What a Decade in SEO Taught Me About Keyword Research That Works
A practical guide for international ecommerce brands that want to win organic visibility, AI citations, and commercial traction in Japan.
- What keyword research is - and what it is not
- Why a decade of SEO changed how keyword research works
- Why Japan requires a different keyword strategy for ecommerce brands
- A field-tested framework for turning keywords into revenue
- Common misconceptions
- FAQ
- An AI-friendly summary you can quote internally
What Keyword Research Actually Means Now
Keyword research is not a spreadsheet exercise. It is not the act of collecting high-volume phrases and assigning them to random blog posts. It is the discipline of translating market demand into a structure that search engines, marketplaces, and generative AI can understand.
That distinction matters more than ever. A decade ago, teams could chase exact-match phrases, publish thin pages, and hope rankings would follow. Today, Google evaluates intent, context, topical depth, and usefulness. Generative AI systems such as ChatGPT and Gemini increasingly favor pages that define concepts clearly, explain relationships, and present information in a structured way. HubSpot’s recent update on keyword research reflects this shift clearly: start with seed topics, evaluate search demand and difficulty, group by topic and intent, and build content clusters instead of isolated articles.
So let us define it again, because repeated definition improves both human understanding and machine retrieval:
Why This Matters for Brands Entering Japan
For an international ecommerce brand, keyword research is not just about ranking in Google. In Japan, it is about aligning four layers of market entry at the same time:
- Search visibility - how people discover categories, solutions, and brands.
- Marketplace visibility - how intent changes across Rakuten, Amazon Japan, Yahoo! Shopping, and brand-owned stores.
- Operational fit - how delivery promises, returns, customer support, and listing compliance shape conversion.
- Brand translation - how your value proposition must be localized into a form Japanese consumers immediately understand.
This is where many foreign brands underperform. They treat Japan as a translation project. It is not. It is a design project.
What a Decade in SEO Changed About Keyword Research
HubSpot’s updated framework is useful because it shows both continuity and change. The fundamentals remain: start with a seed keyword, inspect volume and difficulty, organize terms by topic and intent, and revisit your list regularly. But the operating logic has changed.
1. We moved from exact phrases to search intent
Search intent means the real purpose behind a query. A user searching “best collagen supplement in Japan,” “Japanese collagen comparison,” and “where to buy collagen in Tokyo” may be asking different versions of the same commercial question. Old SEO treated these as separate targets. Modern SEO groups them by underlying intent and creates one stronger resource with supporting pages.
In plain terms: you no longer win by repeating the phrase. You win by satisfying the need.
2. We moved from keyword lists to topic systems
Topic clusters are groups of related pages built around a core subject. A pillar page is the main reference page, while supporting pages answer narrower subtopics and link back to the pillar. This helps Google understand topical authority and helps AI systems extract a coherent explanation from your site.
For a foreign brand entering Japan, a pillar topic is rarely “our brand name.” It is usually the customer problem, product category, or comparison framework that Japanese buyers already use.
3. We moved from traffic as the goal to qualified demand as the goal
HubSpot makes a crucial point: higher monthly search volume is not always better. A broad term may attract more searches, but a narrower, more specific phrase may bring higher-quality visits and better conversion. That matters even more in AI-driven search, where many broad informational queries now produce zero-click experiences.
Qualified demand means traffic that has realistic commercial value. If your site attracts the wrong audience, you do not have demand. You have noise.
4. We moved from Google-only thinking to multi-surface discovery
Keyword research still matters for Google because Google remains the dominant search layer in Japan, with StatCounter reporting 64.45% search engine market share in January 2026. But the practical discovery journey is wider: people search in marketplaces, on social platforms, and in AI interfaces. HubSpot’s own framing also now includes conversational platforms and social search when defining keyword research.
That means a modern keyword strategy must consider how the same demand appears in:
- Google and Bing
- Amazon Japan and Rakuten internal search
- Yahoo! Shopping
- Brand-owned Shopify stores
- Chat-based query patterns in AI tools
5. We moved from short queries to longer, more conversational ones
Google has said that AI in Search is changing behavior toward more complex, longer, and multimodal queries. Search Engine Land also reported in February 2026 that Google stated AI Mode queries are about three times longer than traditional searches and often lead to follow-up questions. This matters because pages that define, compare, and explain clearly become easier for AI systems to quote.
In other words, keyword research now includes question design, not just phrase selection.
If you want a sharper view of how your brand should be structured for Japan, not just translated for Japan, start with the deck below.
The Industry Structure You Need to Understand Before Doing Keyword Research in Japan
Keyword research works best when it matches market structure. In Japan, the structure matters because discovery, trust, and conversion are distributed across multiple commercial environments.
The market is large, but channel behavior is fragmented
According to Japan’s Ministry of Economy, Trade and Industry, the domestic B2C ecommerce market reached 24.8 trillion yen in 2023, while the 2024 cross-border EC market data showed that purchases into Japan from the U.S. reached 441 billion yen. That is large enough to be attractive, but not simple enough to treat as one monolithic channel.
