Richmond's census language figures are striking, and they are also easy to misuse. They show that close to half the city learned Cantonese or Mandarin first, but they say nothing about which language a given resident types into a search box when they need a roofer or a dentist's office. This guide explains how to use the numbers to choose what to test, and how to let your own data decide what to fund.
What the census actually says
Richmond had 209,937 residents in the 2021 Census. The census asks about mother tongue: the first language a person learned at home in childhood and still understands. According to census tables reproduced in Wikipedia with Statistics Canada citations, the leading figures were:
| Mother tongue | Share of residents | Rough number of people |
|---|---|---|
| English | 37.07% | about 77,800 |
| Cantonese | 23.3% | about 48,900 |
| Mandarin | 22.68% | about 47,600 |
| Tagalog | 4.95% | about 10,400 |
| Punjabi | 2.74% | about 5,800 |
| Min Nan | 1.57% | about 3,300 |
The "rough number" column is our own arithmetic, multiplying the share by the total population. It is approximate and is shown only to make the scale concrete.
Two things stand out. Cantonese and Mandarin are close to equal, so treating "Chinese" as a single audience hides a real split. And English, while the largest single mother tongue, is spoken first by well under half the city.
What the numbers don't tell you
Mother tongue is a measure of where someone started, not of how they shop. Four gaps matter for marketing.
Search language varies by product
A resident who grew up speaking Cantonese might search in English for "furnace filter" or "strata lawyer" because they learned those terms in English, and in Chinese for a restaurant, a grocery item or a family service. The pattern varies by person, age and topic. Only search data and your own enquiries reveal it for your category.
Reading script is a separate question
Spoken Cantonese and Mandarin both use written Chinese, but in two scripts: Traditional and Simplified characters. Script preference often follows where and when someone learned to read, not which dialect they speak. Our guide to Traditional and Simplified Chinese covers this decision in detail.
Households are not individuals
Census figures count people. Purchases are often made by households, and a household may include a grandparent who prefers Cantonese, a parent fluent in both, and a teenager who searches only in English. The person who books the service may not be the person who searched.
Your customers are not the city
A tutoring centre in the City Centre and a fencing contractor serving East Richmond farms will have completely different customer mixes. City averages are a starting hypothesis, nothing more.
Using the numbers to choose tests
The right use of census data is to decide which languages are worth testing for your business. A simple scoring approach:
- List candidate languages. For most Richmond businesses: English, Cantonese or Mandarin (and a script choice), and possibly Tagalog or Punjabi.
- Check reply capacity. Can someone on your team answer a call or text in that language during business hours? If not, drop it or treat it as a hiring question.
- Check your own records. Even a rough count of how many recent customers preferred each language is more useful than the census.
- Check search behaviour. For your top two services, look at whether people search in that language at all. Some categories have very little non-English search even in Richmond.
- Pick one test. Choose the single language that scores well on all four checks above, and test it before adding another.
This is the core of our language planning service, and you can do a rough version yourself in an afternoon.
A worked budget test
Here is a test you can run with your own figures. Suppose you run a home-renovation business:
- Average job value: $4,000
- Gross margin: 35%, so $1,400 per job
- Close rate on enquiries: 20%
- Break-even cost per enquiry: $1,400 × 0.20 = $280
You decide to test a Cantonese-speaking, Traditional Chinese audience. Your plan:
| Item | Amount |
|---|---|
| Translation and review of one landing page and 15 ads | $900 one-time |
| Test ad budget | $1,200 per month for 3 months |
| Enquiries needed at break-even to cover ad spend | $1,200 ÷ $280 ≈ 4–5 per month |
If the test brings 8 enquiries a month at $150 each, each enquiry is worth $280 and costs $150, so you are about $1,040 a month ahead of ad cost, and the translation is paid back in the first month. If it brings 3 enquiries at $400 each, it is losing money and should be cut or fixed. In both cases you have learned something specific about your market that the census could never tell you.
One timing note: Google says that from September 2026, Search campaigns no longer use a campaign-level language setting and match on the language of the ads. As of October 2026, a test like this depends on properly written ads in the target script; confirm current behaviour on Google's help page before you launch.
Common mistakes with census data
- Budgeting by percentage. Spending 46% of your budget in Chinese because 46% of residents have Chinese mother tongues ignores everything above.
- Ignoring English for bilingual customers. Many Richmond customers are fully comfortable in English. Losing them to a competitor because your English ads were starved is a real cost.
- Translating once and forgetting. A tested language needs the same maintenance as English: new offers, updated prices, fresh replies to reviews.
- Assuming a smaller language isn't worth anything. For a niche business, a Tagalog or Punjabi landing page and reply templates can be inexpensive and very effective, even without ads.
Where to go from here
Use the census to decide what to test, your records to decide what to fund, and your monthly report to decide what to keep. If you'd like help with any step, read the services hub or the areas page for how this plays out across Richmond, then call, text or email with your numbers.
Frequently asked questions
Is mother tongue the same as the language people search in?
No. Mother tongue is the first language learned in childhood and still understood. Many Richmond residents whose mother tongue is Cantonese or Mandarin search in English for some services and in Chinese for others.
Should I split my budget in proportion to the census figures?
No. The census describes the whole city, not your customers. Use it to decide which languages to test, then let your own enquiry data set the budget split.
What about Tagalog and Punjabi?
They are the next most common non-official mother tongues in Richmond, at roughly 5% and 3%. For many businesses that is better served by translated reply templates and one landing page than by a full ad campaign.
Where do these numbers come from?
Statistics Canada's 2021 Census. The language breakdown on this page is taken from census tables reproduced, with citations to Statistics Canada, in the Wikipedia article on Richmond.