SGE SEO: What Was SGE and How to Optimize for AI Overviews Now
What Was SGE (Search Generative Experience)?
SGE, or Search Generative Experience, was Google's experimental AI-powered search feature that appeared at the top of search results. It used large language models to generate a synthesized answer to a query before showing traditional blue links. Google launched SGE in limited beta through its Search Labs program in May 2023.
Unlike featured snippets, which pull a verbatim excerpt from a single page, SGE generated original prose by synthesizing information from multiple sources. Those sources were then cited as small cards alongside the AI-generated response, giving users the option to click through.
SGE was most active on informational and research-oriented queries. Shopping, local, and navigational searches saw it less frequently. The feature was controversial: publishers worried about reduced click-through rates, while searchers generally found it useful for complex multi-part questions.
SGE Was Renamed to AI Overviews in May 2024
At Google I/O in May 2024, Google officially renamed SGE to AI Overviews and rolled it out to all U.S. users. The rename marked the transition from an experimental feature to a core part of Google Search. By late 2024, AI Overviews were live in over 100 countries.
The underlying mechanics did not change dramatically with the rename. Google still uses its Gemini models to generate answers, and it still selects source pages to cite alongside those answers. However, Google introduced several refinements after the initial rollout — most notably reducing the frequency of AI Overviews on queries where they caused factual errors, and pulling back on health and medical topics following high-profile accuracy incidents shortly after launch.
For SEOs, the rename means that any guidance labeled "SGE optimization" from 2023 is now more accurately called AI Overviews optimization. The strategic principles are largely the same, but the feature is now stable and no longer experimental.
How SGE / AI Overviews Select Content Sources
Google has not published an exact algorithm for AI Overview source selection, but its guidance and observed behavior point to several consistent patterns. The system favors pages that already rank well for a query — being in the top 10 organic results is a strong predictor of being cited in the AI Overview for the same query.
Beyond ranking, Google appears to weight content that directly and explicitly answers the searcher's question. Pages that bury their answer in introductory fluff are less likely to be selected than pages that lead with a clear, specific response. Structure helps: pages with properly marked-up headings, numbered steps, and definition-style formatting give Google more parseable signal.
E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) is a documented factor in Google's quality evaluation. Pages from recognized authorities in their field — organizations with strong brand signals, authors with established credentials, and sites with solid backlink profiles — are more consistently cited than thin or anonymous content.
Freshness matters for fast-moving topics. Google tends to cite recently updated pages when the query implies current information is needed. Keeping your dateModified schema accurate and genuinely updating content (not just bumping a date) is important.
SGE SEO Strategies That Still Apply Today
The core strategies developed during the SGE beta period remain valid for AI Overviews optimization. The most important is targeting question-based queries explicitly. If someone searches "how does X work," your page should contain a direct, self-contained answer to that exact question — not just tangentially related content.
Depth and specificity beat thin coverage. A page that thoroughly explains one topic is more citation-worthy than a page that superficially covers many topics. Google's models are good at detecting when content is substantive versus when it is generic padding.
- Write concise, direct answers to your target queries in the first 100–150 words of each section
- Use H2 and H3 headings that mirror the phrasing of related questions
- Include original data, expert quotes, or first-hand experience to signal E-E-A-T
- Link internally from strong pages to AI Overview candidates to pass authority
- Ensure your sitemap includes all candidate pages so Google can discover and index them promptly
One strategy that gained traction during the SGE era is the "inverted pyramid" format for informational content: lead with the answer, then provide the evidence and explanation below it. This mirrors how journalists write news articles and how search models prefer to extract answers.
Content Formatting for AI-Generated Answers
Formatting is one of the most actionable levers for AI Overview eligibility. Google's models extract structured content more reliably than prose-heavy text. This does not mean you should convert all content to bullet lists — it means you should use the right format for each content type.
For procedural content (how to do something), numbered lists with one action per step are ideal. For comparative content (X vs Y), tables provide clear structure. For definition and explanation content, a short direct definition paragraph followed by deeper explanation works well. For factual questions with a single best answer, a single bold sentence at the top of the relevant section signals the answer clearly.
Avoid walls of text without headings. Google needs structural signals to understand where one idea ends and the next begins. A 2,000-word article with no H2s is much harder for an AI to parse than the same content organized under six clear subheadings.
Short sentences help. AI models are better at extracting crisp, specific claims from short sentences than from long, clause-heavy constructions. Aim for an average sentence length under 20 words in sections you most want cited.
Schema Markup for SGE / AI Overview Eligibility
Structured data does not directly determine whether a page is cited in an AI Overview, but it provides Google with machine-readable confirmation of what your content is about and how it is structured. This indirectly supports eligibility by making your content easier to index and classify.
Article schema is the baseline for any informational content. It tells Google the headline, description, author, publisher, and publication dates — signals that contribute to E-E-A-T evaluation. FAQ schema is particularly valuable because it explicitly marks question-answer pairs, which are exactly the format AI Overviews are designed to surface. HowTo schema serves a similar function for procedural content.
The speakable schema type — originally designed for voice search — marks specific passages as suitable for AI-generated reading. While Google has not confirmed speakable directly boosts AI Overview selection, it is a logical signal to include on key answer passages.
Ensure all schema is valid and error-free. Invalid structured data does not provide positive signals and may create negative ones. Use Google's Rich Results Test to validate before publishing.
What Changed Between SGE and AI Overviews
The most significant change was availability. SGE was opt-in via Search Labs and available only in the United States. AI Overviews are on by default and available globally. This dramatically expanded both the opportunity for citation and the traffic impact of being excluded.
Google also became more selective about which queries trigger AI Overviews after the rollout. During the SGE beta, AI responses appeared on a very wide range of queries. After a series of accuracy controversies in mid-2024, Google pulled back on health, medical, legal, and financial queries. These YMYL categories now see AI Overviews much less frequently, or not at all.
The citation card UI changed slightly — from the large carousel format seen in SGE to a more compact inline citation format in AI Overviews. Pages can now appear as one of several cited sources rather than always as a prominent card. This means a single page being cited may generate fewer clicks than during the SGE beta, but appearing in more AI Overviews increases total visibility.
Google also added a "Show more" toggle in some AI Overviews, and introduced the ability for users to opt out of seeing AI Overviews in their search settings — a significant shift that gives users more control but makes AI Overview impression data in GSC less reliable as an absolute traffic signal.
Tracking Your Appearances in AI Overviews via GSC
Google Search Console added AI Overviews reporting in 2024. In the Performance report, you can filter by "Search type: AI Overviews" to see which queries trigger AI Overview impressions for your site and how many clicks those impressions generate. This is currently the most reliable first-party data available on AI Overview performance.
Interpret GSC AI Overview data carefully. Impressions are counted when your page is cited, whether or not the user expands the AI Overview. Clicks are counted when a user actually clicks through to your page from the citation. The click-through rate from AI Overview citations tends to be lower than from traditional organic listings because many users find their answer within the AI response itself.
Use the data to identify which of your pages are being cited and for what queries. Pages that appear in AI Overviews but generate low clicks may benefit from stronger calls to action or more unique information that incentivizes users to visit the full page. Pages that rank well organically but never appear in AI Overviews are candidates for content restructuring and schema additions.
Ensure your sitemap is complete and error-free so that your highest-quality pages are consistently crawled and indexed — a prerequisite for any AI Overview appearance. A broken sitemap or crawl errors on key pages can silently exclude them from consideration.