How we cost a bilingual site.

The most common quote pattern we see in Geneva is "1× for English, 0.5× for French". Sometimes the ratio is 0.7 or 0.3. The number is almost never right. Because the second language isn't a translation problem, it's an architecture problem.

Here's how the math actually works on our side.

What the second language actually costs

There are three real costs, none of which are "translation words":

Content architecture — the taxonomy, URL structure, canonical + hreflang map, and internal linking rules have to be planned bilingual from day one. Bolted on later, this is where inherited multilingual sites bleed traffic. Cost: real, roughly 15% of the tech build if planned upfront, 40%+ if retrofitted.

Design constraints — French text runs roughly 15-25% longer than English on the same idea. Buttons, cards, nav labels, form field labels, error messages all have to fit both without breaking. Add German or Italian to the mix and the constraint tightens. This is a design tax paid on every component. Cost: baked into a properly-structured design system; painful if the system was designed for one language.

Content operations — someone has to run the second language on an ongoing basis. Translation, review, publish, keep in sync when the primary language changes. This isn't a build cost, but ignoring it is why so many bilingual sites drift out of sync six months in. We plan the workflow at brief time; the ongoing cost is real and belongs on a retainer if the client doesn't have in-house bilingual writers.

What the second language doesn't cost

The translation itself, in most cases. If your team writes in French and English natively (as most Geneva businesses do), the marginal cost of an extra language is your team's time — which is real, but it's not our line item. If we're commissioning translation from a specialist for regulated content, that's a pass-through cost with a source we name in the quote.

What ends up on the invoice

For a typical Estorya bilingual build, the second language adds roughly 20-25% to the base scope — not 50%, not 100%. Broken down: 10-15% on architecture (URLs, hreflang, canonical, sitemap), 5-10% on design system adjustments so nothing breaks under the longer strings, and a small ops line for handoff.

Where we've seen quotes at 50-100%, it's almost always because the studio is quoting the same design work twice (once per language). That's the wrong shape. The right shape is: build once, structured for both, priced accordingly.

The one exception

If the two languages carry substantively different content — different services, different case studies, different legal disclosures per market — then yes, it's closer to two sites and closer to 2× cost. That's a strategic call the client should make deliberately, not one they should discover in month four when the FR site diverges from the EN.

We'll call that decision out in the discovery phase and put a number against each option.