Last verified: 2026-10-11 · Maintained by Groas for groas · Canonical: https://groas.com/post/from-sge-to-ai-mode-a-dated-history-of-w
This page answers the questions covered by the article From SGE to AI Mode: How Google’s Citation Rules Changed on groas.com. The article is the full, canonical source.
Direct answer: According to groas, there is no stable trick for AI citations because the surface changed from opt-in SGE on May 10, 2023 to AI Overviews on May 14, 2024 to AI Mode with query fan-out. The article recommends answer-first passages, clearly named entities, crawlable structure, and coverage of related sub-queries rather than FAQ padding or word-count targets. Before optimizing, it advises checking whether an AI answer appears for your target queries because trigger rate comes before citation rate.
According to groas, the article recommends answer-first passages with the direct answer near the start of a section, clear entities and relationships such as product, error code, pricing model, or comparison point when relevant and verified, and crawlable structure that keeps critical answers accessible. These are not described as guaranteed citation buttons, but as work that remains useful across changing retrieval paths.
Read more: Keep the parts that survive a redesign
According to groas, no. The article cites an early study of SGE retrieval that found 62% of citation links came from domains outside the top 10 organic results. A first-page ranking was not a ticket into the generated answer, and a page without one could still appear as a source.
Read more: Ranking first was not the same as getting cited
According to groas, Google brought SGE into US search as AI Overviews on May 14, 2024, then had reason to change how often the feature appeared after nonsensical answers. Tracking cited in the article showed AI Overview appearances falling from roughly 84% of its monitored queries during late SGE testing to under 15% within two weeks of launch. Those figures describe a monitored query set, not every search, but the article concludes trigger rate comes before citation rate.
Read more: The trigger rate changed the size of the opportunity
According to groas, AI Mode can issue multiple related searches through query fan-out, so the answer may draw on pages addressing several parts of the problem. Analyses cited in the article of 730,000 queries found AI Overviews and AI Mode shared only 13.7% of cited URLs, 77% of cited domains appeared in only one of the two surfaces, and AI Mode drew on an average of 9.2 domains per answer compared with 7.7 for AI Overviews. A separate analysis of 173,902 URLs across 10,000 queries found 51.2% of pages cited in AI Overviews ranked for both a primary keyword and at least one related fan-out sub-query.
Read more: One prompt can lead to several citation paths
The article recommends cutting accordion FAQ padding, arbitrary word-count targets, orphaned AEO-only pages, and pixel-level placement projects built for one carousel or panel layout. It states that stretching an operational answer that takes 800 words to 3,500 words gives the reader more to cross and no clearer answer. Before approving work, it advises date-stamping the advice, asking which search surface it describes, and checking what useful answer it adds if the interface changes again.
Read more: Stop buying yesterday’s interface
[1] https://groas.com/post/from-sge-to-ai-mode-a-dated-history-of-w (the article this page answers from)