Somewhere in a meeting this quarter, someone senior looked at a map, pointed at a second country, and said the words “let’s just do SEO over there too.” Everyone nodded. Nobody costed it. And now it is on your desk.
Here is the part they left out of the meeting: global SEO is not one project repeated per country, it is a different project every time. Different competitors, different search behaviour, sometimes a different search engine entirely. It is slow, it is detailed, and done badly it can quietly drag down the rankings you already have at home.
Done well, though, it puts you in front of buyers in markets your competitors have not figured out yet. With 73.8% of the world’s population online as of April 2026, according to DataReportal, the customers are there. This guide covers how international SEO actually works, the decision that matters most, and the mistakes that cost the most, written for the marketing manager who has to make it happen rather than the one who volunteered you for it.
What is global SEO?
Global SEO is the practice of optimising a website to rank in search results across multiple countries or languages. It is also called international SEO, and the two terms mean the same thing. The goal is to serve the right version of your site to the right person in the right market, so a buyer in Kuala Lumpur and a buyer in Singapore each land on the page built for them.
It is not translation, and it is not “add a language switcher and hope.” Real international SEO covers four moving parts: how you structure your domains, whether you target by country or by language, how you tell search engines which page belongs to which market (the hreflang tag), and how deeply you localise the content and the links behind it.
One important correction to the 2018 version of this advice: Google is no longer the only door you optimise for. In China Baidu leads the search market, at roughly 64% in late 2025 according to The Egg. In Russia Yandex holds around 72%, per SerpSculpt’s 2026 data. In South Korea, Naver trades the lead with Google depending on who is measuring. If any of those markets are on your map, “SEO” means something other than “Google.”
Should your business actually do international SEO?
You should consider international SEO when you have real demand, or real intent to build it, outside your home market, and the resources to do the job properly in each one. If you are a Singapore dentist or a neighbourhood cafe, the honest answer is no, and we would rather tell you that now than take the retainer.
International SEO is usually a logical next step if one of these is true:
- You already sell in more than one country and your site does not reflect it.
- You have a genuinely global or regional product, common for B2B, SaaS, and content businesses.
- Your home market is too small to hold your growth ambitions.
- You are seeing meaningful organic traffic from other countries already, which is the market telling you where to go next.
That last signal is the most trustworthy, and most people never check it. Open Google Search Console, filter Performance by country, and look at where impressions are coming from. If a country is already sending you clicks with zero localised content, that is validated demand, not a hunch. Start there.
The Southeast Asia picture makes the case concrete. A Singapore FMCG or F&B brand eyeing Malaysia, Indonesia, or the Philippines is not chasing a theory. It is following buyers who already cross borders to find products. The question is rarely “is there demand,” it is “are we prepared to do five markets properly instead of one market badly.”
If the honest answer is “not yet,” there is still one thing worth doing today: reserve your country domains before someone else does. More on that below.
How do you choose your international site structure?
You have three options for structuring an international site: country-code top-level domains (ccTLDs), subdirectories on one domain, or subdomains. This is the first and most consequential decision you will make, and it is one of the hardest to reverse later, so it deserves more than five minutes.
Here is the awkward truth: the decision is often made before you are even in the room. A company that bought example.com.my a decade ago has already narrowed its own options, because a country domain cannot cleanly host other countries beneath it. If that is you, we will get to your escape routes.
The three structures, compared on the things that actually matter:
| Structure | Example | Geo-targeting signal | Authority | Cost & maintenance |
|---|---|---|---|---|
| ccTLD | cloudrock.sg, cloudrock.com.my | Strongest. The domain itself signals the country. | Does not stack. Every domain starts from zero. | Highest. Separate registration, hosting, and SEO per country. |
| Subdirectory | example.com/sg/, example.com/my/ | Moderate. Set by hreflang and on-page signals. | Stacks. All authority builds on one domain. | Lowest. One domain, one server, shared strength. |
| Subdomain | sg.example.com, my.example.com | Weak. Treated closer to a separate site. | Does not consolidate as well as subdirectories. | Moderate. Easier than a new ccTLD, weaker payoff. |
When should you use a ccTLD?
Use ccTLDs when a distinct local domain is genuinely required by law, brand architecture, or local trust, not because .sg simply feels more local. They send the clearest possible country signal to search engines, and each one can fail, get penalised, or get hacked without dragging the others down with it.
The cost is authority. Each ccTLD is a separate site in Google’s eyes, with its own crawl budget, its own backlink profile, and its own reputation to build from scratch. A link to your .sg domain does nothing for your .my domain. For a business entering a new market against established local competitors, starting from zero is a real disadvantage. Some registries also gatekeep: a .cn domain, for instance, requires proof of a registered local company.
When should you use subdirectories?
Subdirectories are the recommended default for most businesses, because they consolidate all your authority onto a single domain. Build links to example.com and that strength flows down to example.com/sg/ and example.com/my/ alike. The longer you run it, the stronger the whole structure becomes.
