From the Greek Market to International: Bilingual SEO Without Losing What You Built

The request usually arrives the same way. A Greek business has grown domestically, someone suggests an English version of the site, and a translator is briefed to convert the existing pages. Six months later the English section receives almost no traffic, nobody can explain why, and the conclusion is that "international SEO does not work for us".

What failed was not the ambition. It was the assumption that a second market is the same market in a different language. This article covers bilingual and international expansion with the Semalt platform: deciding whether to expand at all, structuring the site correctly, getting hreflang right, and tracking two markets so that one cannot conceal the other.

First: is a second language actually worth it?

The honest answer for many businesses is no, and it is cheaper to establish that before building than after. Three conditions have to hold.

1. Demonstrable demand exists

   Not "people search in English", which is always true, but volume for your specific service in the markets you could realistically serve. Check the query set before committing, and check it in the target country rather than from Athens.

2. You can fulfil the enquiry

   An English page that generates an enquiry nobody in the office wants to answer produces a bad experience and no revenue. This sounds obvious and is the most common reason English sections quietly die.

3. Someone owns the content

   Translated once and never updated, an English section decays into a stale mirror of a site that has moved on. Budget for maintenance or do not start.

Why translation is not localisation

The central mistake is treating the Greek page as a source document to be converted. It was written for a Greek buyer with Greek assumptions, referencing Greek institutions, prices in a Greek context, and answering the questions a Greek customer asks. The English visitor has different questions in a different order.

A concrete example we see repeatedly. A Greek legal services page explains the process, the documents required and the timeline. Its English counterpart is read mostly by non-residents who are wondering something the Greek page never addresses at all: whether they can do this without being physically present in Greece, and what happens to the process if they cannot. Translating the Greek page perfectly produces a page that answers the wrong question fluently.

The keyword row deserves emphasis because it produces the most avoidable failures. Direct translation of a Greek commercial term frequently yields a phrase nobody searches. The industry term differs, the level of formality differs, and sometimes the concept is bundled differently — what Greeks search as one service is, in another market, two separate services with two separate query sets.

Structure: subdirectory, subdomain or separate domain

For almost every Greek business expanding into one or two additional languages, language subdirectories on the main domain are the correct answer. They inherit the authority the domain has already accumulated, they are simpler to maintain, and they keep everything in one property for reporting.

We cover this in detail in From Rankings to Revenue.

Subdomains and separate country domains have legitimate uses — genuinely separate operations, separate legal entities, separate teams with separate budgets — but they start from zero authority and multiply the maintenance burden. Choosing one because it looks more international is a decision that costs years.

hreflang: the three ways it breaks

The specification is simple and the implementation is broken on a majority of the multilingual sites we audit. Three failure modes account for almost all of it.

Missing return links. The Greek page points to the English one, but the English page does not point back. The annotation must be reciprocal, and a one-way declaration is generally ignored, which means both pages lose the benefit.

Cross-language canonicals. The English page carries a canonical tag pointing at the Greek page — usually inherited from a template or a plugin default. This is the most damaging of the three, because it explicitly tells search engines that the English page is a duplicate that should not be indexed at all. The entire English section can disappear from search results while looking perfectly healthy to a visitor.

Partial translation. Pages that exist in Greek but not in English, where the language switcher points to a page that does not exist or redirects to the English homepage. Every visitor who clicks it lands somewhere they did not intend, and the hreflang cluster is inconsistent.

All three are visible in a crawl and invisible to anyone browsing the site, which is exactly why they persist for years. The audit surfaces missing return links, cross-language canonicals and orphaned language versions as a group, which turns a vague suspicion into a list a developer can work through.

Tracking two markets without blending them

The reporting mistake mirrors the content mistake: one visibility number covering both languages. It is always dominated by the larger market, so the smaller one becomes invisible exactly when it most needs attention.

Set up each language as its own tracked group, with its own location and language parameters and its own competitor set. The competitor point is easy to overlook and matters a great deal: your Greek competitors are usually not your English-market competitors, and comparing yourself against the wrong set produces a comfortable chart and a wrong strategy.

Choosing the market, not just the language

"English" is not a market. It is a language spoken across dozens of markets with different competitors, different search behaviour and wildly different commercial value for a Greek business. Deciding which market you are actually targeting changes almost every downstream decision.

Planning parameters from the sequence described below.

For most Athens businesses, the realistic English-speaking audiences fall into three groups, and they behave differently. Visitors and short-term residents already in Greece search for practical, immediate needs and convert quickly. Diaspora and property owners abroad search for services connected to Greece specifically — inheritance, property, tax, construction — with long consideration cycles and high value. Genuine export customers, who are looking for a supplier and neither know nor care that you are in Athens, are the hardest group and the one where a Greek location is least relevant to the buyer.

There is a full walkthrough in Seasonality and Tourism.

These need different pages. The first group wants logistics and speed; the second wants credibility and process clarity; the third wants specification and proof of capability. A single English section that tries to serve all three usually serves none, and it is the most common reason a well-translated site fails to convert.

Track them separately too. Grouping English queries by these audiences rather than lumping them into one "EN" bucket shows which of the three is actually responding — and in our experience it is rarely the one the client expected when the project started.

See also: The First 30 Days with Semalt.

A realistic sequence

Expansion done properly is slower and much cheaper than expansion done all at once.

Start with the commercial core: the three to five service pages that generate revenue, localised rather than translated. Do not translate the blog. Informational content is the largest volume of work, the slowest to pay back, and the part most exposed to being answered directly in the results page without a click.

Once those pages have three months of data, you will know whether the market responds. If it does, extend into the supporting content that the commercial pages need — the questions that arise before a decision. If it does not, you have learned it at the cost of five pages rather than a hundred.

Throughout, keep the two language sets separate in reporting, and review them separately in the monthly meeting. The moment they are averaged together, the smaller market stops being managed.

The realistic view

International expansion is not a translation project with an SEO component. It is a market entry with a content component, and the businesses that succeed at it treat the second language as a second business rather than a second version of a website.

The technical part is genuinely straightforward: subdirectories, reciprocal hreflang, self-referencing canonicals, separate tracking. The hard part is editorial — accepting that the English pages need to be written for a different reader, and that a smaller number of properly localised pages will beat a complete but translated mirror every time.

If you already run a bilingual site and have never audited the hreflang relationships, that is the first thing to check, and it takes minutes. Open the dashboard, crawl both language versions, and look specifically for missing return links and canonicals that point across languages.