Vietnamese and Chinese language search in San Jose and San Francisco: what changes and what does not

Below are the population figures with their margins, and then the two things Google actually documents about a business operating in two languages. What we will not do is bridge from one to the other, because no source in this library does.

How many people are we talking about?

Enough that the ordering matters, and the ordering is the opposite in the two cities.

From the American Community Survey 2024 one-year estimates, table B16001, language spoken at home for the population five years and over:

These are estimates with sampling error, not counts. Publish the margin whenever you publish the figure.

San Jose has 13,790 more Vietnamese speakers than Chinese speakers. San Francisco has roughly nine Chinese speakers for every Vietnamese speaker. The San Francisco ratio computes to 9.4, and we are not going to write 9.4: running the Vietnamese estimate across its own margin alone moves that ratio between about seven and thirteen. Roughly nine to one is what the data supports.

The San Jose inversion, on the other hand, survives its margins comfortably. Even taking the unfavorable end of both, Vietnamese still exceeds Chinese in San Jose. That is a finding you can plan around.

Santa Clara County, same release: Chinese 197,041, Vietnamese 123,953. San Francisco County is coterminous with the city, so the city figures serve for both.

How many of them have limited English?

A little over half in each of the larger groups, and it is the number that decides whether any of this is worth doing.

Of San Francisco's 131,407 Chinese speakers, 74,357 report speaking English less than "very well." Of San Jose's 98,606 Vietnamese speakers, 63,401 do. In San Jose's Chinese-speaking population the figure is 39,098, and in San Francisco's Vietnamese-speaking population, 8,683.

Speaking a language at home is not the same as needing a page in it. Somebody bilingual and comfortable in English is counted in the first column and is not the person a translated page is for. The second number is closer to the audience, and it is still not a measure of what anyone searches.

It is worth being precise about what "less than very well" is. It is self-reported, on a four point scale, by one person filling in a household form, and it says nothing about reading. Somebody may answer the phone in English comfortably and still prefer to read a menu, a price list or a consent form in Vietnamese. So treat the figure as the outer edge of the group a translated page could help rather than as its size.

One more thing the table cannot tell you. Chinese here is a single category covering Mandarin and Cantonese together, and those are not interchangeable to a customer choosing where to go. The Census release does not split them at this level. Your staff can, and in San Francisco that distinction usually matters more than the total does.

What can you conclude about search from this?

Nothing, and this is the part that gets fabricated most.

The table counts language spoken at home. It is not a search behavior source. There is no bridge in this library from a language spoken at home to a query typed into Google, and we have not found a primary source that builds one. Anyone quoting you a share of Bay Area searches in Vietnamese should be asked for the table it came from.

What the figures do support is a decision about whether to invest at all, and where. A restaurant in San Jose weighing one translated page has a materially different case from the same restaurant in San Francisco, and the direction of that difference is the opposite of what a single Bay Area strategy would assume.

Measure your own demand rather than the region's. Your Business Profile performance report and your Search Console query list are yours, they are free, and they describe your customers rather than a city. That is the same argument as reading a monthly local report critically: the number that decides anything is the one from your own account.

What does Google actually say about a business name in two languages?

That you may not use both, and the rule names the exact reason people do it.

This is the sharpest thing in this article and almost nobody knows it. Google's guidelines for representing your business list, under what a name must not include:

"Repeated Bilingual Names / Script Transliterations (Repeating the same business name in multiple scripts or languages, even if it appears this way on physical storefront signage)"

Read the parenthetical twice. The instinct is to put both names in the field because both are on the window, and that is the case the rule describes and rejects. The name field takes the real-world name once.

Marketing taglines are prohibited separately, and Google's own example is "TD Bank, America's Most Convenient Bank" as unacceptable against "TD Bank." A name field is not a positioning statement in either language.

What does carry the second language is everything else: the description, the services, the posts, the photos, the replies to reviews. Those are where a practice or a restaurant shows it operates in Vietnamese or Cantonese, and the field-by-field version is in adding services and products to your profile correctly.

What does hreflang do, and what does it not do?

It maps versions of a page to each other. It does not tell Google what language a page is in.

The second half of that sentence is Google's, verbatim, and it is the most commonly reversed fact in multilingual SEO:

"Google doesn't use hreflang or the HTML lang attribute to detect the language of a page; instead, we use algorithms to determine the language."

So hreflang cannot rescue a page whose language Google reads from its content. If your Vietnamese page is mostly English boilerplate with a translated paragraph, the annotation does not change how it is read.

The requirement that catches most small sites is mutual linking. Google's wording: "Each language version must list itself as well as all other language versions." And the consequence, also verbatim: "If two pages don't both point to each other, the tags will be ignored." A one-way annotation is not weaker. It is inert. Add `x-default` for the languages you do not cover.

None of that is complicated on a small site, and the mechanics are worked through in hreflang for small multilingual sites. What it is not is a reason to build the second language in the first place. That decision comes first and is covered in whether your website should be in more than one language.

What would we do first, in each city?

Different orders, and that is the entire practical point of the population figures.

In San Jose, start with Vietnamese and start with the profile, not the site. Description, services, hours, posts and review replies in Vietnamese cost nothing and go live the day you write them. A translated page can follow if the profile shows the interest.

In San Francisco, the same work in Chinese, deciding Mandarin or Cantonese from who actually walks in rather than from the table, which does not split them.

Fix the name field first in both cities, because a bilingual name is a prohibited name, and a prohibited name is a suspension risk rather than a ranking problem. If yours repeats itself in two scripts, that is a ten minute edit and it is the highest value ten minutes on this list.

Then check whether your city page is real. A Vietnamese page for San Jose that says the same thing as the English one with the city swapped is a thin page in two languages instead of one, which is the failure described in local landing pages: how many before they start hurting.

And keep the neighborhood question separate from the language question. They feel related and they are not, which is why ranking in one part of town and not another is its own problem with its own fix.

Want a read on which language, if either?

Email me at eric@seod.com with your city and your Business Profile link, and I will tell you three things: whether your name field is carrying a prohibited bilingual repeat, what your own performance report already says about who is finding you, and whether the profile or the site is the cheaper place to start. Takes about ten minutes and there is no proposal attached.

I will not tell you what share of your market searches in another language. Nobody can, from any source I have been able to read.

If you want the profile, the pages and the local work handled together, that is Search visibility. There is more on profiles and local search in the library.