How to Get Recommended by ChatGPT

How ChatGPT Actually Works

ChatGPT is a large language model trained on a massive corpus of web pages, books, and structured data up to a training cutoff. When it answers a question, it predicts the most likely helpful response based on patterns in that training data. For evergreen questions ("what is an accounting firm that handles GST in Singapore?"), it leans on what it already learned.

For questions needing current information, ChatGPT can browse the web using Bing. It issues a search, reads the top results, and synthesises an answer grounded in those live pages. This means two paths lead to a recommendation: being part of its training data (strong, repeated presence on well-crawled sites) and being findable via Bing at the moment a user asks.

Why Bing Is ChatGPT’s Hidden Gatekeeper

Here is the part most Singapore SMEs miss: when ChatGPT searches the web, it does not use Google — it uses Bing. OpenAI’s partnership with Microsoft means ChatGPT’s live browsing is powered by Bing’s index. Studies show that roughly 87% of ChatGPT’s citations match Bing’s top results for the same query.

The implication is stark: if your business is not in Bing’s index, ChatGPT cannot cite you — no matter how good your content is, no matter how strong your Google rankings are. Bing is the gatekeeper behind ChatGPT’s browsing mode, and almost nobody in Singapore has invested in Bing visibility. That asymmetry — everyone optimizes for Google, almost nobody optimizes for Bing — is the cheapest arbitrage in GEO right now.

Bing’s reach extends beyond ChatGPT. Microsoft Copilot is literally built on Bing — same index, same gate. Many smaller AI search products also license Bing’s index rather than crawling the web themselves. Perplexity is the notable exception, running its own crawler. But for ChatGPT and Copilot combined, Bing’s index is the single most important non-Google data source in the AI search ecosystem.

What ChatGPT Prioritises When Recommending

ChatGPT does not rank businesses the way Google ranks pages. It recommends entities it can describe with confidence. Confidence comes from consistency: the same business name, service, and location appearing across multiple trusted sources.

• A clear entity definition — who you are, what you do, where you operate — stated the same way everywhere.
• Mentions in "best of" lists, comparison articles, and round-ups on authoritative sites.
• A Wikipedia or Wikidata entry (where notable) that anchors your entity.
• Structured data (schema.org) that lets crawlers parse your services, area served, and reviews.
• Direct-answer content on your own site — pages that plainly state "we do X for Y in Singapore".

The Optimization Logic for ChatGPT

Optimizing for ChatGPT is fundamentally about entity clarity and citation-worthy presence. You are not chasing a keyword rank; you are making it easy for the model to confidently name you.

Start by making your own site machine-readable: consistent NAP (name, address, phone), schema markup for LocalBusiness and Service, and a /llms.txt file that summarises what you do for AI crawlers. Then build external corroboration — get listed in directories, earn mentions in articles and listicles, and ensure your business appears the same way everywhere.

Common Mistakes Singapore SMEs Make

Many businesses assume a pretty website is enough. ChatGPT cannot "see" your design — it reads text and structure. Sites built with images for headings, no schema, and inconsistent business details across the web are effectively invisible to the model, no matter how good they look to a human.

• Using different business names or phone numbers across platforms (kills entity matching).
• No schema markup — the model cannot reliably parse your services or location.
• Zero presence on third-party sites — no listicles, no directories, no PR.
• Service pages written as marketing fluff instead of clear, direct answers.

Prioritized Action Plan

Sorted by impact and effort — start with the quick wins at the top.

Frequently Asked Questions