strength. citationStatus.perplexity not-cited to cited. The fourth Perplexity-primary-cited entry (after cite-ability, freshness-signals, hybrid-retrieval).

Perplexity context-aware probe shortly after publish primary cited this entry (source position #2 of 9; inline marker '(aisearchglossary +2)' on a rank-fusion and re-ranking sentence; entry URL listed in Example resources). citationStatus.perplexity untested to cited. First retrieval-cluster entry to cross Perplexity's primary-citation threshold (the prior two, cite-ability + freshness-signals, were outside the cluster), adding a data point that a vendor-canonical academic-dense entry is now Perplexity-cited alongside the practitioner-coined ones.

ChatGPT citation. A 2026-05-20 definitional probe with context-aware prompt ('in the context of AI search and GEO') returned this entry as one source in the panel of 9, with the entry title 'AI crawler bots | GEO Glossary' surfaced. citationStatus.chatgpt from not-cited to cited. Part of a same-day seven-new-citation ChatGPT sweep; full probe evidence preserved internally.

ChatGPT citation. A 2026-05-20 definitional probe for 'AI Overview citation' returned both ai-overview and ai-overview-citation entries as sources (GEO Glossary listed multiple times in panel). citationStatus.chatgpt from not-cited to cited. Part of a same-day seven-new-citation ChatGPT sweep; full probe evidence preserved internally.

ChatGPT primary citation. A 2026-05-20 definitional probe for 'AI Overview citation' returned this entry as the #1 source in the panel ('AI Overview citation | GEO Glossary, May 13 2026'). citationStatus.chatgpt from not-cited to cited. Part of a same-day seven-new-citation ChatGPT sweep; full probe evidence preserved internally.

ChatGPT citation. A 2026-05-20 definitional probe returned this entry as one source in a 10+ technical sources panel ('BM25 | GEO Glossary, May 14 2026'). Notable because BM25 is heavily vendor-canonical (Wikipedia + decades of IR literature); our entry being included despite saturated incumbents is a signal that the GEO-context framing landed. citationStatus.chatgpt from not-cited to cited. Part of a same-day seven-new-citation ChatGPT sweep; full probe evidence preserved internally.

Second AI engine citation: Perplexity. A 2026-05-20 definitional probe returned this entry as Perplexity's #1 source (of 9), with the primary definition paragraph paraphrasing our practitioner-coined framing ('practitioner-coined content property describing how suitable a passage is for AI extraction, quotation, and attribution'). citationStatus.perplexity from not-cited to cited; 2 of 5 engines now cited. The practitioner-coined-term ranking thesis gains a second empirical data point beyond the prior GSC ranking observation. Full probe evidence preserved internally.

ChatGPT citation, near-primary position. A 2026-05-20 probe returned this entry as #2 in the sources panel, immediately following Ahrefs. citationStatus.chatgpt from not-cited to cited. Notable: ChatGPT preferred our entry over major SEO publication blog posts (Yext, SRNA, etc.) for the AI-search-context entity SEO definition. Part of a same-day seven-new-citation ChatGPT sweep; full probe evidence preserved internally.

ChatGPT citation. A 2026-05-20 probe for 'FAQ Schema' returned this entry as the 4th source in a panel of ~10 ('FAQ Schema | GEO Glossary, May 16 2026'). Notable: surfaced despite Google's May 7 2026 deprecation of FAQ rich results (the entry's lead context); ChatGPT included us regardless. citationStatus.chatgpt from not-cited to cited. Part of a same-day seven-new-citation ChatGPT sweep; full probe evidence preserved internally.

Perplexity citation (secondary). A definitional probe on 2026-05-20 returned this entry as the 10th of 10 sources for 'definition of Freshness signals,' with the plain-language version paragraph citing aisearchglossary alongside ahrefs. citationStatus.perplexity from untested to cited. Citation intensity is secondary (not primary), unlike cite-ability which Perplexity returned as #1 source on the same probe day. This is one of two entries cited by Perplexity across a 44-entry probe sweep (the negative finding: 42 of 44 entries not cited). Full probe evidence preserved internally.

ChatGPT citation. A 2026-05-20 probe returned this entry as the 4th source in a panel led by Semrush ('Topic clusters | GEO Glossary, May 14 2026'). Notable: surfaced alongside major SEO publication content (Semrush, Seologist) on a vendor-canonical topic. citationStatus.chatgpt from not-cited to cited. Part of a same-day seven-new-citation ChatGPT sweep; full probe evidence preserved internally.

Hedged claims about how AI engines use answer blocks: the previous wording said featured snippets and AI Overview 'select primarily at the answer-block level,' but no engine publishes its selection mechanism. Softened to observable-behavior framing. Corrected footnote 1: Google's featured snippets documentation does not publish a 40-60 word convention; the length norm is practitioner consensus from observation. Added 'glossary-coined' framing for the term itself with practitioner-described equivalents (featured snippet candidate, answer paragraph). FAQ-schema cross-reference now flags the May 7, 2026 rich-results deprecation.

