The Short Version
Plenty of Malaysians search in Bahasa Malaysia or Chinese, but a translated website is not automatically a better website. Extra language versions help when your buyers genuinely search in those languages and you can keep the pages as good as the English ones.
If you go ahead, put each language in its own subfolder, connect the versions with hreflang, research keywords separately for each language, and have a fluent person review every page. Bulk machine translation with no review is the version I would not publish, especially after Google's September 2026 spam update.
Multilingual SEO comes up in almost every discovery call I have with retailers, clinics and property firms in Penang, Johor Bahru and the Klang Valley. The question is usually some version of "should we have a BM version, and maybe Chinese as well?" Often a half finished plugin translation already sits on a subdomain, untouched for years.
My own Search Console is a fair example of why the question is reasonable. Over the last three months this site picked up impressions for Malay queries such as "servis seo malaysia" (129) and "perkhidmatan seo" (70), and for Chinese ones such as "seo顾问" (85) and "马来西亚seo服务" (45), even though every page here is in English. People do search this way. The harder question is whether a separate language version would serve them better than a good English page, and what it takes to build one that Google treats as a real page rather than filler.
💬Search Behaviour
How Malaysians Search Across Languages
Malaysian search behaviour does not split neatly into three groups. A lot of commercial searching happens in English, even among people who speak Malay or a Chinese dialect at home. Then there's the mixed query: "kedai aircond near me", "klinik gigi Bangsar", "renovation rumah teres". Google handles these mixed queries reasonably well, and an English page with the right local terms in it can rank for them.
Where a separate language version starts to matter is when the whole journey happens in that language. Someone searching "harga pasang solar rumah" is likely to want to read the answer in Malay. Someone searching in Chinese for a tuition centre in Cheras probably wants a Chinese page and a reply in Chinese on WeChat or by phone.
| Language | Typical query style | Where I see it matter most |
|---|
| English | Short, commercial, often with a city name | B2B, professional services, most of the Klang Valley |
| Bahasa Malaysia | Price and how to questions ("harga", "cara", "kedai") | Home services, automotive, government related services, east coast and northern states |
| Chinese (Simplified) | Service plus area, often very specific | Education, property, health, retail in Penang, Johor Bahru, Klang and Cheras |
| Mixed | Malay or English words combined with a place name | Local services everywhere; usually fine to target from an English page |
One more trap that shows up in Malaysian data: Indonesian terms. My own reports include queries like "jasa seo", which is the Indonesian phrasing. A Malaysian searcher would more often write "perkhidmatan seo" or "servis seo". If you translate your pages with a generic tool, you can easily end up with Indonesian vocabulary that reads slightly off to a Malaysian buyer and targets searches from the wrong country.
⚖Decision
Is a Second or Third Language Worth It for You?
I'd rather talk a client out of a translation project than watch them publish 60 pages nobody maintains. Each language version is a second website in practice. It needs its own keyword research, its own updates when prices or services change, and someone who can answer enquiries in that language. Here is how I usually frame the decision.
| Situation | My usual advice |
|---|
| Your sales team already handles enquiries in BM or Chinese every week | Worth building, starting with your top service or product pages |
| Search Console shows steady impressions for queries in another language | Worth testing with a few strong pages before translating everything |
| B2B buyers are procurement or management teams | Usually English is enough; spend the budget on better English content |
| Nobody on the team can review or reply in the language | Not yet; a page that leads to an unanswered enquiry does more harm than good |
| You want more pages mainly to "look bigger" to Google | Don't. That is the pattern spam policies are written for |
If you sell to Malaysian consumers in specific communities, a language version can be the most useful thing you build all year. If you sell enterprise software to finance directors, it is probably the least. The B2B SEO guide covers what usually matters more for that second group.
🌐Setup
Site Structure and Hreflang
Once you decide to build, the technical setup is not complicated, but it has to be consistent. Google does not guess language versions from a dropdown or a cookie. Each version needs its own URL that Googlebot can crawl without clicking anything.
