On screen, a second language looks like one more menu item. From the planner’s chair it is a commitment that everything you build from now on comes in two. Two pages per page, two edits per correction, two rounds of review. So the first question is not “how do we build it” but “do we actually need it”.
This series starts with that judgement and then works through addresses, indexing and operations across eight parts.
Decide from data, not instinct
“We might get enquiries from abroad” is not a reason. There is evidence already sitting in front of you.
One caution: overseas traffic figures are not demand in themselves. Crawlers and spam are mixed in, and even a real person may not be someone who can buy what you sell. Think about payment methods, currency and the range you can actually serve. If your service is delivered remotely, borders are not a barrier; if the work needs someone on site, adding a language will not produce conversions.
Pick the purpose and the method follows
Two different goals hide inside the word “multilingual”. Deciding which one you have changes the size of the job completely.
Blur that distinction and you get cost without result. If search reach is the goal, bolting on automatic translation means you are still findable in exactly one language. If usability is the goal, having people translate every page means maintaining screens nobody reads. Part three covers why the two behave so differently.
Cut the scope first
If you go ahead, start with the conversion path, not the whole site. The screens a visitor genuinely needs in order to decide are usually few: what you sell, how it runs, what it costs, how to apply. Once those four read properly in the new language, conversion works.
A plan to translate several hundred articles rarely survives its first month. And a plan that does not survive leaves behind a half-translated site — handling that state honestly is the subject of part seven.
The wider picture of search visibility lives in the SEO archive, and if you would like the technical foundations reviewed at the same time, a diagnostic is part of our optimization program.
Next part
Once you have decided to go ahead, the address scheme comes first. The next part compares three approaches — but there is one rule to settle before any of them: language never lives in a cookie.