Searching the internet used to mean typing a few keywords into a search engine and getting a page of links to sort through. Companies of every size competed to appear near the top of the page, spending what they could afford to stay visible.
However, consumers today are increasingly bypassing the search box entirely and instead posing their questions to AI assistants like ChatGPT or Perplexity and receiving a single, synthesised answer rather than 10 blue links. According to Euromonitor International, generative AI platforms accounted for just 0.3% of inbound web referrals to retail product pages in September 2024. By November 2025, that figure had reached 3.5%, a tenfold rise in 14 months.
In response, companies are retooling their strategies around what practitioners call generative engine optimisation (GEO), aiming to appear inside AI-generated answers rather than traditional search rankings. What this looks like in practice is often less straightforward. Terrence Ngu, founder and CEO of Singapore-based digital marketing agency Hashmeta, says companies often treat GEO as a quick add-on. “[Most] assume it’s a thing they can buy and deploy quickly. So, the conversation usually starts with: we’ve already got search engine optimisation (SEO) sorted, now we want GEO. That’s the misconception right there.”
In reality, much of what determines visibility has not changed. Ngu’s view, drawn from a decade in search optimisation, is that AI systems tend to surface the same sources that already perform well in conventional search. Credibility carries over. A company with thin content, weak third-party validation and an inconsistent digital presence is unlikely to appear in AI-generated answers simply by commissioning a GEO audit. “No SEO foundation, no AI visibility,” Ngu asserts.
That makes GEO a slower discipline than many expect, with results often taking months to show. “There are no 30-day hockey sticks. Meaningful citation presence by month six, and competitive query presence between nine and 12 months. Anyone promising faster needs to show you the proof,” Ngu tells The Edge Singapore.
GEO is also less uniform than it appears. Different AI platforms have distinct preferences for which sources they trust and cite, so visibility in one does not guarantee presence in another. As more platforms compete for queries that once flowed almost entirely through Google, that divergence is likely to widen.
See also: Meta and EssilorLuxottica bring AI to eye level with smart glasses in Singapore
Even so, many brands continue to focus their efforts on Google AI Overviews as they are familiar with Google search. Ngu cautions against that approach, noting that across his client base, ChatGPT accounts for most AI-driven referral traffic, so having a narrow focus on one platform is a risk. This echoes an earlier mistake, when companies built their digital presence around a single social platform only to be exposed when algorithms changed. “Being strong in Google AI Overview and invisible in ChatGPT is a significant gap. User behaviour is still shifting, so concentration risk in AI search is real,” he warns.
The enemy within
The external platform challenge is only part of the problem. For many organisations, especially in regulated industries like healthcare and finance, compliance is often the binding constraint. As Ngu puts it: “You can have the best content strategy in the world and still spend three months in legal review before a single word goes live.”
See also: IMDA steps up support for local university students entering tech careers
Beyond compliance, much of the expertise within large organisations — from product knowledge to internal research — tends to sit in formats AI cannot easily access. “Large enterprises often have deep expertise locked inside PDFs, white papers and sales decks [which are] invisible to AI. Unlocking and restructuring them for machine extraction at scale is harder than any technical fix,” he explains.
Smaller businesses face a version of this challenge too, even though the dynamics can work in their favour. Instead of citing the biggest brands, AI platforms prioritise the most credible and specific answer to a question. A company that builds genuine depth around a narrow set of queries can, therefore, outperform a global rival spread too thin, Ngu says.
Google’s response
For Google, the shift is both a threat and an opportunity it cannot afford to mishandle. The company still processes more than 5 trillion searches annually and commands close to 90% of the global search market. However, its traditional advertising model built on the back of the click is becoming increasingly strained as users can get answers directly from generative AI platforms or apps without visiting a website.
Rather than cede ground to rival AI platforms, Google has been folding AI-generated answers directly into its search engine. Called AI Overviews, it is now reaching more than 2 billion monthly users across more than 200 countries. It generates synthesised responses powered by Gemini AI directly on the results page.
The same model does double duty on the commercial side. Dan Taylor, the company’s vice president of global ads, says Gemini has solved a monetisation problem that long eluded the business. “Longer and more complex questions were historically much harder to monetise and figure out the relevant ads to show. But today, Gemini helps us deliver the right answer for a consumer by better understanding user intent and matching the right ads,” he says at a media briefing.
Advertisers can pay to place ads above, below or within those AI-generated answers, with placements matched not just to the query but to the context of the response itself. For instance, a travel insurer can surface inside a response about flight cancellation rights, reaching someone at a moment of genuine need rather than vague pre-trip browsing. “Advertisers are realising there are entirely new opportunities for their brand to be discovered that didn’t previously exist,” says Taylor.
To stay ahead of the latest tech trends, click here for DigitalEdge Section
What looks like a new opportunity for advertisers, however, also puts Google in a more delicate position. As more commercial messages are built into AI-generated answers, Google has to work harder to ensure those responses still feel useful, independent and trustworthy.
Taylor pushes back on concerns that advertising could shape the answers themselves. “There is zero opportunity for the advertising business to influence the organic results, and that continues to be true in these new AI-powered experiences.” Whether that assurance holds as the financial stakes rise is something marketers and regulators will be watching closely.
Measuring what matters
Even brands that navigate all of this successfully face a problem that nobody has quite solved yet: how to know whether it is working. The metrics that have governed digital marketing for two decades (such as traffic, clicks, and conversions) are becoming poor proxies for what is actually happening.
What matters now is whether a brand is mentioned in the answer the user gets from AI platforms, says Ngu. When someone asks ChatGPT which company to consider or which product to buy, and a brand appears in the response, that moment carries real commercial weight, even if it is not captured by existing customer analytics.
Consumers who click through from AI answers may be more likely to convert as they arrive with clearer intent. However, part of the path to purchase is increasingly unfolding somewhere the data cannot fully see. Customers who eventually buy through a conventional channel may first have encountered a brand through an AI response that was never recorded in the customer relationship management system. “Chief marketing officers who solve that attribution problem early will be able to make the case for GEO investment far more effectively than those still waiting for traffic to recover,” says Ngu.
Citation authority also requires maintenance. AI systems update and preferences shift, and a brand that is well cited today can be displaced if competitors are more active.
Stopping GEO maintenance is like stopping public relations (PR) activities and expecting your reputation to hold indefinitely. For a while, it does. Then it quietly erodes, and you often don’t notice until the damage is already done.Terrence Ngu, founder and CEO, Hashmeta
For organisations asking where to begin, Ngu’s counsel runs counter to the instincts of most marketing departments. Most corporate content is written for humans who browse. AI platforms, however, favour direct answers to specific questions placed clearly at the top of a page, requiring a rethink of how content is structured.
Additionally, securing placements in publications that AI platforms already trust directly builds citation authority. This makes PR a more strategic investment than many organisations currently treat it.
On content volume, Ngu argues for depth over quantity. “AI systems are increasingly good at identifying depth [so ensure the content you develop is] something that reflects real operational knowledge and answers a specific question better than anything else available.” One genuinely expert piece per month, he argues, will outperform a strategy built on quantity every time.
The harder question is whether organisations are moving quickly enough. “The cost of waiting keeps going up. Enterprise players are building citation authority right now that will become harder to displace over time. Starting small and focused is still starting. And in GEO, starting early is a real advantage,” says Ngu.