Pick one URL structure and stick to it
| Option | Example | Trade off |
|---|
| Subfolders | example.com.my/ms/ and example.com.my/zh/ | Easiest to manage; shares the authority of your main domain. My default for SMEs |
| Subdomains | ms.example.com.my | Works, but often treated as a separate project and forgotten |
| Separate domains | example.my for BM | Only makes sense for a separate brand; doubles the link building work |
| Parameters or cookies | ?lang=ms, or a switcher with no URL change | Avoid. Google may never see the other versions |
Connect the versions with hreflang
Hreflang tells Google which pages are language versions of each other, so it can show the BM page to someone searching in Malay and the English page to someone searching in English. For a Malaysian site the common codes are en (or en-MY), ms (or ms-MY) and zh-Hans for Simplified Chinese. Add an x-default entry for the version you want shown when nothing else fits, usually English.
Three rules cause most of the problems I find in audits. Every version must list every other version, including itself. The links must be reciprocal, so if the English page points to the BM page, the BM page has to point back. And every URL in the set must return a 200 status and be the canonical version, not a redirect or a noindexed page. Google's own guide to telling Google about localized versions covers the three ways to add it: HTML tags in the head, HTTP headers, or your sitemap.
For sites with more than a few dozen pages, I prefer putting hreflang in the XML sitemap because it keeps the page templates clean and is easier to check in one place. If you go that route, my notes on XML sitemaps explain how Google reads them.
One detail people find surprising: Google says it does not use hreflang or the HTML lang attribute to work out what language a page is in. It detects that from the text itself. So a page with a lang="ms" tag and English body copy is still an English page to Google. Translate the visible content, including titles, headings, image alt text and buttons.
💡
Hreflang is a hint, not a ranking boost
Hreflang helps Google pick the right version for each searcher. It does not make a weak translated page rank. If the BM page is thin, it will lose to a competitor's better BM page whether your tags are perfect or not.
⚠Quality
Translation Quality and Google's Spam Policies
This is the part I'd stress most right now. Google rolled out its September 2026 spam update from 24 September, and its Search Status Dashboard showed it complete on 8 October. Google doesn't publish a list of what each spam update targets, so I won't pretend to know exactly what it changed. What I can point to is the written policy.
Google's spam policies describe scaled content abuse as producing many pages mainly to manipulate rankings, and they list automated transformations such as translating as one of the ways it happens when little value is added for users. A plugin that pushes 300 English pages through machine translation overnight, with nobody reading the output, fits that description uncomfortably well.
That does not mean machine translation is banned. Many good translated sites start with a machine draft. The difference is what happens next.
- A fluent reviewer reads every page. Ideally someone who sells to that audience, not only someone who speaks the language.
- Examples are adapted, not just translated. A Chinese page for Penang buyers may need different examples, areas and payment methods than the English page.
- Only pages worth reading get translated. Your top service pages, key product categories and the FAQ people actually ask. Not every blog post from 2019.
- Contact details match the language. If the page is in Chinese, say who will reply and how.
This also matters for AI search. In my testing, assistants such as ChatGPT usually reply in the language of the question, so a well written BM or Chinese page gives them something relevant to cite. I wrote about what tends to get cited in how to rank better in ChatGPT.
🔎Keywords
Keyword Research Per Language
The most common mistake is translating the English keywords. Search terms don't translate one to one. "Aircond service" in English might be "servis aircond" in Malay searches, not the textbook "perkhidmatan penyaman udara". Chinese searchers often use shorter terms plus an area name.
- Start from your own data. Filter Search Console's Queries report to Malaysia and scan for Malay and Chinese queries you already appear for. My Search Console guide shows where to find them.
- Check volumes in each language separately. Set the language and location properly in your keyword tool. I explain the settings for Malaysia in the Google keyword tool guide.
- Ask your sales or front desk team. The words customers use on the phone and in chat messages are often better keywords than anything a tool suggests.
