Skip to content
Back to Blog
Insights

Generative Engine Optimisation in Malaysia: How to Actually Do It, Step by Step

By Bernard Leong

Co-Founder, Geo One

24 August 2026
Generative Engine Optimisation in Malaysia: How to Actually Do It, Step by Step

Generative engine optimisation in Malaysia comes down to seven jobs, done in order: confirm AI crawlers can reach your site, fix your business identity so it matches everywhere, add structured data, rewrite key pages into question-and-answer shape, earn mentions on sites you don't own, measure what AI platforms actually show, then repeat monthly. Based on the audits we conduct, a significant number of the Malaysian SMEs we see skip step one and undermine the steps that follow — though exact figures vary by industry and audit sample. This guide is the working sequence Geo One runs for clients in Kuala Lumpur (if you'd like to know more about our approach and credentials, visit our services page), written so you can do it yourself if you'd rather not pay anyone.

What is generative engine optimisation, in one paragraph?

GEO is the practice of making your business easy for ChatGPT, Claude, Gemini and Perplexity to find, understand and recommend. Search engine optimisation fights for a blue link on a results page. GEO fights to be the sentence inside the answer. The difference matters because an AI assistant doesn't show ten options. It usually names two or three businesses, and if yours isn't one of them, the customer never learns you exist. If you want the longer explanation of how answer engines pick who to quote, we've covered that in our guide to answer engine optimisation.

Step 1: Check that AI crawlers can even reach your website

This is the step that quietly undermines everything else — getting this right is the foundation every subsequent step is built on. AI platforms send their own bots (GPTBot, ClaudeBot, PerplexityBot and others) to read your pages. If your hosting or CDN blocks them, you are invisible no matter how good your content is.

Cloudflare has reportedly split its old single AI toggle into three separate controls covering Search crawlers, Agent crawlers and Training crawlers — though this should be verified against Cloudflare's official documentation, as settings and policies may have changed. Check Cloudflare's documentation directly for the latest confirmed details. Plenty of Malaysian SMEs sit on Cloudflare's free tier because their web developer set it up years ago and nobody has touched it since.

What to do: open yoursite.com/robots.txt in a browser and look for Disallow rules naming GPTBot, ClaudeBot, PerplexityBot or Google-Extended. Then log into Cloudflare and check the AI crawler controls against their current published settings. If you don't have the login, ask whoever built the site. Completing this step is essential, as the steps that follow rely on AI crawlers being able to reach your content in the first place.

Step 2: Make your business identity identical everywhere

AI models build a picture of your business from every mention they can find. If your SSM-registered name, trading name, address, phone number and service description differ across your website, Google Business Profile, Facebook page, LinkedIn, industry directories and old press mentions, the model gets low confidence and names a competitor instead.

Pick one canonical version of these five things: legal name plus trading name, full address, one phone number, one email, and a one-sentence description of what you do and who you serve. Then go and make every profile match, word for word.

To illustrate why this matters: consider a freight forwarder in Port Klang listed as "logistics services" on one profile and "customs brokerage and haulage" on another. To an AI model, that reads as two half-businesses rather than one clear entity. Described consistently — with a single, specific service description used across every profile — the same company becomes far easier for a model to understand and recommend. If you have a case study or client example that demonstrates this outcome, it is well worth publishing on your site as a third-party-citable reference.

Step 3: Add schema so machines read the facts, not the design

Structured data is code in the background of your page that states facts plainly: this is an Organization, this is its address, these are its services, this is a FAQ and here are the answers. AI systems lean on it because it removes guesswork.

The practical minimum for a Malaysian SME is Organization or LocalBusiness schema on your homepage, Service schema on each service page, and FAQPage schema wherever you answer real customer questions. If you're on WordPress, a schema plugin gets you most of the way. If your site is custom-built, your developer can add JSON-LD in an afternoon. Test it with Google's Rich Results Test before you move on.

