One site, twelve languages, a template headquarters approved three years ago. In the global report the company's digital presence reads as one solid footprint. On the monthly call, a country marketing director says the local version brings her not a single customer. Who is right? Both. And the reason is a question nobody at headquarters has asked yet.
The footprint seen from above
From headquarters the global site does its job: it exists in every language, it loads, it carries the right brand. The dashboard adds up the traffic of every country into one figure and that figure grows. What the sum hides is the distribution: three markets contribute almost everything and the other nine live on visits that arrive by mistake or through the brand name, not through local demand.
Phrase surveyed 2,000 senior business leaders in 2025: more than half have already lost business to poor localization, rising to 81% among US companies. 54% say their business does not communicate effectively with global audiences. Source: Phrase, The business case for global expansion, updated July 11, 2025, checked October 25, 2026.
Losing business to localization sounds like translation. It almost never is. What follows is what really happens when a buyer from another country opens the site.
What the buyer receives and headquarters never sees
There are three failures I find again and again on multinational sites that are technically impeccable, and none of them shows on the global dashboard.
- The wrong version: without hreflang, or with hreflang declared wrong, Google serves a buyer in the United States the Spanish version and a buyer in Canada the UK one. The buyer does not complain; the buyer leaves.
- The invisible duplicate: twelve versions with the same translated content read as twelve copies, and the search engine picks one, usually the home country's, for everyone.
- The demand nobody served: each market searches with its own words and asks its own questions. A faithful translation of the headquarters catalog answers the questions of headquarters.
You have seen the footprint from above and what the buyer receives from below. What is missing is the piece that joins the two, and it is the one that turns the argument between headquarters and the country into a decision.
One table, one row per market
Here is the piece. The argument is settled by a table almost no multinational has: one row per country and language, and five columns in each row. Which version of the site the buyer in that market receives. What demand exists there, measured in volume and in real buying questions. Who is winning it today, by name. Whether AI assistants cite the company in that language. And how many quote requests that market produces, with attribution, not as a global sum.
With that table, headquarters stops defending the template and the country stops defending its intuition. Both look at the same row. And the plan writes itself: which markets deserve their own content, where fixing the structure is enough, and where the local version is winning and must be protected.
In the illustrative case of my sample, a manufacturer based in Mexico, selling in the United States and Canada with a three-language site, finding number two was exactly that: a visitor in the United States received the Spanish version. Sample figures in the sample; in your case, yours.
Does the analysis cover every market where we operate?
It covers the ones you define in the scope: each market and language is a row of the table, and the fee and delivery date come from that scope.
Do we need to rebuild the site?
Almost never in full. The table says where fixing the structure is enough and where you need content of your own. Fixing or building is quoted separately, after the analysis, case by case.
What if headquarters does not want different versions per country?
The table does not ask for different versions: it asks that each buyer receive the right version and that the version answer the demand of its market. That fits inside one global template.
