International Keyword Research: From Search Intent to Content Architecture

International keyword research is not the act of translating a Chinese keyword list into English, simplified Chinese, or another Chinese variant. Each market needs a fresh view of how people search, compare options, and make decisions. A Hong Kong company expanding overseas may need to work across traditional Chinese for Hong Kong and Taiwan, simplified Chinese for Singapore and Malaysia, and English-language markets at the same time. If the keyword choice is wrong, even well-written content may attract no organic traffic. Ahrefs’ study of more than one billion pages found that 90.63% of pages receive no organic traffic from Google. This article sets out a practical framework from search intent and metric interpretation to local wording and content architecture. It is a working method, not a guarantee of rankings or traffic.

Why Translation Is Not Keyword Research

Translation and localisation are not the same. TransPerfect’s hotel example is a useful warning: “cheap hotel” had about 13,500 monthly searches and the French phrase “hôtel pas cher” had about 8,100; but “budget hotel” had about 14,800 searches while the literal French translation “hôtel budget” had zero. “Low cost hotel” had about 480 searches, but “hôtel bas coût” also had zero. A phrase can be understandable and still be wrong for search.

Even the same language changes by market. US searchers may use “vacation,” “sneakers,” and “cell phone,” while UK searchers may use “holiday” and “trainers.” In Germany, “Handy” can mean mobile phone. CSA Research also reports that 75% of buyers in non-English-speaking countries prefer to buy in their native language. The first rule is therefore simple: rebuild the keyword set in the language of the target buyer, not in the language of the internal team.

First Decide: Country Targeting or Language Targeting

Before building an international SEO plan, decide whether the site is targeting countries or languages. Country targeting treats Hong Kong, Taiwan, Singapore, Malaysia, and other markets separately. Language targeting treats variants such as zh-Hant, zh-Hans, and en as the primary units. This affects URL structure, hreflang, keyword volume sources, and whether content needs to be rewritten for each market.

Choosing the URL Structure

For many multilingual websites, subdirectories such as example.com/en/ and example.com/zh-hant/ are the most practical structure. This is an industry practice consensus rather than a Google rule, but it is easier to scale and usually keeps domain authority more concentrated than ccTLDs or subdomains. For hreflang, use hyphens, such as zh-Hant-TW, not underscores. Each language version should reference itself and all other versions, with reciprocal links and full http or https URLs. Google also states that it primarily uses algorithms to determine language, not only hreflang or the lang attribute.

Understand Search Intent Before Choosing the Page Type

Keyword research should not start with volume. It should start with intent. The common categories are informational, navigational, commercial, and transactional. Semrush and Yoast use similar classifications. Andy Crestodina’s practical estimate is that roughly 80% of searches are informational, 10% transactional, and 10% navigational. This is not an official statistic, but it is a useful reminder that not every keyword should become a sales page.

Ahrefs’ Business Potential score is especially useful for content planning. The question is not only “How much traffic could this keyword bring?” but “How naturally can the business’s solution appear as part of the answer?” For example, “international keyword research process” suits an educational article, while “SEO consultant pricing” may be closer to commercial investigation. Google also understands synonyms, entities, and context, so modern content has to answer the full problem rather than repeat a phrase.

Reading the Metrics: Volume, KD, and Trends

Search Volume Is an Estimate

Google Keyword Planner generally reports a 12-month average. It rounds figures and may group similar terms, so “HDMI” and “High-Definition Multimedia Interface” can display the same volume. It also requires a Google Ads account and may hide or show zero for sensitive or emerging terms. Search volume is therefore a directional estimate, not an absolute truth.

Keyword Difficulty Is Not a Google Metric

Keyword Difficulty is usually expressed from 0 to 100, but Ahrefs, Semrush, and Moz use different methods. The same keyword may be Ahrefs 23, Semrush 58, and Moz “medium.” Ahrefs mainly bases KD on backlink and referring-domain counts for the top 10 ranking pages. Semrush uses a broader formula including median referring domains, Authority Score, and search volume. In practice, the opportunity is often a mid-volume, achievable, commercially valuable keyword. A 400-search keyword with KD 28 and strong buying intent can be more valuable than a 50,000-search keyword with KD 85.

Google Trends Is Not Search Volume

Google Trends shows relative interest from 0 to 100. It is normalised against the term’s own peak and the comparison set. Countries are normalised separately, so numbers should not be compared directly across markets. Trends is best for seasonality, regional interest, and directional changes, not for replacing Keyword Planner or Search Console.

Chinese Variants Across Hong Kong, Taiwan, Mainland China, Singapore, and Malaysia

Traditional and simplified Chinese are not only writing systems. They also carry different vocabulary and market habits. Mainland China, Taiwan, and Hong Kong may use different words for the same idea: 軟件／軟體／軟件, 網絡／網路／網絡, 視頻／影片／視頻, 打印／列印／打印, 用戶／使用者／用戶, 信息／資訊／資訊. Singapore and Malaysia often use simplified Chinese, but their wording is not always the same as mainland China, and English-Chinese mixing is common.

