As AI reshapes how consumers discover brands, South African marketers face a new challenge: being visible in a world where Google is no longer the first stop.
Sam Davis, senior director of solutions engineering at Yext, unpacks the rise of generative engine optimisation (Image supplied)
Customer search is changing, with nearly half of consumers now using AI tools for brand research. In this shifting digital landscape, what does it mean to win at discovery?
Sam Davis, senior director of solutions engineering at Yext, a New York-based international digital location management software company, unpacks the rise of generative engine optimisation (GEO), in a conversation with South African agency Sauce Advertising.
Search is being redefined
Davis says that over the past year, the most striking changes being observed are the fragmentation of the search journey.
There is a shift from keywords to conversations, and from a “Google only” mindset to a world where discovery happens across multiple AI-led applications - ChatGPT, Perplexity, Gemini - and even social platforms like TikTok.
This has big implications for businesses. People’s behaviours are changing - not only how they search, with more complex, conversational queries, but also where they search. '
Brands that fail to adapt to these shifts risk becoming invisible in the very places their audiences are now spending their discovery time.
SA can’t ignore the global shift
The fragmentation of the search journey is being driven by global shifts in technology and consumer behaviour, says Davis.
He highlights that AI-led discovery tools, conversational search, and social-first search habits are influencing users everywhere - South Africa included.
"The pace of change may differ depending on external factors such as technology penetration, adoption, and cultural preferences, but the direction of travel is the same.
"Younger demographics in particular are already bypassing traditional search entirely, favouring platforms like TikTok, Instagram, or ChatGPT as their first step.
He adds that forward-thinking brands are even starting to explore model context protocol (MCP) strategies powered by a knowledge graph,- initially for enriching and streamlining internal knowledge GPT-enabled applications.
"But with an additional eye on future-proofing for how public large language model (LLM) end-points could evolve to ensure brands can surface the most accurate, up-to-date brand-verified content to LLM users."
For brands in South Africa, adopting this kind of approach early could be a powerful way to future-proof discoverability in the AI era.
Rewriting the rules of discoverability
He says that one of the biggest mistakes he is seeing is that marketers are still treating AI search as if it behaves exactly like traditional search.
"They focus purely on optimising for Google’s blue links, when in reality AI-led discovery surfaces content in completely different ways - summarised answers, conversational flows, and even responses that combine text, images, and other rich content.
"If content is not structured for machines to understand and contextually reuse, it's effectively invisible in those channels," says Davis.
There is a marked behavioural shift - from single keyword searches, to short-tail queries like “restaurant near me”, to long, highly specific conversational requests that AI can remember and personalise over time, notes Davis.
"AI search can now handle complex, conversational requests such as booking a restaurant near a hotel while factoring in prior knowledge like dietary preferences, location, and family context."
For businesses, this changes the game.
Without accurate, rich, structured data—menus, dietary attributes, opening hours, proximity, reviews, FAQs, etc.— businesses simply won’t appear in how LLMs present summarised, structured content that aligns with these in-memory preferences.
And when it comes to measurement, marketers need to move beyond rankings and clicks.
Phil Manners,
Sauce Advertising 24 Mar 2025 Staying ethical
Every time a new search paradigm emerges, some will try to manipulate it. In the early days of SEO, it was keyword stuffing; in local search, it was fake reviews and location spam.
The real danger is that generative results carry an implicit sense of authority, stresses Davis.
"If a brand appears in an AI-generated summary, the user assumes it’s there because it’s the best or most relevant option.
"That’s why content relevancy and E-E-A-T principles - experience, expertise, authoritativeness, and trustworthiness - are more important than ever.
"This is especially critical for industries like healthcare, finance, and legal, where inaccurate or misleading answers can cause real harm.
The golden rules for staying on the right side of ethics in generative results are:
- Be relevant: create content that directly and accurately answers the questions your audience is asking, in the context they’re asking them.
- Be truthful: don’t exaggerate capabilities or withhold important details to get picked up in an answer.
- Be specific: break down exactly what your brand stands for, the products or services you provide, and who they’re for. Avoid vague positioning in favour of factual, verifiable detail.
- Show your credentials: where applicable, cite experts, reference reputable sources, and demonstrate real-world experience to strengthen your trust signals.
- Manipulating AI search can backfire faster than traditional SEO: LLMs are continuously retrained, and platforms are actively building guardrails to detect low-quality or misleading input.
The next chapter of GEO
Davis remarks that by 2026, the GEO landscape will look very different.
"The rate of LLM development and maturity is increasing rapidly - compare the performance leap between ChatGPT-4 and ChatGPT-5.
"Search will be increasingly conversational, context-aware, and hyper-local, with results shaped by a mix of AI-generated summaries, structured data feeds, and in-memory personalisation."
For brands, the mandate is clear, he says: stay AI-ready.
"That means ensuring your listings are complete, accurate, and consistent across every surface; maintaining a strong reputation management programme; publishing highly relevant, E-E-A-T-driven content with structured markup; and keeping a regular cadence of local social posting to feed engagement signals."
Forward-thinking brands will also lay the groundwork for MCP (model context protocol) strategies, preparing to connect their trusted, structured data directly into AI systems should LLM platforms open their APIs to brand-registered MCP servers.
"That will give them a secure, first-party route into AI assistants, ensuring their information is authoritative, current, and contextually relevant at the moment of discovery," he says.
He adds that by 2026, GEO success won’t just be about being found - it’ll be about being trusted, contextually accurate, and AI-ready in a way that makes you the obvious choice when an AI answers a query. The brands that start now will own that space.
If there’s one takeaway Davis explains, it’s this: the search journey is fragmenting, AI is accelerating, and consumer behaviour is shifting faster than most brands can adapt.
"The winners in 2026 will be those who prepare now - with trusted data, relevant content, and an AI-ready strategy that ensures they’re found, trusted, and chosen in the moments that matter."
Continuing the GEO conversation
Yext is a key partner of Sauce Advertising, and together they are unpacking the GEO narrative as headline sponsors for the IMC later this month. where Bryony Rose, director of Yext Enterprise International Business, will deliver a keynote on search fragmentation, the rise of AI, and the hyperlocal advantage.