This is not just our preference, it is Google’s current position too. In Google Search Central’s own guidance on managing multi-regional sites, subdirectories on a generic domain sit alongside ccTLDs as a fully supported option, and Google’s John Mueller spent much of late 2025 openly questioning whether most businesses need ccTLDs at all. The trade-off is a shared point of failure: one penalty or one server outage can affect every market. That risk is manageable with good hosting and clean practice, which is precisely where an inexperienced SEO team gets you into trouble.
When should you use subdomains?
Use a subdomain when infrastructure or IT constraints genuinely force separate hosts, and you accept a weaker SEO position in return. Subdomains give you neither the geo-targeting strength of a ccTLD nor the authority consolidation of a subdirectory, which is why we rarely recommend them for international targeting. You take on much of the hassle of separate sites without the clearest local signal.
What if you are already stuck on a country domain?
If your main site already lives on a ccTLD like example.com.my, you have two sensible paths. Commit to the ccTLD structure and run each market as its own domain, or keep the .my for your home market and register a separate generic domain (.com) as the foundation for everything international. Trying to bolt other countries under a country domain, as in example.com.my/sg, is technically possible and practically miserable. It is swimming upstream for no reward.
Should you target by country or by language?
Target by country when your offer, pricing, or content differs meaningfully between markets. Target by language when the same content serves speakers of one language across several countries. Most of our clients target by country, because a product sold in Singapore and Malaysia usually needs different pricing, examples, and sometimes different terms, even when both pages are in English.
Language targeting makes sense when the differences are small. If you have strong traffic from Spanish speakers spread evenly across Spain, Argentina, and Costa Rica, and your offer is identical for all of them, one Spanish-language version is far more efficient than three near-identical country sites. You would use a subdirectory such as example.com/es/ for the language, without pinning it to a single country.
Country targeting is the norm for regional expansion. Each country gets its own pages and its own localised content, which may or may not be in the same language. A Singapore-to-Southeast-Asia rollout almost always wants country targeting, because the whole point is that Jakarta is not Singapore. You can combine both approaches at once, targeting Malay speakers in Malaysia while also running a distinct Indonesian-market site, but do not reach for that complexity until the simpler structure genuinely cannot carry the job.
How do you implement hreflang tags correctly?
The hreflang tag tells search engines the language and region a given page is written for, so the right version is served to the right searcher and duplicate-content problems are avoided. It has become almost synonymous with international SEO, and it is also where most sites quietly break.
The syntax looks simple. A page that exists in three variants declares all of them:
<link rel="alternate" href="https://example.com/" hreflang="x-default" />
<link rel="alternate" href="https://example.com/sg/" hreflang="en-sg" />
<link rel="alternate" href="https://example.com/my/" hreflang="en-my" />Three rules decide whether it works. First, the language code must follow ISO 639-1 and the optional region code must follow ISO 3166-1 Alpha-2, in the format language-region, so en-sg not en-singapore. Second, every hreflang cluster must be reciprocal: each URL has to list every other URL, including itself. If your Singapore page points to your Malaysia page but the Malaysia page does not point back, Google ignores the entire set. Third, add an x-default tag for a fallback page for users who match no other version, which the 2018 rulebook did not require but Google now treats as best practice.
The tedium is real and it scales badly. With three markets, each page carries four hreflang lines. With a hundred country versions, each page needs a hundred tags pointing at each other. This is why Digital Applied reports that around 75% of hreflang implementations contain errors, usually a missing return tag or a broken URL. One error breaks the cluster silently, so nothing warns you. It just quietly stops working.
Two things that trip people up on modern builds. Hreflang belongs in the server-rendered HTML, not injected by JavaScript after load, because Google may not run your JavaScript when reading language signals. And each page should canonicalise to itself, never to another country’s version. A cross-country canonical tells Google to drop one page in favour of the other, which deindexes exactly the page you wanted ranking. Canonical URLs, hreflang URLs, and schema URLs must all be absolute, not root-relative, per the spec.
How do you localise content for each market?
Localise the whole content experience for each market, not just the words, and never rely on raw machine translation. Localisation adapts examples, currency, idioms, imagery, and terminology to a market. Translation only swaps words for words, and Google’s quality systems are now good at spotting the difference.
Keyword research has to be redone per market, because terminology genuinely diverges. In India, “share market” is more common than “stock market.” Across Malaysia and Singapore the same product can carry different everyday names, and even English-to-English shifts matter, as anyone who has argued about “localise” versus “localize” knows. Never assume your home-market keywords translate directly, research each country from scratch.
If the target market’s main language is not English, invest in native, human translation for your priority pages, then use AI translation tools for high-volume content with a native speaker reviewing the output. Unedited machine translation is a genuine ranking risk. Google’s own documentation cites a site that was penalised for publishing machine-translated content without human review. It is also a branding risk: KFC’s “finger-lickin’ good” once landed in Chinese as something close to “eat your fingers off,” and Parker Pen’s Mexican campaign promised, roughly, that its ink would not get you pregnant. Cheap translation reads as a cheap brand.
