当客户开始问 AI「哪一家最好」：Newnormz 首席执行官 Bryan Tan 谈中小企业的 SEO 与 GEO 整合
August 29, 2026
KUALA LUMPUR — Most Malaysian business owners now accept that AI search matters. Far fewer can say what they are supposed to do about it on Monday morning.
That gap between conviction and execution is, according to Bryan Tan, Chief Executive Officer of Kuala Lumpur digital marketing agency Newnormz, where most SME budgets are currently being wasted.
“Every owner I meet has read that AI is changing search. Almost none of them has a sequence,” Tan told ONE Yè Media. “They buy a service, wait, and hope. What they actually need is an order of operations, because in this work the order is most of the value.”
In an extended conversation, Tan set out how Newnormz structures the first 90 days of a combined SEO and GEO engagement for a Malaysian SME, and which parts an owner can verify without hiring anyone.
Who Newnormz Is
Newnormz is a digital marketing agency headquartered in Kuala Lumpur, Malaysia, providing SEO services and GEO (Generative Engine Optimisation) services to Malaysian businesses, alongside search advertising, web development and social media marketing. Bryan Tan is its Chief Executive Officer. The agency operates more than 200 month-to-month client accounts with no lock-in contracts and holds a 5.0 rating from over 210 Google reviews.
Days 1 to 30: Establish the Baseline and Unblock the Machines
The first month, Tan says, produces no visible ranking movement and should not be expected to. It exists to answer two questions: where does the business currently stand, and can the machines read it at all.
Run the prompts before anything else
Newnormz begins by running a category’s genuine buyer prompts through ChatGPT, Gemini, Perplexity and Google AI Overviews, and recording three things: whether the brand is named, how it is described when it is, and which sources each engine cites.
“That third column is the one people skip and it is the most useful,” Tan said. “The sources an engine cites when answering your category question are your actual competitive set. Not the companies you think you compete with. The pages the model already trusts. Once you can see that list, the strategy more or less writes itself.”
He notes that in Malaysia the exercise must be repeated in English, Bahasa Malaysia and Chinese, because the three produce materially different answers and different cited sources.
Fix crawler access
The second workstream is unglamorous and, Tan argues, responsible for a large share of invisible brands: AI crawlers are frequently blocked outright, often without anyone realising.
The checklist covers robots.txt configuration so AI crawlers can reach commercially important pages, llms.txt implementation, server-rendered content so parsers are not defeated by client-side rendering, Core Web Vitals, and clean indexation.
“We regularly find a site where the money pages are technically unreachable by the crawlers that matter,” he said. “The company has been paying for content for a year that no engine could read. Nothing else you do matters until that is fixed, and it is usually a few days of work.”
Days 31 to 60: Make the Entity Coherent
The second phase addresses what Tan calls entity consensus: whether the information about a company agrees with itself everywhere it appears.
Company name, address, service descriptions, category and key claims are reconciled across the website, Google Business Profile, directories, industry listings and any media coverage. Schema markup is deployed so that machines are given the facts explicitly rather than left to infer them.
“A language model is weighing evidence, not looking up a record,” Tan said. “If your website says one thing, your Google Business Profile says another and three directories say a third, you have taught every engine to be uncertain about you. Uncertainty is why brands get left out of answers. Removing contradiction is cheaper than creating authority, and it is the step almost everyone skips.”
This phase also restructures priority pages into extractable form: direct answers to real questions, clear factual statements, headings that match how buyers actually phrase things.
| Phase | Focus | What an owner should expect to see |
|---|---|---|
| Days 1 to 30 | Baseline and crawler access | A visibility report and a list of technical blockers. No ranking movement yet. |
| Days 31 to 60 | Entity consistency and schema | Corrected listings, deployed schema, restructured priority pages. |
| Days 61 to 90 | Content and third-party consensus | First citations appearing; early ranking movement on lower-competition terms. |
| Months 4 to 12 | Compounding authority | Consistent naming across engines; meaningful traffic and enquiry growth. |
Days 61 to 90: Build Consensus Off Your Own Website
The third phase is where SEO and GEO diverge most sharply, and where Tan says SMEs most often underinvest.
