Google AI Overviews (AIO) is the product name for the AI-generated answer blocks that have been broadly deployed across Google's SERPs since 2024 — following the test phase under the name Search Generative Experience (SGE). In Germany and several other European countries they have been available since March 2025. AIO condenses information from multiple sources into a single summarising answer with numbered links.
The result: for informational and comparative queries, a substantial share of impressions migrates from the classical organic slots into the AIO block — with corresponding click losses for positions one through three: according to Ahrefs, the Position 1 CTR is 58 % lower on keywords with an AI Overview (December 2025), and Seer Interactive measures a 61 % drop in organic CTR on informational AIO queries.
How do Google AI Overviews work technically?
Google AI Overviews generate an AI answer from pages that are already in Google's index and link those pages as sources. According to Google's documentation AI features and your website, AI Overviews and AI Mode may use a "query fan-out" technique for this: the system issues several related searches across subtopics in parallel to gather additional supporting pages. Google says there are no additional technical requirements and no special optimizations needed to appear; a page has to be indexed and eligible to show in Search with a snippet. Display is controlled through the normal preview controls such as nosnippet, max-snippet or noindex, not through Google-Extended. Google's guide to optimizing for generative AI features (updated July 10, 2026) adds that llms.txt files and artificial "chunking" can be ignored for Google Search, and that pages built mainly for individual fan-out variations violate the scaled content abuse policy. Helpful, self-contained content remains what counts.
Our working model of the AIO pipeline (reconstructed from tests and public Google statements, not a confirmed Google fact): AIO works on Google's own corpus with a passage-based retrieval model. The model has six steps: (1) query classification — is the query AIO-eligible? (2) sub-query generation to cover the different facets of the question, (3) retrieval of relevant passages from Google's index, prioritised by classical signals plus chunk readability, (4) passage ranking at the semantic level, (5) answer synthesis via Gemini with citation attribution, (6) rendering as an AIO block with links.
The decisive difference from classical SEO: AIO operates at the passage level, not the document level. A page with mid-tier ranking but structurally clean passages can be preferred for citation over a position-one page with poor chunk quality.
CTR for position one on keywords with an AI Overview (Ahrefs, Dec 2025)
optimization layer — no longer the document
structural signals that shape AIO citation
The eight levers for AIO citations
1. Passage engineering. Paragraphs that stand on their own (usually 200-400 tokens in our tests), the first sentence defining the claim, evidence in the next two, explicit entity mentions, concrete numbers. This is not technical "chunking", which Google says it does not need, but clear structure for readers. See passage ranking.
2. FAQ content with real questions. Not generic marketing FAQs but the actual queries from Search Console and "People Also Ask", with precise, self-contained answers. The lever is the answer structure, not the markup: FAQPage schema is worth keeping as a semantic declaration, but it no longer brings a SERP feature or a proven citation boost.
3. Step structure for procedural content. Clearly numbered, self-contained steps, especially strong on tutorial and instruction queries. HowTo schema can be kept as a supplementary semantic declaration; rich results for it no longer exist.
4. Article plus author entity. Every editorial piece carries Article schema with author reference by @id to a Person schema and publisher-@id to Organization. See author entity for E-E-A-T.
5. Reinforce E-E-A-T signals. Experience, Expertise, Authoritativeness, Trustworthiness. Decisive on YMYL topics (health, finance, legal). Author bios with credentials, external citations, an E-E-A-T-aligned content architecture.
6. Knowledge-Graph coherence. Wikidata item with references, Schema-@id graph, sameAs cluster. AIO resolves entities against the Knowledge Graph — an unambiguous entity is more likely to be cited than an ambiguous one.
7. Freshness signals. dateModified in the schema, current statistics, an XML sitemap with an accurate lastmod (Google only uses the value if it is verifiably accurate). In our observation AIO prefers current sources for time-sensitive queries.
8. Query-intent matching. A page has to cover the exact sub-query dimensions that Gemini generates when building the AIO. That requires fan-out query analysis — not only the head keyword, but the sub-questions Gemini derives from the main query. This means coverage within one page or topic cluster, not a separate page per fan-out variation, which Google treats as scaled content abuse.
| Query type | CTR decline on position one | AIO trigger rate | Primary lever for citation |
|---|---|---|---|
| Informational (definition, what-is) | 40-60% | High | Claim-evidence passages + genuine question-answer structure |
| How-to / tutorial | 30-50% | High | Step structuring with self-contained steps |
| Comparison / vs. | 25-45% | Medium | Comparison tables + entity @id graph |
| Local | 10-20% | Low | LocalBusiness schema + GBP maintenance |
| Transactional | 5-15% | Rare | Product schema + reviews |
| YMYL (health/finance) | 15-30% | Controlled | E-E-A-T signals + author entity |
How many of your queries trigger AIO?
A 30-minute live analysis of your top-50 informational queries: AIO trigger rate, own citation presence, competitive share and the three biggest structural levers.
What AIO does NOT increase
Four common misconceptions. (a) "More content lifts AIO probability." Wrong: AIO selects by passage quality, not word count. In our tests, unstructured walls of text make it harder to find a citable passage. (b) "Keywords in title/H1 are enough." No — AIO works at the semantic embedding layer. Keyword density is meaningless; entity salience and claim clarity decide.
(c) "AIO is just a short-term trend." No — Google is investing massively, AIO is expanding into further query types, and the infrastructure is being integrated more deeply. Skip optimization now and you lose structurally over 18 to 36 months. (d) "An llms.txt file or special markup earns AIO citations." Not according to Google: such files are ignored for Google Search, and no special structured data is required for AI Overviews.
Measurement: tracking AIO visibility
Combine three data sources. (1) Google Search Console: the generative AI performance report, announced on June 3, 2026 and available to all websites since August 31, 2026, reports impressions from AI Overviews, AI Mode and generative Discover features separately. Clicks, CTR and queries are still missing, and so is API access; citation tracking still needs your own prompt sampling. Details in our article on the GSC generative AI report. (2) Proprietary prompt tracking: geo-IP-controlled queries via a headless browser against Google Search, automated parsing of the AIO block, storage of citation URLs. (3) Third-party tools: AlsoAsked, seoClarity, Similarweb AI Insights.
In our LLM citation monitoring we combine all three sources into weekly AIO citation-rate reports with competitive comparison.
Bottom line: AIO is the new position zero
Featured snippets were called "position zero". AIO is the new position zero — but with much wider consequences: multiple sources per answer, structurally different signal weighting, a major click shift. Brands that do not actively play AIO lose impressions to competitors that do.
The optimization is not a content-production exercise but a structural refactor: passage engineering, schema graph, entity consolidation. The 90-day protocol from the ChatGPT SEO guide applies here in structurally analogous form — with AIO-specific extensions for question-and-answer structure and Google Search Console monitoring.
Sources
- Google Search Central: AI Features and Your Website (continuously updated)
- Google Search Central: Google's Guide to Optimizing for Generative AI Features on Google Search (last updated July 10, 2026)
- Google Search Central Blog: Introducing Search Generative AI performance reports in Search Console (June 3, 2026; rolled out to all websites on August 31, 2026)
- Google Search Central: Build and submit a sitemap (continuously updated)
- Google: AI Overviews Europe: 9 New Countries, 4 New Languages (March 25, 2025)
- Ahrefs: Update: AI Overviews Reduce Clicks by 58% (February 4, 2026)
- Seer Interactive: AIO Impact on Google CTR: September 2025 Update (November 4, 2025)