Skip to main content

MULTILINGUAL & COMMUNITY MARKETING · September 2026 · ~10 min read

Translation versus localization for small business

Translation converts your words. Localization changes what you say. A translated page makes the same argument to a different customer, and a localized page makes the argument that customer actually cares about, which is usually a different one.

On this page

Most small business language work stops at translation because translation is a quotable, finishable task. Localization has no clean end point, so it gets skipped. Then the page underperforms and the owner concludes the second language was not worth it.

The second language was worth it. The copy was answering a question nobody asked.

01

What is the actual difference?

Translation preserves the sentence. Localization preserves the intent.

Here is what that looks like in practice. Your English service page leads with speed, because your English-speaking customers are comparing you against three other quotes and want to know who can start Monday.

Your Spanish-speaking customers, in many cases, are not comparing three quotes. They came from a referral and already have a reason to trust you. Their question is different: will you charge what you said, and will the person who shows up treat their home properly.

Translate the English page and you have a Spanish page arguing about speed to a reader who was never worried about speed. Every word is correct. The page still does nothing.

Localization means asking what this customer's actual hesitation is, then writing to that. It is a copywriting job wearing a translation job's clothes.

There is a demographic reason the arguments differ, and it is measurable. In the Census Bureau's December 2023 release of the 2018 to 2022 American Community Survey five-year estimates, among people who spoke Spanish at home, 58.3% of those aged 18 to 64 spoke English "very well," against only 41.6% of those 65 and older. Among Chinese speakers of all ages, 48.2% did. Those are two different readers inside one language, and a single translated page is written for neither of them.

02

What specifically changes when you localize?

More than the language, and less than a full rebuild. The usual list:

The objection. What worries this customer before they buy is often not what worries your English-speaking customer. Ask your staff. They have heard it a hundred times.

The proof. Reviews from customers with similar names and similar concerns are worth more than a generic star count. So is a photo of a staff member who speaks the language.

The contact method. Some customer bases call, some text, some will only use a messaging app. A page that offers only a web form loses the ones who wanted to call.

The names of things. This is the biggest single lever for restaurants, because a dish name carries information no translation can reproduce. Getting menu translation right without losing the dish is its own discipline and it is worth doing carefully.

The register. How formal you are, how you address the reader, whether humor lands. This is the part machines get wrong most consistently, and the part native speakers notice fastest.

The holidays and hours. If your busiest day for one customer group is a holiday you do not currently mention anywhere, that is a localization gap, not a translation gap.

03

How do I find out what my page is actually missing?

Worked example

Count it. The whole exercise is one afternoon and you can rerun it every year.

Step one, build the question inventory. Ask three people who talk to second-language customers to write down every question they get before someone buys. Not after. Before. Say the three lists come back with 34 distinct questions once you merge the duplicates.

Step two, score your page against it. Open your second-language page and mark each question answered or not answered. On a page that was translated from the English rather than written for this reader, a typical result is nine answered. That is 26%derived coverage, and the 25 unanswered questions are the reason the page does not convert.

Step three, look at which ones are missing. They cluster. In my experience the missing block is almost always price certainty, who physically shows up, and what happens if something goes wrong. Those are trust questions, and the English page usually does not answer them either. It just never had to, because the English customer arrived with three other quotes and used those as the reference point.

Step four, price the two roads. Say a translation vendor quotes $0.14 a word for 1,900 words across four pages, which is $266 and buys you a Spanish version of a page that answers nine of 34 questions. Rewriting one page around the 25 missing answers costs you two hours of a bilingual employee's time and one hour of your own. For most operators that comparison is not close, and the cheap option is the one that fails.

Rerun the inventory in twelve months. The question list drifts as your offer changes, and that drift is the thing an annual translation contract cannot see.

04

Can I use machine translation at all?

Yes, for internal work and for scaffolding. Not for the pages where money changes hands.

Machine translation is genuinely useful for reading incoming reviews, understanding a customer email, or producing a rough draft a native speaker then rewrites. Used that way it saves real time.

Where it fails is anywhere a customer forms an impression of you. The output is grammatically fine and tonally dead, and it consistently gets exactly the words wrong that your specific customers care most about: your product names, your category names, and the phrases with cultural weight.

