Harnessing AI for Effective Marketing and Branding – SME Centre@SCCCI x SMU Academy

On 20 August 2026, GeoVector co-founder and CEO Razvan Iordache spoke at Harnessing AI for Effective Marketing and Branding, a half-day workshop run by SME Centre@SCCCI and SMU Academy at Singapore Management University. His session was called From Google to ChatGPT: How SMEs Get Discovered When Customers Ask AI, and the room was mostly Singapore SME owners and marketing leads who had come with one version of the same question: what happens to my business when customers stop searching and start asking?

AI has moved into the buyer's funnel

For two decades, being found online meant ranking on Google. That assumption is now breaking. The sequence Razvan drew on screen was deliberately mundane: a buyer has a question, opens an AI assistant, gets back a shortlist of names, visits only the sites on that shortlist, and buys. The website visit still happens. It just happens after the decision has narrowed, and only for the brands that made the list.

Three numbers set the scale of that shift:

• 900M+ weekly active ChatGPT users (OpenAI, February 2026; Reuters reported 1B monthly app users in June 2026)
• 58.5% of Google searches end without a click to any website (SparkToro, 2025)
• 94% of B2B buyers use LLMs somewhere in their buying decision (6sense, 2025)

So the competition is no longer for a position on a results page. It is for a seat on a shortlist that the customer never sees you missing from. Nobody sends you an email to say ChatGPT recommended three of your competitors and not you. Most Singapore businesses have no way to check.

Which AI engines Singapore customers actually use

A fair question from the floor was whether any of this is real in Singapore yet, or whether it is a US story arriving late. The local numbers are lopsided in a way that makes the answer simple. ChatGPT is not merely present here, it is dominant, and the two Google AI surfaces ride on a search engine that almost everyone in the country already uses.

Chatbot shares are StatCounter's Singapore figures for July 2026; the search share is StatCounter Singapore for 2026. The practical reading: if you only have the appetite to care about one AI assistant, care about ChatGPT. But because AI Overviews and AI Mode are bolted onto the search engine Singaporeans were already using, they reach people who never chose an AI product at all.

Even on Google, the click is vanishing

The point that got the strongest reaction was that this is not only ChatGPT pulling traffic away from Google. It is happening inside Google. Someone searches as always, an AI Overview writes the answer at the top so there is no need to open a site, and a follow-up question turns the page into AI Mode, a chat almost nobody leaves. The search still happens. The visit does not.

• ~58% of US Google searches end with no click to any website (SparkToro / Datos)
• 15% → 8%: a result's click rate roughly halves when an AI Overview appears above it (Pew Research)
• ~93% zero-click rate inside Google's AI Mode (Bain & Company)

Those three figures are the ones presented on the day, attributed to their original publishers. Read together, they describe a search page that has quietly become an answer page. Razvan's line for it: being left out of the answer is the new version of being on page two of Google, except there is no page two to scroll to. We wrote about what that means for local sites in the zero-click reality.

SEO and GEO are not the same job

Generative Engine Optimisation, or GEO, is the practice of getting a brand mentioned, recommended and cited inside AI-generated answers, as opposed to ranking a page in a list of links. Both disciplines are about being found, and the tactics overlap, but the thing you are trying to win is different — which changes how a small marketing budget should be spent.

SEO ranks pages. GEO gets you recommended.

What makes an AI recommend a business

This was the part the room had come for, and the answer was less technical than most expected. Razvan grouped it into three levers, none of which need an engineering team.

• Be clear about what you do. Consistent content about your offer and who you serve, so a model can understand you at a glance rather than infer you from fragments.
• Fix your website's content and structure. Content and technical markup organised so LLMs can read and retrieve it, not just render it for a human.
• Be present where AI looks. Show up in the places these engines actually pull their answers from, which is usually not your own homepage.

For SMEs without an agency, the first steps are smaller than most people expect.

