Global SEO

I start with a traffic audit in Google Search Console. I pull the last 12 months of performance by country and filter for pages that already rank in languages I do not actively target. That tells me where demand exists without me having built anything. In one case, I found a client was getting 8,000 clicks a month from France on their English pages. The bounce rate was 78%. That is a signal, not a mistake. I then prioritise markets by a combination of existing search volume, business potential, and operational ease. I do not use a point system. I ask: can I get local content written, can I handle shipping or service, and is the search intent different enough to need a separate page?

Once I have a shortlist of countries and languages, I decide the URL structure. I default to subdirectories with a two-letter country code: example.com/de/, example.com/fr/. The reason is simple. ccTLDs (example.de) are harder to migrate, slower to build authority on, and expensive to maintain. I have seen a site with 15 ccTLDs struggle to get any of them indexed properly because each domain had to be treated as a separate project. Subdirectories let me transfer existing link equity and keep the crawl budget concentrated. But I make an exception for China. There, a ccTLD or a .com.cn is practically mandatory because Baidu treats it as a trust signal. That is a call I had to make after learning about [Baidu SEO](/baidu-seo/) the hard way.

Next, I do keyword research per market. I do not translate my English list. I give a native speaker a brief: what terms do people in this market actually use to search for this product? Then I check those terms in a local tool like Yandex.Wordstat for Russia or Naver's keyword tool for Korea. I have seen a global brand spend six months ranking for a term that nobody in Japan actually uses, because they translated 'car insurance' literally. The Japanese term is entirely different. I also look at the SERP for each keyword. If the results are dominated by local portals or government sites, the intent is probably informational. If it is a marketplace, it is transactional. That changes how I build the page.

Finally, I implement hreflang. But I do it only after I have a clear mapping of every language-country combination. I have seen teams add hreflang tags pointing to pages that do not exist yet, which causes Google to ignore the signal. I use the 'sitemap method' rather than inline tags because it is easier to audit. I also test with the hreflang validator in Search Console. A mistake I made early on: I used 'x-default' on a page that was not actually a default. That confused Google and traffic dropped by 30% for a month. I now only use x-default for a genuinely language-agnostic landing page, like a login portal.

The biggest watch-out is assuming that hreflang alone solves everything. I have seen sites with perfect hreflang tags still serve the wrong page because the URL structure was inconsistent. For example, example.com/en/ and example.com/uk/ targeting the same language but different countries. Google sees them as separate pages competing for the same query. The result is that neither ranks well. The fix is to use a single URL per language-country pair, and use hreflang annotations to point to the correct localised version. If you have a page that covers both UK and US English, you need to decide whether to use a region-specific tag or let Google decide based on the user's location.

Another watch-out is site speed. A page that loads in 2 seconds in London might take 8 seconds in Sydney because the server is far away. I have seen a client lose 40% of their Australian traffic after moving to a server in the US without a CDN. I now use a CDN from day one for any international expansion. I also lazy-load images and serve fonts locally to avoid cross-origin delays. But I do not use a CDN as a substitute for a good host. The backbone matters.

Currency and pricing pages are a common tripwire. If you show prices in USD on a page targeting Mexico, users will bounce. Worse, Google might interpret the page as irrelevant to the Mexican market. I always ensure that prices are in the local currency, and that the page header or footer shows the correct country selector. I also recommend using a geolocation redirect only if the user explicitly chooses a country, otherwise you risk sending users to a page they did not intend. I prefer a soft banner: 'You appear to be in Germany. Would you like to switch to our German site?' That preserves choice and avoids the 'forced redirect' penalty that some users report.

Finally, do not forget about link building. International links are not the same as [international link building](/international-link-building/). You need links from local domains and local languages. A .gov.uk link to your UK page is worth more than a .com link from the US. I have found that local PR, guest posting on regional blogs, and partnerships with local influencers work better than trying to buy links. But I also check the link profile of my competitors in each market. If they have links from 20 local news sites and I have none, I know the gap is not just technical.

I used to think that a single global SEO strategy could work for all markets. I applied the same keyword research, the same content structure, and the same link building tactics to every country. That failed. The French market wanted long-form, detailed guides. The German market wanted comparison tables and prices. The Japanese market wanted short, visual pages with trust signals. I wasted six months optimising for a style that worked in the US but not in Europe. Now I start each market with a fresh content audit and a local brief.

I also got hreflang wrong for a long time. I used to put hreflang tags on every page, including those that were not localised, thinking that more tags meant better targeting. That created a mess. Google saw a page with a hreflang tag pointing to itself and another page that was identical, and it could not decide which one to show. I learned that hreflang should only be used on pages that are actually localised, and that you should never include a page that is a near-duplicate unless you have a genuine language variant. I also discovered that hreflang is not a ranking signal; it only helps with serving the correct version. If the content is poor, no amount of hreflang will save you.

Another mistake: I ignored the importance of country-specific search engines. I focused entirely on Google, assuming that everyone uses it. Then I worked on a project targeting Russia, where Yandex has a 60% market share. My client's pages were not indexed in Yandex at all, because I had not set up Yandex Webmaster Tools or managed the robot.txt file for YandexBot. I had to learn about [Yandex Search Engine](/yandex-search-engine/) from scratch. Similarly, in China, Baidu requires different meta tags, a different sitemap format, and a Chinese hosting license. I now ask myself for every new market: is Google the dominant search engine here? If not, I spend time learning the local one.

Finally, I overestimated the value of translation. I thought that if I translated the English content into Spanish, I would rank in Spain and Mexico. But the same word in Spain and Mexico can mean different things. 'Ordenador' vs 'computadora' is a trivial example. More importantly, the cultural references, the calls to action, and the way of addressing the user differ. I now use professional [localization SEO](/localization-seo/) services that adapt the content, not just translate it. I also test the localised page with a small group of users before publishing. That has saved me from publishing pages that felt foreign, even if the words were correct.