Be precise about the search side, because the sloppy version of this claim is everywhere. Google's documentation on localized versions says that "localized versions of a page are only considered duplicates if the main content of the page remains untranslated." A genuinely translated page is not a duplicate of your English page in Google's terms. The risk is a quality and trust risk, and it belongs to you and your customer before it belongs to a search engine.

If you do publish in two languages, tell search engines how the versions relate. hreflang handles that and it is simpler than it is usually made to sound, especially on a small site.

05

Is a translation widget the same thing as a translated site?

No, and this is the most expensive misunderstanding in the category.

The claim to check is "we localized your site," delivered as a dropdown in your header that translates the page in the browser. It is cheap, it installs in minutes, and it looks exactly like the thing you asked for. What it produces, for search purposes, is nothing.

Here is the trace. Google's multilingual guidance states that Google "recommends using different URLs for each language version of a page rather than using cookies or browser settings to adjust the content language on the page." A browser widget does not create a URL. There is no second address for Google to crawl, no second page to index, and nothing to point an hreflang annotation at. Your site has exactly as many pages as it had before.

It is worse than neutral on the human side too. The widget translates your navigation, your legal text and your form labels with the same indifference it translates your offer, and it does that live, in front of the customer, with no reviewer between the machine and the reader. That is the exact failure mode a localized page exists to avoid.

A widget is a reasonable accessibility convenience for a visitor who wants to skim your About page. It is not a second-language strategy, it is not localization, and nobody should be charging you for it as either.

06

Who should do the writing?

Someone who talks to your customers, edited by someone who writes.

The best source material in most small businesses is a staff member explaining the service out loud. Record it, transcribe it, edit it into copy. You will get the actual vocabulary, the actual objection, and the actual register, none of which a translation vendor can invent for you.

Then have a second native speaker read it for anything that sounds off. Not for grammar. For fit.

What to avoid: handing your English page to a freelance translator with no context and no access to your business. That produces a technically correct document that reads like a legal notice. It is the cheapest option and it usually returns nothing.

Localization is also not a one-time project. Your offer changes, your prices change, your customer base changes. The version that goes stale first is always the second language, because nobody in the building is reading it every day.

07

Does this apply outside the website?

Yes, and often the highest-return localization is not on the site at all.

Your Google Business Profile description. Your review requests. Your voicemail greeting. Your signage. Your review responses, which are read far more by future customers than by the person who complained, and a review response is customer service performed in public regardless of which language it is in.

One caution on the profile, because it is where an enthusiastic localization effort can cost you the listing. Google's guidelines for representing your business prohibit "Repeating the same business name in multiple scripts or languages, even if it appears this way on physical storefront signage," and state that including unnecessary information in the name "could result in the suspension of your Business Profile." Localize the description, the services and the posts. Leave the name alone.

It also extends into how you show up outside of search entirely. Sponsoring, showing up, and being present in a community is a form of localization that no page can substitute for, and community marketing works differently from advertising precisely because it is not something you can translate into existence.

08

What to do this week

Take your single most important page and ask a bilingual staff member one question: what would a customer in the other language worry about here that this page does not address. Write the answer down.

Build the question inventory. Three people, every pre-purchase question they hear, merged into one list. Score your page against it and write the fraction down with today's date.

Rewrite that page in the second language around the missing answers. Not a translation of the English. A new page making a different argument to a different reader.

Then localize your Google Business Profile description in the same voice. That is an hour and it often gets more traffic than the page.

If you are opening soon, do this before launch rather than after. The first ninety days set a lot, and a new restaurant's early visibility work compounds in a way that is hard to catch up on later.

Be honest with yourself

When you do not need this

If your customer base is monolingual, none of this applies. Spend the money on photos and reviews instead.

If you have a second language page that is already written by a native speaker who knows your business, you are probably fine. Leave it alone and check it quarterly.

If your question inventory comes back with your page answering most of the list already, stop. You have a localized page and you did not know it.

And if the second language is a language you cannot serve in, localization is the wrong problem. Staffing is the problem. No amount of good copy fixes a phone nobody can answer.

Sources

Related reading

12

Questions about a page that is not converting?

Email me at eric@seod.com with a link to your second-language page and tell me the language. I will send you back the three things on it that read as translated rather than written, plus the one argument I think it is missing. Korean and Spanish I read myself.

I answer these personally, and if the page is genuinely good I will say so and tell you where the actual bottleneck is instead.

Or keep reading more on multilingual and community marketing.

Call Eric Email Eric