Generative engine optimisation for AI chat visibility

Yes, Google is still the leading search engine, holding more than 85% of the total global share of the search market, according to Statcounter. Users rely on it for finding brand platforms, navigating to specific websites, and making purchases. For now, Google dominates searches with high intent, including:

• Navigational searches: users looking for a specific brand or website.
• Commercial searches: users researching brands and services.
• Transactional queries: people ready to purchase.

Meanwhile, generative AI platforms are rapidly growing in popularity for informational and research-focused queries. They excel at top-of-funnel experiences where users are exploring and learning rather than transacting immediately.

In Q2 2025, 49% of searches on Google were navigational, compared to 34% on ChatGPT. Meanwhile, 52% of searches on ChatGPT were informational, compared to 36% on Google. This shows AI tools are more focused on helping users learn or explore topics.

From 2015 through 2022, Google’s share of the global search market never fell below 90%—but that streak ended when ChatGPT exploded onto the scene in late 2022. As of Q2 2025, Statcounter and other trackers still peg Google at roughly 87–89% of all queries, and the rate of decline has since slowed. Even so, the emergence of generative AI platforms represents a genuine inflection point in how people find information.

Conversational AI tools like ChatGPT, Gemini, Claude, Perplexity, DeepSeek and Grok are carving out a niche for top-of-funnel research. Users value their instant, conversational answers and are increasingly kicking off informational searches there rather than in a traditional search bar. As generative AI continues to improve, that shift will likely accelerate; but for now, Google’s dominance remains largely intact.

Unlike traditional search engines that lean on signals such as domain authority, backlinks, and click‑through data, generative AI chatbots craft answers by synthesising information they have already learned. Their sense of what counts as “reliable” or “relevant” comes from consensus signals in their training datasets. These training backbones include public web crawls, licensed datasets, and proprietary training materials.

Depending on their design, generative AI platforms fall into two broad categories:

• Pure language models: Generate free‑form text from internal training data knowledge without live browsing. No citations are provided.
• Retrieval-augmented systems: Run live queries against an index of the web (or other data stores), augment responses with the data retreived, and return answers with citations and links.

Because AI models evaluate passages of text on contextual relevance and content quality, and not merely on old-school authority and SEO metrics, well‑written specialist pages can rise to the top when they best address a user’s question. Nevertheless, there is still substantial overlap with high‑ranking Google results. Authoritative domains feature heavily in the datasets that language models learn from and in the indices that retrieval‑augmented systems search.

Estimates vary across studies and the landscape is constantly evolving. As of Q2 2025, ChatGPT accounts for about 80% of the generative AI search market. However, Google Search still handles roughly 350 times more queries than all generative AI tools combined, making it the primary search destination overall.

It’s important to remember that many people who search on Google have already conducted initial research using generative AI. These users often arrive more informed and are primed to engage with specific brands, unique selling points (USPs), or solutions they’ve discovered earlier in their journey.

AI Overviews are Google’s proprietary generative AI snapshots. They appear at (or near) the top of search resuls pages for most informational queries.

Powered by Google's large language models (LLMs), including Gemini, these overviews synthesise information drawn from multiple sources across the web to produce a cohesive, conversational summary that directly answers the user’s query. The feature was formerly referred to as the Search Generative Experience (SGE) during its experimental rollout phases.

UX and Presentation

Visually, AI Overviews appear as a highlighted box at or near the top of the Search Engine Results Page (SERP). The content may include:

• A short paragraph summarising the answer.
• Bulleted lists or step-by-step instructions.
• Suggested follow-up questions.
• Links to relevant sources.
• Sometimes images videos, or visual cards extracted from web pages.

This experience is designed to help users get quick, multi-faceted answers without needing to click through multiple pages. As a result, they often reduce clickthrough rates.

From a technical perspective, AI Overviews operate as a query rewriting and multi-step reasoning system. When a user searches Google, here's a simplified overview of what happens under the hood:

1. Analyse intent: Google parses the query (breaks it down), identifies entities, and classifies the likely intent (e.g. informational, commercial, navigational). This process uses a combination of foundational algorithms like RankBrain and Neural Matching (introduced in 2015 and 2018) together with more recent models such as BERT, MUM, and Gemini, which enhance semantic understanding and context awareness.
2. Generate sub-questions: For complex queries, the engine may break the query into sub-queries to retrieve more granular facts.
3. Retrieve relevant web content: Google surfaces high-quality pages, often focusing on authoritative sources, structured data, and entities from its Knowledge Graph.
4. Generate a summary: Using generative AI, Google composes a concise natural-language answer. Unlike traditional featured snippets, AI Overviews can combine details from multiple documents and weave them into a unified narrative, underpinned by multi-step reasoning.
5. Provide source links: Typically, the AI Overview includes a panel of source links (or "citations") adjacent to the summary. However, the presence and format of these links vary depending on confidence scores and content policy constraints.

In short, AI Overviews are an embedded AI chat interface that uses “retrieval-augmented generation” (RAG) to deliver tailored, up-to-date, and verifiable answers designed to address the user’s precise intent, while remaining closely aligned with the traditional organic results for the same query. It can issue hundreds of searches behind-the-scenes, reason across disparate pieces of information, and create an expert-level, fully-cited report.

A 2025 study by Ahrefs reported an average 34.5% drop in clickthrough rates for the #1 organic result when AI Overviews were shown. This means fewer people click on individual organic website links. The actual links in AI Overviews, including the citations, also do not get clocked as frequently as traditional organic results. Despite this, appearing in AI Overviews still helps brands gain visibility while users are reseaching a topic.

No, brands should not ignore Google Search. Google is still crucial for brand visibility and high-intent searches. However, brands also need to optimise for generative AI platforms to stay visible where users are increasingly searching. Fortunately, there is a huge overlap between GEO and traditional SEO.

Brands should think about GEO the way they think about mobile SEO, voice search, or SERP feature optimisation... it's simply an additional consideration within an already multi-faceted seamless search strategy.