OpenCC is a useful external tool for conversions such as s2tw, s2twp, and s2hk, but it cannot replace keyword research. It can help with scripts and some vocabulary, but it cannot decide whether Taiwanese buyers actually search “網路,” whether Hong Kong users prefer “網絡,” or what intent sits behind the query. Chinese variants can be marked with zh-Hant, zh-Hans, and country-level combinations such as zh-Hant-HK, zh-Hant-TW, and zh-Hans-CN, but the content still needs local judgement.

Choose the Right Search Battlefield

Before investing in Google SEO, confirm the local search landscape. Based on the research notes, Google is still dominant in markets such as Singapore, Malaysia, Indonesia, Thailand, the Philippines, India, and the United States, with shares ranging roughly from the mid-80s to the high-90s. Where figures differ, StatCounter should be treated as the more authoritative source.

But Google is not the default everywhere. Baidu remains important in China, and .cn sites usually involve ICP filing. In South Korea, Naver and Google compete closely. In Russia, Yandex is dominant. In Southeast Asian ecommerce, Lazada and Shopee may also function as discovery and search platforms. These are external platforms and market-context considerations, not Chan & Chung products or services.

Three Free External Tools to Start With

The following are third-party tools, not company products. Google Trends is useful for seasonality, regional interest, and broad comparison, but not absolute search volume. Google Keyword Planner can expand seed terms and provide volume ranges, but its 12-month averaging, grouping, and hidden-term behaviour must be understood. Google Search Console is the most honest first-party data source after launch. It shows up to 16 months of Clicks, Impressions, CTR, and Average Position, although impressions are not the same as search volume and some low-click queries may be anonymised.

A common quick win is to identify queries in positions 5 to 20. They already have visibility, so improvements to titles, descriptions, content sections, or internal links can move faster than creating entirely new pages. For a new site, expect roughly three months before Search Console data becomes meaningful enough for stronger decisions.

From Keywords to Content Architecture

The output of keyword research should not be a spreadsheet only. It should become a page map. Each page should have one primary keyword, supporting secondary keywords, and semantically related terms. As a rule, one primary keyword maps to one page to reduce cannibalisation. A pillar-cluster structure works well for international content: the pillar page covers a broad topic such as “international SEO strategy,” while cluster pages target long-tail questions such as “how Hong Kong companies should research English keywords.”

HubSpot and other SEO resources often suggest starting with 5 to 10 pillar themes, with pillar pages around 1,500 to 3,000 words. A practical clustering method is SERP overlap: if two keywords share three or more URLs in the top 10 results, Google may consider them answerable by the same page.

Keyword Strategy in the AI Overviews Era

From 2024 to 2026, AI Overviews changed the click structure of SEO. seoClarity reports that after the US rollout began in May 2024, AI Overviews appeared on around 30% of US desktop keywords by September 2025. About 84% appeared on informational queries, and more than 99% of cited content came from top 10 pages. Ahrefs also reported that when an AI Overview appears, even the first organic result may lose around 34% of clicks. Pew found that users who see an AI Overview rarely click sources.

SparkToro reported that in the first four months of 2026, 68.01% of Google searches ended without a click. Methodologies differ, but the direction is clear: chasing short-head keywords alone is becoming weaker. Google still says SEO matters, but its guidance emphasises query fan-out, RAG grounding, and non-commodity content: original viewpoints, first-hand experience, and material worth citing. International SEO should therefore move from isolated keywords to complete question clusters.

Seven-Step Implementation Checklist

1. Decide whether to target by country or language, then define URL structure and hreflang.
2. Select 3 to 5 priority markets and check the local search engine landscape.
3. List Chinese and English seed terms, then rebuild them with local native-language input; validate 15 to 40 terms per market at the start.
4. Mark search intent for each keyword and map it to the buying journey.
5. Use KD, search volume, and business value to find achievable opportunities rather than vanity keywords.
6. Cluster keywords and map them into pillar and cluster pages.
7. After launch, use 16 months of GSC data and prioritise position 5 to 20 opportunities.

Common misconceptions include: translating into English or simplified Chinese is enough; bigger volume is always better; KD is an objective truth; Google Trends is search volume; one keyword list works across all Chinese markets; and Baidu alone reaches global buyers. Real international keyword research is judgement, not automatic tool output.

Next Step

To discuss international keyword research, SEO content architecture, and outbound market growth content consulting across traditional Chinese for Hong Kong and Taiwan, simplified Chinese for Singapore and Malaysia, and English markets, visit https://chanchung.com/zh-hant/services. Actual outcomes depend on the market, competition, website condition, and execution quality; no specific ranking or traffic result is guaranteed.

The tools and platforms mentioned in this article, including Ahrefs, Semrush, Google Keyword Planner, Google Trends, Google Search Console, OpenCC, Baidu, Naver, Yandex, Lazada, and Shopee, are external third-party tools or platforms and are not company products. Where Hong Kong company formation or regulated trust or company services (TCSP) are involved, those regulated services are provided by Intelligent Services Limited (TC010349), not Chan & Chung Consultancy Services Limited. Decisions should be made case by case, with professional tax, legal, or compliance advice where appropriate.