- Look at who ranks. If the top results for a Malay query are all English pages, a BM page may not be needed for that term. If they are all in Malay, you need one to compete.
This is the same process I follow in keyword research projects, just repeated per language.
☑Checklist
Launch Checklist and What to Do This Week
Adding a language version is closer to a site migration than a content task, because it changes your URL structure and templates. Treat it with the same care as the steps in my website redesign SEO checklist.
| Check | What good looks like |
|---|
| URL structure | Every language in its own subfolder, crawlable without JavaScript or cookies |
| Hreflang | Reciprocal, self referencing, includes x-default, only points to 200 status canonical URLs |
| Canonicals | Each language page canonicalises to itself, never to the English version |
| Titles and descriptions | Written in the page's language, using that language's keywords |
| Language switcher | Plain links to the equivalent page, not back to the homepage |
| No forced redirects | Don't redirect visitors by IP or browser language; let them choose |
| Review | A fluent person has read every published page |
The canonical row matters more than it looks. If your BM pages point their canonical to the English versions, you are telling Google they are duplicates, and they will mostly disappear from results. It's one of the first things I check in a technical SEO review.
What to do this week
- Export the last three months of queries from Search Console, filtered to Malaysia, and count how many are in Malay or Chinese.
- Ask your team which language your best enquiries arrive in.
- If there's an old plugin translation on your site, check whether those pages are indexed and whether a fluent person would be happy to put their name on them. If not, noindex or remove them before building new ones.
💡
Start small and measure
Ten well reviewed pages in BM or Chinese, tracked for three to six months, will tell you more than 300 machine translated pages ever will. If they earn impressions and enquiries, expand. If they don't, you've lost very little.
Multilingual pages matter a great deal for shops with walk in customers, which is why I cover language versions of outlet pages in my guide to retail SEO in Malaysia.
Where to Start
Multilingual SEO is worth doing when your customers search and buy in more than one language, and when you can keep each version as good as your English pages. Look at your own Search Console data and your enquiries first. They will tell you whether BM or Chinese demand is real for your business or only a nice idea.
If the answer is yes, the setup is a technical job: subfolders, correct hreflang and canonicals, and clean sitemaps. If you'd like a second pair of eyes on an existing translated section, that is part of the technical SEO work I do, and the content side is covered under content SEO.
Frequently Asked Questions
Do I need a Bahasa Malaysia version of my website for SEO?
Not always. If your customers search and buy in English, a BM version may add little. It is worth building when your enquiries or Search Console data show real demand in Malay, and when someone on your team can review the pages and reply to Malay enquiries. Start with your most important service or product pages.
Is machine translation bad for SEO?
Machine translation on its own is not the problem. Publishing large numbers of machine translated pages with no human review can fall under Google's scaled content abuse policy, because little value is added. Use a machine draft if it helps, then have a fluent person edit every page and adapt the examples for Malaysian readers.
Which hreflang codes should a Malaysian website use?
Most Malaysian sites use en or en-MY for English, ms or ms-MY for Bahasa Malaysia, and zh-Hans for Simplified Chinese, plus an x-default entry for the fallback version. Every page in the set should reference all the others and itself, and each URL must be a live, canonical page.
Should I use subfolders or subdomains for language versions?
For most SMEs I recommend subfolders such as /ms/ and /zh/. They are simpler to manage, share the authority of the main domain, and are less likely to be forgotten. Subdomains can work, but they are often treated as separate projects. Avoid switching language with cookies or parameters alone.
Can an English page rank for Malay search terms?
Yes, especially for mixed queries that combine Malay or English words with a place name. Google can match an English page to these searches if it uses the right local terms. For queries written fully in Malay where the top results are Malay pages, you will usually need a proper BM page to compete.
How long does it take for new language pages to rank?
It varies with competition and how strong your site already is. New language pages on an established site often start getting impressions within a few weeks, while meaningful traffic usually takes three to six months. Check the Pages report in Search Console to confirm Google has indexed each version.
