What Is Generative Search?
Generative search is the application of large language models to the task of answering search queries. Instead of the traditional model of matching a query to indexed documents and ranking them by relevance, a generative search system retrieves candidate content from its index and then uses an AI model to synthesise a composed, readable answer. The result appears at the top of the search interface, often with inline citations linking back to the source pages.
Google introduced this approach with the Search Generative Experience in 2023 and then rolled out AI Overviews to all users globally in 2024. Perplexity AI operates as a standalone generative search engine. Microsoft Copilot in Bing uses a similar approach. Each of these products is an implementation of generative search, and together they represent a significant change in how users interact with search results.
The distinction from traditional search is important for digital marketing strategy. In traditional search, ranking well means appearing as a clickable result that users choose to visit. In generative search, the AI system decides whether to cite your content within its composed answer. High organic rankings help but do not guarantee citation. What matters is whether the content on a page directly, specifically, and credibly answers the user's question in a format the AI can extract and use.
Generative engine optimisation is the practice of adapting content so that generative search systems will cite it. This includes structuring content with clear headings, using answer-first paragraph formats, including verifiable facts and named entities, and adding structured data markup. South African businesses competing in sectors where generative search is increasingly common need this approach built into their content strategy from the start.
Generative Search In Practice
The scenario below is an illustrative example, not a Juicy Designs client result. The figures indicate the scale of effect that generative search optimisation work typically produces, so treat them as indicative rather than measured.
Picture a Cape Town accountancy firm targeting the query "how does SARS provisional tax work". Not long ago this query would have shown ten blue links in Google. Today it is likely to trigger an AI Overview that explains provisional tax in four short paragraphs, with citations to SARS.gov.za and two accountancy resources. The firm's own explainer article could easily be missing from those citations, despite ranking at around position four organically.
The firm's marketing consultant would then analyse what makes the cited sources different. SARS typically uses a numbered list format with exact amounts and deadlines. A cited accountancy resource would usually open with an answer-first sentence ("Provisional tax is a method SARS uses to collect income tax in advance from taxpayers whose income is not subject to employees' tax.") followed by specific steps. The firm's article, by contrast, might use flowing prose, bury the definition in paragraph three, and include no specific figures or dates.
If the firm rewrites the article using the same structure, answer-first opening, numbered steps, specific rand thresholds, named SARS forms, and an FAQ section with schema markup, it could plausibly begin appearing as a citation source in the AI Overview for multiple provisional tax queries once Google reindexes the page. It might also pick up citation appearances on Perplexity for related long-tail queries, generating brand visibility among users who may never click organic results.
What generative search is
Generative search is search that uses generative AI to produce answers, generating a synthesised response to a query (often drawing on multiple sources) rather than only returning a list of links, so that the search experience includes AI-generated summaries and answers. It describes the broad shift in search towards AI-generated answers, encompassing features like Google's AI Overviews and AI Mode, and AI-powered answer engines, where, instead of (or alongside) the traditional blue-link results, the search surface presents an AI-composed answer that synthesises information, usually with citations to sources. Generative search represents a significant change in how search works and how people get information: rather than clicking through several results to piece together an answer, users can get a synthesised, comprehensive answer directly, generated by AI from the underlying sources. For businesses, generative search matters because it changes how their content is surfaced and how users interact with search results: being one of the sources an AI-generated answer draws on and cites becomes a form of visibility, and the AI answer may satisfy some queries directly (potentially reducing clicks) while still citing and linking sources for users who want more. Generative search is closely related to, and overlaps with, terms like generative engine optimisation (optimising for these AI-generated results), AI Overviews, AI summaries and answer engines, all part of the AI-driven, answer-generating evolution of search. Understanding generative search matters because it is the broad, significant shift towards AI-generated answers in search, which is reshaping how people find information and how businesses gain visibility, so knowing what generative search is, search that generates AI answers from sources, helps a business understand this evolution and that being a clear, trustworthy, cited source is the way to remain visible as search becomes increasingly generative.
Visibility and traffic in generative search
Two important questions about generative search are how it affects website traffic and how to be visible in it, and both come back to being a clear, trustworthy, cited source, since generative search generates answers from sources and links to them. On visibility: appearing as a source in generative search rests on the same foundations as broader AI visibility, being indexed and accessible to the systems generating answers, providing clear, specific, well-structured, fact-based answers to the questions people ask, offering accurate and verifiable information (since generative systems favour and cite trustworthy sources), and being a well-regarded, authoritative entity, with Google confirming no special files or AI-specific schema are needed. Content that clearly and comprehensively answers the relevant questions, with quotable, self-contained statements and concise answers, is most readily drawn on and cited, so the way to be visible in generative search is to be genuinely excellent, clear, authoritative content on the questions in your field. On traffic: generative search does change the traffic dynamic, because an AI-generated answer can satisfy some queries directly (a user getting their answer from the AI summary without clicking through), which can reduce clicks for simple queries fully answered by the generated answer, a real shift that businesses should be aware of. However, generative search does not eliminate website traffic: the AI answers cite and link sources, so users who want more detail, to verify, or to act (buy, contact, explore) still click through, and many queries, especially those where users want depth, options, or to transact, continue to drive clicks, while being a cited source provides both potential traffic and brand visibility. So generative search tends to reduce clicks for simple, fully-answered informational queries while still driving traffic for queries where users need more than the summary, and being a prominently cited source captures the clicks that do occur plus brand exposure. The practical response is to focus on being a cited source (through excellent, clear, authoritative content) so as to capture the visibility and clicks that generative search still provides, and to recognise that content offering genuine depth, options or transactional value, beyond what a summary conveys, continues to attract clicks. For a South African business, the guidance is to optimise for generative search by being a clear, trustworthy, well-indexed source that thoroughly answers its audience's questions with fact-based, quotable content, so it is cited in AI-generated answers, while understanding that generative search may reduce clicks for simple queries but still drives traffic where users need more, and that being a cited source secures both visibility and the clicks that remain. Because generative search is reshaping how information is found and traffic flows, adapting by being genuinely excellent, citable content, and offering value beyond what a summary provides, is the sound way to remain visible and continue attracting traffic as search becomes increasingly generative.
FAQ
Does generative search reduce website traffic from Google?
Studies show that queries answered by AI Overviews receive fewer clicks on organic results below the generated answer. This is especially true for informational queries where the AI answer fully resolves the user's question. However, commercial and transactional queries, where users need to take action or compare options, often still produce strong click-through rates. South African businesses should audit which of their target queries now trigger AI Overviews and adjust their content strategy accordingly.
What content format works best for appearing in generative search results?
Generative search systems favour content that is structured, specific, and citable. Use clear headings, short paragraphs, and direct answers at the start of each section. Include verifiable facts such as statistics, specific process steps, named entities, and dates. Avoid vague generalisations. FAQ schema and DefinedTerm schema also help AI systems identify and extract your content as a reliable answer source.