Selling into a second country looks like a commercial decision and becomes a technical one quickly.
The store works. The products ship. Then search starts behaving oddly. Your UK page outranks your Australian page for Australian shoppers. Google indexes one market and ignores another. Rankings you had already earned drift sideways into the wrong country.
This is what international SEO services exist to resolve, and it is genuinely different work rather than ordinary SEO applied to more countries. Here is what actually changes.
The problem hreflang solves
A single-market site has one job. Rank for the right terms with the right audience. There is only one audience.
Two market versions introduce a question the search engine has to answer. These pages are broadly the same. Which audience gets which one?
Unanswered, search engines guess. Often they show a US page to a UK shopper, or treat two near-identical pages as duplicates and index one while suppressing the other. Hreflang answers that question directly. It says these are alternate versions of the same page for different audiences, which removes the duplication risk and tells Google which page belongs to which country.
It is most valuable on subfolder setups, where every market shares one domain and the versions look most alike to a crawler.
How the annotation is built
Every market version carries a set of references listing every other version, and the whole set appears identically on each page. An Australian store selling into the US, UK, Canada, Germany and Sweden would have a block naming all six, repeated on all six pages.
Three things about that set matter more than they appear to.
Self-reference. Each page must include itself in its own set, not only the other markets. A page that names every version except its own is one of the most common broken setups, and it tends to be ignored entirely rather than partially honoured.
The fallback. One entry, marked as the default, nominates which version to serve anyone who does not match a listed market. It usually points at the primary market rather than a generic landing page, and choosing which market carries it is a real decision rather than a formality.
Language before region. Each entry names a language first and a country second, and the two are not always the same letters. Swedish content for Sweden is the Swedish language code paired with the Swedish country code, which do not match. Assuming they do is a frequent and silently ineffective mistake.
Tip: Each market page should treat itself as the canonical version, not defer to the primary market. A market page that points its canonical at another country removes itself from search in its own, and nothing on the page looks wrong while it happens.
Choosing a structure
Hreflang needs somewhere to point. There are three options and the choice is long-lived.
| Structure | Example | Strengths | Trade-offs |
|---|---|---|---|
| Country domains | example.com.au | Strongest geographic signal, clean separation | Authority builds separately per market |
| Subfolders | example.com/uk/ | Shares domain authority, simplest to run | Weaker inherent geographic signal, needs hreflang |
| Subdomains | uk.example.com | Clear separation | Authority sharing unreliable, complexity without full benefit |
Subfolders suit most stores, because a new market inherits authority the main domain has already built. Country domains suit large markets or categories where local presence genuinely affects buyer trust. Mixed setups exist and work, with a primary country domain carrying x-default and other markets living as subfolders.
Choosing deliberately matters, because changing later means a migration and migrations put existing rankings at risk.
Localisation is not translation
The language a market speaks is the starting point, not the answer.
Content should be genuinely native rather than translated. That means British English for the UK, US English for the US, German for Germany. But it goes further than spelling. Vocabulary diverges in ways that decide whether a page ranks at all.
The same product is chips in the UK and fries in the US. The same service is pregnancy massage in Australia and prenatal massage in the US. Translate the Australian page for the American market and you will rank for a term nobody there searches, while a competitor using the local word takes the traffic.
This is where keyword research per market earns its place. Not translating a keyword list, but researching each market on its own terms and finding what people there actually type.
The same logic applies to language selection itself. A German market can be served in English, and sometimes brands do it because it is easier. But German keywords are what German buyers search, in far higher volume, and moving from English targeting to native German targeting can be the difference between a market that stalls and one that grows.
Tip: Two English-speaking markets still need separate versions. Pricing, shipping, availability and vocabulary all differ even when the language does not.
The commercial side of a market version
A market version that ranks and cannot convert has solved half the problem.
Currency, payment methods and delivery information all need to match local expectation, since preferred payment methods in particular vary considerably between countries. Authority also builds per market, which makes country-specific directories, listings and local links part of the work rather than an afterthought.
Detecting cannibalisation across markets
With several markets in the same language, the versions compete unless keyword territory is assigned deliberately. This is the hardest version of the problem and it is not visible from looking at the pages, because each one looks fine.
It shows up in the data instead. Search Console filtered by country reveals when a core keyword ranks in a market it was not built for. Analytics shows traffic arriving from the wrong geography. Rank tracking across markets surfaces the same term being contested by multiple pages, and sometimes by multiple country versions of the same page.
That pattern, one keyword and several competing pages across markets, is the signal. Once it is visible, the fix is territory rather than technology. Each market version needs its own defined set of terms to own, informed by what people in that market genuinely search.
Launching a new market
Before building anything, the market needs verifying.
That means real search demand for the category in that country, an honest look at who already ranks there and how entrenched they are, and an understanding of what performs locally rather than assuming the home market translates. Vocabulary research belongs here too, since the terms that carry volume may not be the ones you expect.
Tracking should be decided at the same time rather than retrofitted. Separate Analytics and Search Console properties per market give clean per-market reporting, which matters because a blended figure hides exactly the market-level problems you need to see.
The sequence that tends to work is technical foundation first, then native language content and local keyword targeting on top of it. Building content before the structure is sound produces pages that cannot perform, and diagnosing that afterwards is considerably harder than setting it up correctly.
Where the difficulty is
The concepts are learnable. Executing them across a live store selling into several countries is where it becomes specialist work, because every decision has to hold in every market at once and mistakes surface as gradual drift rather than obvious breakage. That is the work I do. If you are expanding into new markets, or running several already and watching them compete, my international SEO services page explains how I approach it. To see where your current setup stands, request an SEO audit.
If you sell online, the ecommerce SEO guide covers the store foundations international work sits on top of.
Frequently asked questions
What does hreflang actually do?
Hreflang tells search engines that pages are alternate versions of each other for different languages or regions. It prevents market versions being treated as duplicate content and helps the right version surface for the right country. It does not boost rankings on its own.
Does every page need to reference itself in hreflang?
Yes. Each market version must include itself in its own annotation block alongside the others. Missing self-references are one of the most common reasons hreflang is ignored.
What is x-default for?
The x-default value nominates which version to serve users who do not match any listed market. It usually points at the primary market version rather than a generic page.
Should the canonical point to the main market?
No. Each market version should canonicalise to itself. Pointing a market page at another market’s URL removes it from search in its own country, and the effect is easy to miss because nothing appears broken.
Do two English-speaking markets need separate pages?
Usually yes. Currency, shipping, availability and vocabulary differ between English-speaking countries even though the language does not. A single English page serving every English market tends to underperform in all of them.
How do I know if my country sites are competing with each other?
Search Console filtered by country shows core keywords ranking in markets they were not built for, and rank tracking reveals the same term contested by multiple pages. The fix is assigning distinct keyword territory per market rather than a technical change alone.
Aashish Maharjan is an SEO specialist focused on ecommerce and international organic growth across global markets. Learn more about my work or request an SEO audit.