Step 4: Rewrite your key pages into answer shape

AI assistants quote passages, not pages. A page written as one long block of marketing copy gives them nothing clean to lift.

Restructure your top five pages like this. Use a real customer question as the H2, for example "How much does a company audit cost in Malaysia?" Then answer it in the first two sentences, in 40 to 60 words, with a specific number, timeframe or condition. Add the detail underneath for the human reader. Do this for pricing, turnaround time, coverage area, what's included, who it's for and how to start.

Be specific. "Competitive rates" is unquotable. "Between RM1,200 and RM3,500 depending on transaction volume" gets quoted. Ranges are fine. Vagueness isn't.

Step 5: Earn mentions on websites you don't own

This is where most DIY attempts stall, and it's the part that moves the needle hardest. Models weigh third-party corroboration heavily. Your own site says you're good. A roundup article, a directory listing, a supplier's case study and a trade association member page saying the same thing is what turns a claim into a fact.

Start with the ones you can control this month: industry association memberships, chamber of commerce directories, supplier and partner pages that list you as an authorised dealer, local business directories with real editorial standards, and any journalist or blogger already writing roundups in your category. Ask satisfied clients for reviews that describe what you actually did, not just "good service". A review that says "handled our SST filing and EA forms for 38 staff" carries specific, extractable information. This is the slowest part of GEO and the one that compounds. Our authority and citation building work exists precisely because owner-operators rarely have the hours for it.

Step 6: Measure it, or you're guessing

Two measurements, both free.

First, Google Search Console has introduced dedicated Generative AI Performance Reports. Confirm current availability and full details via Google Search Console Help or the official Google Search Central blog, as features and rollout timelines may vary by account and region. These reports provide an isolated view of impressions inside AI Overviews, AI Mode and Discover's generative features, filterable by page, country and device. Click data is still absent, so treat it as a visibility signal rather than a traffic report. Many Malaysian businesses have not yet opened this report. Open yours today.

Second, test the prompts your buyers actually type. Write down ten of them in the language your customers use, including Bahasa Malaysia and Mandarin if that's your market. Run them across ChatGPT, Gemini, Claude and Perplexity, log which businesses get named, and repeat monthly. That log is your scoreboard. If you'd rather not do it by hand, our free AI visibility scanner runs the same check across all four platforms and shows you where you currently stand.

Step 7: Repeat monthly and cover your languages

GEO isn't a project you finish. Models refresh, competitors publish, and answers shift. Set a monthly cycle: one new question-shaped page, one new third-party mention chased, one crawler and schema check, one prompt test.

One Malaysia-specific point worth acting on. If your customers switch between English, Bahasa Malaysia and Mandarin in the same WhatsApp conversation, your content should exist in more than one language. Very few local businesses publish answer-shaped content in Bahasa Malaysia, which makes those queries far easier to win than the English ones.

How long does this take and what does it cost?

Doing it yourself costs time, roughly 8 to 15 hours to set up properly and 3 to 4 hours a month after that. Expect the first movement in AI answers around 60 to 90 days, sometimes longer in crowded categories like property and F&B. Nobody can promise you a specific position in an AI answer, because these systems are probabilistic and change without notice. What you can control is whether you're readable, consistent, corroborated and measured. Businesses that get those four right get named far more often than those that don't.

If you'd rather have it done, Geo One offers tailored GEO solutions for Malaysian SMEs and solo professionals, designed to be accessible well below where traditional agency retainers start. See what's included on our services page and pricing page, or run the free scan first and decide afterwards. Either way, go and check your robots.txt today. That one takes five minutes.

Bernard Leong

Co-Founder, Geo One

Nearly 20 years across energy, capital strategy and applied AI, including large-scale operational data at BP. Founded SkillsMe and Cryptrain.

More about the team

Check Your AI Visibility

See how AI platforms currently view your business with a free scan.

Free AI Scan