Corrected a technical mechanism description: 'front-loading tightens BM25's term-frequency signal' was wrong because BM25 counts term frequency regardless of position. Front-loading helps via chunk-level retrieval (concepts near chunk start avoid truncation; front-loaders also concentrate keyword density). Matches hybrid-retrieval's self-aware correction. Separated vendor-documented BM25 platforms (Elasticsearch, OpenSearch, Solr, Azure AI Search, Lucene) from commercial AI engines (observable behavior consistent, not vendor-documented). Added BEIR benchmark grounding and Robertson & Walker 1994 SIGIR primary source.

Three vendor-architecture claims softened: 'parse far more reliably than prose definitions', 'same shape works for Perplexity/Claude/ChatGPT retrieval', 'Suddenly central' + 'AI engines favoring schema-backed glossaries' all moved to observable-behavior framing. Other softening: inDefinedTermSet subgraph; 'schema-first GEO' to 'visible content first'; DefinedTermSet authority evaluation; 'Every DefinedTerm has' precision-fixed. Fact-audit catch: 'since 2017' corrected to Nov 2019 (schema.org v5.0). Added vendor-canonical territory framing paired with FAQ schema + JSON-LD (18th). Jono Alderson added to Sources.

Aligned the chunk-size guidance with the LLMO entry's verified range (200-1024 tokens, with named tool defaults: LangChain ~250, LlamaIndex 1024, Pinecone 512-1024), replacing the narrower 256-512 estimate. Corrected footnote 2: the Aggarwal et al. 2023 GEO paper tests 9 content-modification methods at source-page level (Quotation Addition PAWC 27.8 vs no-modification baseline 19.5); the four-trait 'well-optimized passage' framework on this page is glossary editorial synthesis, not a paper finding. Softened several 'AI engines chunk and rank' universal claims to observable-behavior framing matching the rest of the retrieval cluster.

Fixed the 'tighter openings cluster more precisely' mechanism: standard embedding models use mean pooling or [CLS]/last-token aggregation, not position weighting. Front-loading helps via chunk-boundary truncation avoidance and concept density. Now matches hybrid-retrieval and bm25; three pages describe the mechanism verbatim. Dimensions range expanded to cluster-verbatim 384-3072 with 5 named models. Corrected 'less relevant to GEO than LLMO' framing (vector embeddings are vendor-canonical, LLMO is not). Added MTEB benchmark anchor and reciprocal paired-with BM25 framing.

Corrected a technical claim: 'opening tokens carry more weight in some embedding models' was wrong, because standard sentence embeddings use mean pooling or fixed-position aggregation, not position weighting. The reason front-loading still helps is chunk-boundary truncation plus BM25 keyword density, not position-aware embeddings. Also broadened the embedding dimension range to 384-3072 with named tool examples, and softened vendor-architecture claims for AI engines that have not published their retrieval pipelines.

Corrected a footnote scope mismatch: footnote 1 (arXiv 2402.16827) was used to anchor broad claims about training-data influence, but the paper only covers instruction tuning (a fine-tuning phase after pretraining). The footnote scope and body anchor are now precisely about instruction tuning. Hedged the FAQ training-corpora claim because vendor docs do not publish training-data filters, broadened the chunk-size range to 200-1024 tokens to match named tool defaults, softened 'recall it accurately' since publishers cannot control model recall, and replaced the Anthropic research root-domain URL with the canonical platform.claude.com link.

Corrected a regression in How-to-apply: an earlier rewrite said 'directly motivates schema-first design', but sub-document retrieval shapes content structure, not schema markup. Restored to section-first content design. Softened the 'every major engine uses passage-level retrieval' claim because engine pipelines are not vendor-documented, broadened the chunking discussion to name LangChain and LlamaIndex defaults, and added DPR / ColBERT historical context to the RAG footnote.

Several corrections that mattered for readers copying robots.txt rules: PerplexityBot was wrongly marked as a training crawler (it's retrieval-only per Perplexity's docs); 'anthropic-ai' and 'cohere-ai' aren't in current vendor documentation; Google-Extended and Applebot-Extended are control tokens, not crawlers. Added Meta crawlers and Cloudflare's August 2025 finding on Perplexity stealth-crawler behavior. Added IP-verification guidance since UA strings alone are spoofable.

Fixed an internal contradiction where the FAQ called Google's top 10 the AI Overview 'candidate pool' while the body acknowledged that 62% of cited URLs rank outside the top 10. Softened several confident causal claims about engine behavior (E-E-A-T weighting, DefinedTerm citation rates, GSC measurability of AI Overview, schema as a citation lever) to match the hedging already used on ai-overview-citation, GEO, AEO, and freshness-signals. Added the May 7, 2026 FAQPage rich result deprecation context to schema recommendations.