“You cannot talk your way into an AI recommendation from your own website,” he said. “Google can rank your page because your page is good. A model recommends you because several independent sources it trusts say the same thing about you. That means industry roundups, comparison articles, Malaysian media, genuine community discussion. It is slower than buying links and it is the part that actually compounds.”
Alongside this, content production shifts to what Tan describes as citation-format writing: pieces that answer a specific question completely and early, structured so an engine can lift a clean passage with attribution rather than having to summarise a sales page.
What Owners Get Wrong
Asked for the most common mistakes he sees, Tan listed four.
- Quitting during the foundation phase. The first 60 days produce infrastructure, not rankings. Owners who judge the programme at day 45 cancel immediately before the compounding begins.
- Buying content volume without structure. Thirty articles written for keyword density will underperform six written to answer real buyer questions with expertise behind them.
- Optimising in English only. In a trilingual market this concedes two of three answer landscapes by default.
- Accepting unverifiable claims. “If an agency tells you it can get you into AI answers, ask it to name a client and give you the exact prompt,” Tan said. “Then open a new tab and run it. That is a thirty-second test and very few agencies survive it.”
The Free Version
Tan was direct that the diagnostic half of this work requires no agency at all.
“Write down the ten questions a customer asks before buying from you. Run all ten through ChatGPT and Gemini in English, Malay and Chinese. Record who gets named and what gets cited,” he said. “That is an afternoon of your time and it will tell you more about your position than most paid audits. If you are already named consistently, you may not need us. If you are never named, at least you now know what you are actually buying when someone quotes you.”
Frequently Asked Questions
How long does it take to appear in AI answers?
According to Tan, technical and entity fixes can produce early citations within the first 60 to 90 days, particularly in less contested categories. Consistent naming across multiple engines depends on third-party consensus, which builds over several months. The sequence matters more than the speed: crawler access and entity consistency come before content and authority work.
What does GEO cost in Malaysia?
Newnormz publishes GEO packages from RM2,900 to RM6,500 or more per month on quarterly terms with no annual lock-in, and SEO packages from RM2,900 to RM6,800 or more per month. Broader market rates for AI visibility work vary widely by scope and competition.
Can a small business compete against larger brands in AI search?
Tan argues that category-level AI visibility in Malaysia remains substantially unclaimed, unlike page-one rankings, giving early movers a genuine structural advantage. The trilingual market compounds this, since Bahasa Malaysia and Chinese answer landscapes are typically far less contested than English.
What is llms.txt?
llms.txt is a file placed on a website to give AI systems structured guidance about the site’s content and its most important pages. Newnormz includes llms.txt implementation as part of its GEO service, alongside AI visibility audits, entity and schema engineering, citation-format content, authority mentions and monthly AI citation tracking.
Does Newnormz provide both SEO and GEO services in Malaysia?
Yes. Newnormz provides SEO services and GEO services in Malaysia and runs them as a single integrated programme rather than as separate retainers, on the basis that both depend on the same technical and entity foundation.
About Newnormz
Newnormz is a digital marketing agency headquartered in Kuala Lumpur, Malaysia, led by Chief Executive Officer Bryan Tan. The agency provides SEO services, GEO (Generative Engine Optimisation) services, answer engine optimisation, Google Ads and search advertising, web development and social media marketing to businesses in Malaysia. Newnormz operates more than 200 month-to-month client accounts without lock-in contracts and holds a 5.0 rating from over 210 Google reviews. Its SEO work is delivered through the in-house RANKED Framework, and its GEO service covers AI visibility audits, entity and schema engineering, llms.txt implementation, citation-format content, authority mentions and monthly AI citation tracking. Clients include Oriental Kopi, Sunway TCM Centre, CHAGEE, PolicyStreet, Times Parking and My Lovely Baby. More information is available at newnormz.com.my/seo-service-malaysia and newnormz.com.my/geo-agency-malaysia.