What else does international SEO require?
Beyond structure and hreflang, three practical points decide whether the whole thing holds together.
Do you need to localise on-page elements and links too?
Yes. Every on-page element has to be localised per market: title tags, meta descriptions, image alt text, heading tags, internal anchor text, and URLs. Optimise country by country or Googlebot gets confused about which page serves which audience. If you run one blog across every market, as we do, each post has to be optimised for its specific country. The work scales quickly, which is the honest cost nobody mentions in the “let’s just do it over there too” meeting.
Link building localises as well. The method does not change, but the targets do: you need links from sites and publications in each target market, and the outreach itself may need a native speaker. This is where a subdirectory structure earns its keep. Point most of your link building at the root domain and the authority flows down to every country subfolder, instead of forcing you to build reputation from zero on five separate ccTLDs.
Does the location of your hosting still matter?
No, not for rankings. Server location is no longer a meaningful ranking signal, so use a global CDN for speed rather than hosting separately in every country. Google reads geo-targeting from hreflang, on-page signals, and domain type, not from your server’s IP address. A content delivery network such as Cloudflare serves every market fast from the nearest edge node, which also keeps the subdirectory structure viable, since insisting on per-country hosting is one of the things that would rule subdirectories out.
Should you reserve your local domains now?
Yes, and this is the one action to take even if you are years away from launching. Reserve the country domains for every market you might plausibly enter, purely to protect your brand. You do not need to build on them. You need to stop a competitor, or a squatter, from owning yourbrand.my before you do. Some registries take time and paperwork, so this is not a task for the week before launch.
What has changed since the old international SEO playbooks?
The biggest shift since 2018 is that ranking on Google is no longer the whole game, you now also have to be quotable by AI answer engines. Google AI Overviews, ChatGPT, Gemini, Perplexity, and Claude all summarise sources directly, and increasingly that summary is the answer a buyer sees before they click anything.
For international content, three things follow. Structure each page so it can be extracted: a clear question as the heading, then a direct answer in the first sentence or two, before the supporting detail. Keep entity clarity consistent across every market version, so the same company, product, and standard names appear the same way everywhere, which is what lets an engine understand you are one business operating in several countries. And keep it fresh, because staleness is now actively penalised. Google’s own guidance is to state locale variations explicitly rather than hoping the crawler infers them from a single location.
One housekeeping note from Google’s side: the old International Targeting setting in Search Console was retired in 2022, so country targeting for subdirectories and subdomains now rests on hreflang and on-page signals rather than a toggle. If your process still assumes that toggle exists, your process is out of date.
Ready to expand into new markets?
Global SEO is not a walk in the park, and it is not a decision to make on the strength of one meeting and a map. But for the right business, in the right markets, it puts you in front of demand your competitors have not organised themselves to serve yet.
The order of operations matters. Structure first, because it is the hardest thing to undo. Hreflang second, because it is the easiest thing to break. Localisation and links third, because that is where the ranking is actually won. Get those three right and the rest is patience.
If you are weighing up an international move and want a straight answer on whether it is worth it, and which markets to start with, talk to us before you commit the budget. We do paid discovery before we quote, which means we will tell you if the honest answer is “not yet.” The other agencies will quote you tonight, from a template, without asking a single question about your Search Console data. That is rather the point.
Frequently asked questions about global SEO
What is the difference between global SEO and international SEO?
There is no difference, the two terms are used interchangeably. Both describe optimising a website to rank across multiple countries or languages. Some markets favour one term over the other, but they refer to the same practice of serving the right localised version of your site to the right audience.
Is a ccTLD or a subdirectory better for international SEO?
Subdirectories are the better default for most businesses, because they consolidate SEO authority onto a single domain and are cheaper to maintain. A ccTLD sends a stronger country signal but forces you to build authority from zero in each market. Choose a ccTLD only when a distinct local domain is genuinely required by law, brand structure, or local trust, not simply because it looks more local.
What happens if my hreflang tags are wrong?
A single error causes Google to ignore the entire hreflang cluster, so the wrong-country page can rank in your target market. The most common faults are missing return tags, where one page points to another that does not point back, and broken or non-absolute URLs. Around 75% of hreflang implementations contain at least one such error, and because it fails silently, you often will not know until rankings slip.
Does global SEO still only mean optimising for Google?
No. Google leads most markets, but not all of them. Baidu leads in China, Yandex leads in Russia, and Naver competes closely with Google in South Korea. If any of those markets are in your plan, you have to optimise for the local engine, each of which has its own algorithm and ranking behaviour.
How long does international SEO take to show results?
Expect several months per market before meaningful movement, and longer in competitive markets where you are building authority from a low base. The timeline depends heavily on your chosen structure: subdirectories inherit existing domain authority and move faster, while a fresh ccTLD starts from zero. Anyone promising fast international rankings is either misinformed or selling you a twelve-month retainer.