Being mentioned is only half the story

An AI assistant does not simply list you. It frames you — as the budget pick, the premium choice, the safe bet, or the risky one. Two businesses can both appear in the same answer and be sold very differently:

• Positive: “a reliable, well-reviewed choice”
• Neutral: “another option in the space”
• Negative: “less established than rivals”

Neutral is the trap. It does not look like a problem in a mention count, and it does not win anyone over either. The goal is to be mentioned, and mentioned well.

Three things SME owners kept getting wrong

Razvan spent a slide on objections he hears in almost every room of business owners.

• “I rank on Google, so I'm fine.” AI often answers the question outright, without sending anyone to a search results page at all.
• “This is only for big tech brands.” AI recommends small businesses constantly. Clarity and trust beat size, which is the opposite of how domain authority worked in classic SEO.
• “There's nothing I can do about it.” AI visibility can be measured and improved, and the starting moves are basic.

What our own platform data shows

Rather than argue the theory, Razvan showed three findings from brands GeoVector already tracks.

Visibility moves, and it moves a lot. Two businesses were plotted on how often ChatGPT named them across the same six months. One worked at it and climbed. The other did nothing in particular and slipped by roughly half. Same window, opposite results — which is the argument against treating this as a one-off project.

Roughly one mention in two says nothing memorable. For one tracked business, AI almost never described them badly. It described them flatly. About half the time it named them without saying anything that would make a buyer choose them, which is the neutral trap in practice.

AI assistants are already sending real visitors. One business received 103 website visits referred directly from AI assistants over a 90-day window, about 71% of them from ChatGPT alone. Small numbers today. The trajectory is the point.

The live demo: six engines, three stages, and a chat with the data

The session closed with a live walkthrough on two entities everyone in the room already knew, Singtel and the CPF Board, used purely as public examples. Visibility is measured across six engines that customers actually use — ChatGPT, Claude, Gemini, Perplexity, Google AI Overview and Google AI Mode — and split across the buyer journey into Discovery, Research and Decision. That split is the part SME owners tend to find most useful, because a business can look healthy on broad discovery questions and disappear completely at the decision question, which is the one that converts. Seeing which prompts you lose, and which competitor takes the slot, turns an abstract worry into a list of things to fix.

The last few minutes went somewhere the room did not expect: Razvan connected GeoVector to an AI assistant over MCP and simply asked it questions about the visibility data in plain English. No dashboard, no filters. For an SME owner who does not want to learn another analytics tool, that turned out to be the demo people asked about afterwards.

The rest of the programme

Razvan's talk sat in the middle of three. Aaron Toh, Senior Business Advisor at SME Centre@SCCCI, opened with Helping Organisations Adapt to AI and Manage Changes, on why AI adoption so often fails to deliver results and how to build a workforce that can absorb it. David Setiawan, Trainer at Tribe Academy, closed with Harnessing AI for Hyper-Discoverability, a practical playbook of tools and tactics for moving a digital strategy toward AI-driven discovery. The three sessions lined up better than a shared theme usually does: adopt AI internally, then make sure AI can find you externally.

If you remember nothing else

1. Discovery has moved. Customers increasingly start with an AI assistant, not a search bar.
2. Recommendation is not ranking. Being named by an AI is a different game from ranking on Google, with different levers.
3. You can improve it. AI visibility can be measured, and it can be deliberately grown.

A free AI visibility report for every SME in the room

Every SME attending received a complimentary AI visibility report for their own business, plus one month of free access to the GeoVector platform, so they could see how they currently appear in AI answers rather than take our word for it. Setting it up takes four steps — enter your brand and website, confirm the competitors we find automatically, review the customer questions we generate for your industry, and optionally pick an audience to model. GeoVector then runs those questions across all six assistants and builds the report. The same free brand visibility report is open to any business that wants one.

Thank you

Thanks to SME Centre@SCCCI and SMU Academy for organising, and to the owners and marketing leads who stayed behind afterwards with questions about their own brands. The most common one was also the simplest: so what does ChatGPT say about my company right now? Nobody in the room could answer it, which is more or less the whole point.

If you want to know what AI assistants say about your business, and which competitors are taking the shortlist spot, get in touch with our team.