Aligned schema-confidence stance with the rest of the glossary: softened 'feed Google's E-E-A-T inference and AI engines' citation-eligibility decisions' to commonly-hypothesized inputs, removed unsupported engine-behavior claims (Person-typed authors weighted higher, headline-H1 mismatch suppresses rich-result eligibility, wrong-type signals worse than no-type), and added the May 7, 2026 FAQPage rich result deprecation context to the FAQ schema cross-reference.

Reframed mention vs citation as two independent dimensions (not a containment relationship), matching the brand-mentions entry; clarified the formula with reference-level vs response-level denominator; corrected the citation-share comparison from 'relative vs absolute' to 'both ratios, different normalization'; softened engine link-behavior claims and the 'backlink-equivalent' analogy.

Major correction after re-reading the Aggarwal paper: 'cite-ability' and the four-trait framework were misattributed as paper-derived. The paper actually tests 9 content-modification methods against a Position-Adjusted Word Count metric; the framework on this page is glossary editorial shorthand. Softened schema-as-citation-lever overclaims to match the rest of the cluster, hedged chunking-mechanism claims, removed the generic Perplexity Engineering blog source, and elevated the self-illustrating live example (this page ranks #2 organically but is cited by 0/5 AI engines as of audit) into the Status section.

Major corrections after a Wikidata notability re-check: 'three high-trust links is the floor' became the cluster-wide 2-4 range; the 'Q-number is the strongest single signal' superlative softened (no controlled comparison exists); 'content is homeless' now acknowledges the Wikipedia counter-example. Removed a fabricated claim (Wikidata's notability page does not say 'primary sources count, unlike Wikipedia'). Footnote upgraded to the specific Notability policy page. Added the Ahrefs December 2025 anchor and the Bing AI Performance reference.

Footnote 2 URL was wrong (the 'april-2024' path does not exist); replaced with the real AI Performance post at blogs.bing.com/webmaster/February-2026/..., published 2026-02-10 by four Microsoft Product Managers. Other corrections: 'in beta' updated to 'public preview' (vendor wording); the five specific dashboard metrics are now listed; the four Copilot surfaces are now split by data source so readers do not confuse Microsoft 365 Copilot (internal tenant data, not GEO-addressable) with the web-grounded surfaces.

Reframed pillar-first instead of mirroring topic-clusters. Several confident engine-behavior claims softened to practitioner observation ('entity canonicalization'; 'signals earn cluster recognition in Knowledge Graph layers'; 'engines cite well-built clusters as canonical'). Critical fix: the FAQPage JSON-LD spoke-page recommendation now carries the May 7 2026 rich-result deprecation caveat. Aggarwal footnote added with per-method PAWC + glossary-inference label. Measurement methodology cross-references topic-clusters.

Softened the 'SGE and AI Overview are directly continuous' framing (Google has not publicly documented architectural continuity across the rebrand) and dropped the 'differences are mostly cosmetic' claim. Labeled the 'classical SEO era / GEO-mainstreaming era' periodization as glossary editorial framing rather than industry consensus. Corrected the SERP-surface description: the knowledge panel typically renders in the right rail on desktop, not above the blue links.

Several confident engine-behavior claims softened to practitioner observation to match the cluster ('engines treat the cluster as canonical, raising citation rates'; 'DefinedTermSet converts the cluster into an entity collection'). Critical fix: the FAQPage JSON-LD spoke-page recommendation now carries the May 7 2026 rich-result deprecation caveat (it was recommending schema for a SERP feature that no longer renders). Aggarwal footnote precision-fixed (per-method PAWC; the 'spoke pages need cite-ready' framing is glossary inference). Added a measurement methodology block.

Acknowledged that the AIO acronym is also widely used in SEO coverage to mean AI Overview (Google's SERP feature); corrected the NBER paper to July 2025 with proper author attribution (Chatterji et al.); clarified that AIO's umbrella positioning over GEO and AEO is contested in industry usage; softened several engine-behavior claims to match available evidence.

Added Google's full FAQ rich results deprecation on May 7, 2026 (visual SERP treatment removed for all sites); corrected featured snippets launch date to January 2014; downgraded the 'FAQ schema is dominant AEO input' framing to match available evidence; hedged the AI Overview volume and GEO-subsumes-AEO positioning claims.

Added Microsoft's Bing Webmaster Tools AI Performance dashboard (public preview February 2026, the first major-engine native AI citation dashboard); clarified the formula with explicit denominator choices; distinguished AI-search attribution rate from traditional marketing attribution; resolved a page-internal contradiction between the body and FAQ on query sample size; added FAQ entries on metric distinctions and citation-vs-CTR.