International SEO · 12 September 2026 · 9 min
International SEO and hreflang: rank a market, do not clone a site
International SEO ranks a market you can fulfil. Hreflang, country targeting and localised research, without cloning the site you already have.
International SEO is the decision to rank in a market you can actually serve, with pages written for that market's language, competitors and proof. It is not a plugin that emits hreflang, and it is not a translated homepage with the currency symbol swapped. Google can exchange-rate a price. It cannot invent a local operation, a local vocabulary, or a reason for that country's buyer to prefer you over the firm already there.
A market is a fulfilment problem before it is a template
Before anyone draws a site tree, answer the unglamorous questions. Can you contract, invoice, support and deliver in that country, in the language you plan to publish, on the timelines a local buyer expects? If the website will say yes and the office will improvise, do not build the market. A ranking you cannot fulfil becomes a refund, a review and a branded query you will not enjoy.
Choose the market by demand you can see and by rivals you can name, not by a slide that lists continents. Search Console, and a paid account if you have one, will show which countries already send impressions. A handful of queries researched properly will show whether a commercial vocabulary exists. Paid and organic sharing one map helps: the queries people pay for in that country teach faster than a keyword list translated at head office.
Country targeting is this choice made explicit. You are not targeting Europe. You are targeting Germany in German, the UAE in English, or Canada in English with French decided later. Write down country, language, URL pattern, and the person who may approve copy. Anything vaguer becomes a clone with extra folders.
International SEO fails when the site is a mirror
The failure mode is familiar. A strong UK site is duplicated under `/us` and `/au`. Titles are swapped, a few spellings change, and hreflang is added so the duplication feels technical. The pages compete, or Google folds them together and shows the original. You have paid for templates and earned no new relevance.
A market page needs something find-and-replace cannot fake. Local proof: clients, regulations, units, seasons, and the objections sales hears on calls there. Local internal links, so a US guide does not dump the reader on a UK service URL with a VAT price. If you will not publish a price, publish the pricing logic, the currency you invoice in, and what is out of scope. Sameness is the bug. Hreflang is not a patch for it.
English is the trap because it looks portable. A solicitor and an attorney are not the same query. Holiday and vacation are not the same intent. A city page that was a fair service-area page in one country is often a doorway in another. Read the local results before you brief. If page one is directories, marketplaces and a national brand with years of mentions, say so, and decide whether that term is an SEO job this year.
Hreflang is a hint you earn by having distinct URLs
Hreflang names which URL is the alternate for which language and region. It does not confer rankings, and it does not rescue identical copy. Use it only when the alternates exist and a buyer would notice the difference. Each URL references itself and its alternates. The annotations reciprocate. Codes are real: `en-GB`, not `en-UK`. `x-default` points at a selector or a genuinely international page, not at the market finance prefers.
Put the annotations in the HTML or in a sitemap you maintain. A sitemap of URLs you later redirected is worse than silence. Check return tags when a URL is added or retired. A one-way annotation is easy for Google to ignore, and it is the usual source of reports that say hreflang was implemented and nothing changed.
IP redirects are the other self-inflicted wound. Sending every American address to `/us`, including a crawler, hides the rest of the site. If you must nudge people, use a banner they can dismiss, and keep every market URL directly crawlable. A country selector with real links beats a script that decides the visitor is in the wrong place. Auto-translate widgets that rewrite the page are not a market either.
Localized keyword research is not a dictionary pass
Start from the live results and from the language sales already uses. Take the head terms, the problems and the comparisons. Have someone who sells in that market reject the nonsense. A translation tool will offer a correct phrase nobody searches, and it will miss the colloquial term that carries the intent. Group queries by page. Do not force a row-by-row twin of the home site. Some home pages should not exist in the new market. Some new pages have no parent.
Be consistent inside the market: spelling, currency, tax language, units. Mixed conventions on one URL tell the reader, and the engine, that nobody local edited it. Build fewer pages than the original site until those pages deserve links. A shallow complete clone loses to a deep partial catalogue. The content programme should be briefed from the local map, with local examples, and edited by someone who can hear a wrong register. A machine draft is only a start when a person who knows the market rewrites the claims.
Architecture, links and the measurement that keeps you honest
For most firms adding a second or third market, a subfolder on the existing domain is the clean default: one crawl, equity that can flow, and hreflang that stays honest. A country-code domain earns its cost when the market expects a local company, when legal or hosting rules push you there, or when you already have a site with its own links. A subdomain is rarely the best of either choice. Pick one pattern and stop inventing a new one each year.
Give each market its own Search Console view so a win in one country cannot hide a stall in another. Track impressions, rankings and enquiries by market. Do not average them into a global vanity line. Technical health is often shared: a slow template hurts every country. Fix it once, in the WordPress build or whichever stack you run, rather than adding a plugin per locale.
Earn links in the market whose page should move. A UK trade title pointing at the UK URL is not a plan for the Australian service page. Editorial link building in a new country means a new list, a local story, and anchors a local editor would write. Cross-link markets only where the reader is choosing a region. Keep the firm one organisation and the offers local, which is entity SEO. Where an overview appears, give it a market page worth citing, the same discipline as Google AI Overviews citations, written in that market rather than translated after the fact.
What we will not pretend hreflang can do
We will not annotate duplicate pages and call it international. We will not auto-redirect crawlers, launch ten languages because a menu made it look easy, or build city doorways to fake country targeting. We will not promise that a new folder ranks on a date. A new market is closer to a new site that shares a domain: indexation, internal links, proof and mentions, judged on that market's results.
Confirm you can fulfil. Research the language of demand. Design the URL pattern. Write or seriously adapt the priority URLs. Add hreflang once the alternates are real. Earn a few relevant mentions. Read that market's Search Console before you open the next country. Teams that reverse the order spend a quarter debugging tags on pages that should not exist.
The firms that get this right look narrower than their competitors' flags, and they rank where they chose to. SEO Jackers Global plans that choice with the ranking work, not after the translations have shipped. If you are opening a market rather than cloning one, brief it through classic SEO and the order form.