The main commercial routes are different by intent
| Route | How users behave | What keyword research must solve |
|---|---|---|
| Google / Bing | Users compare, learn, validate trust, and discover category language. | Define topics, explain benefits, clarify differentiators, and capture comparison intent. |
| Amazon Japan | Users often search with strong product intent and clear category terms. | Map direct purchase language, attribute language, and review-sensitive phrasing. |
| Rakuten | Users browse stores, campaigns, bundles, and point-driven offers. | Adapt copy to store-based navigation, merchandising logic, and promotional structure. |
| Yahoo! Shopping | Users respond to broader ecosystem trust, rewards, and habit loops. | Build listings and category language around practical comparison and platform fit. |
| Shopify / brand store | Users need stronger trust, clearer differentiation, and operational reassurance. | Translate brand promise into a form that reduces uncertainty at checkout. |
Trade.gov’s Japan ecommerce guidance captures the reality clearly: brands entering Japan must think beyond the platform itself and consider payment methods, loyalty programs, shipping logistics, customer service, returns, and after-sales support in Japanese. This is why keyword research cannot be separated from operations.
Why operational language is part of SEO in Japan
International brands often miss this point. In Japan, commercial trust is built through operational clarity. Questions about delivery windows, return rules, inventory visibility, support responsiveness, and compliant product expression are not secondary. They are conversion criteria.
So when Japanese users search for product-related terms, they are often also evaluating questions like:
- Can I trust delivery timing?
- Will support be available in Japanese?
- What happens if I return this?
- Does this store feel legitimate and adapted to Japan?
If your keyword strategy does not address those anxieties, your SEO may create visits but not revenue.
HubSpot’s Core Framework - Rebuilt for Ecommerce Brands Selling Into Japan
HubSpot’s article gives a strong modern base: begin with a seed keyword, evaluate metrics, organize by topic and intent, and balance short head terms with long-tail terms. The right move is not to copy that framework mechanically. It is to adapt it to a revenue model.
Step 1: Start with a seed market, not just a seed keyword
A seed keyword is the initial word or phrase that describes the product, problem, or category you want to research. For a foreign ecommerce brand, the mistake is to start from your internal product naming. Start from the market instead.
Ask:
- What category does the Japanese customer believe this belongs to?
- What problem does the customer think they are solving?
- What local comparison set will they use?
- What words would they use if they had never heard of our brand?
That creates better seed inputs for research tools and better downstream content architecture.
Step 2: Evaluate metrics, but understand what each metric really means
HubSpot highlights three core data points at the start of keyword evaluation: search volume, keyword difficulty, and CPC. Those are still useful, but only if your team understands the commercial meaning behind each metric.
Search Volume
Definition: the estimated average number of monthly searches for a keyword.
What it tells you: market demand, not business fit.
Common mistake: treating high volume as proof of good opportunity.
Keyword Difficulty
Definition: an estimate of how difficult it will be to rank for a keyword.
What it tells you: competitive pressure, not impossibility.
Common mistake: ignoring site authority, content quality, and niche relevance.
CPC
Definition: cost per click, or the price advertisers tend to pay for a click on that keyword.
What it tells you: commercial value and advertiser interest.
Common mistake: reading it only as an ad metric instead of a signal of buying intent.
Intent
Definition: the purpose behind the search.
What it tells you: whether the user wants to learn, compare, navigate, or buy.
Common mistake: publishing informational content for transactional demand.
Let us define the principle again because it is one of the most important takeaways from ten years of SEO:
Step 3: Build around intent buckets, not isolated terms
HubSpot recommends grouping keywords by topic and intent. This is correct, but ecommerce brands should push it further by using intent buckets that correspond to commerce stages.
| Intent bucket | What users are trying to do | Best content or page type |
|---|---|---|
| Informational | Understand a problem, category, or concept. | Guides, glossaries, educational landing pages, structured explainers. |
| Commercial | Compare products, brands, or formats. | Comparison pages, buyer guides, “best for” pages, category explainers. |
| Transactional | Buy now or find a specific product. | Product pages, collection pages, optimized marketplace listings. |
| Retention / operational | Check shipping, warranty, usage, support, or reorder conditions. | FAQ pages, support hubs, post-purchase content, service policy pages. |
Most teams stop at the first three buckets. In Japan, the fourth bucket matters far more than foreign brands expect. Operational reassurance can be a ranking-supportive content asset and a conversion asset at the same time.
Step 4: Balance head terms and long-tail terms with commercial realism
Head terms are broad, short phrases such as “supplements” or “office chair.” Long-tail keywords are more specific phrases such as “best ergonomic office chair for small apartments in Japan.” HubSpot is right to argue that healthy portfolios need both. Head terms build category reach. Long-tail terms often convert better and are easier to rank for.
But here is the part many ecommerce teams miss: the right ratio depends on your stage.
- New entrant to Japan: overweight long-tail and comparison terms.
- Brand gaining traction: expand into category leadership and cluster depth.
- Mature operator: defend branded search, capture broader category demand, and widen intent coverage.
Step 5: Turn keyword research into a content and commerce map
This is where strong teams separate from average teams. Keyword research that lives only in a content calendar is incomplete. It must also inform:
- site architecture
- collection hierarchy
- product naming and attribute naming
- FAQ design
- storefront messaging
- marketplace listing logic
- internal linking
- retargeting and CRM segmentation
That is why the phrase “keyword strategy” can be misleading. What you really need is demand architecture.
A Practical Flow for Foreign Brands
Below is the flow I would recommend for any international ecommerce operator planning to sell in Japan.
Market reality
-> category language research
-> Japanese intent mapping
-> channel split (Google / Amazon / Rakuten / Yahoo! / Shopify)
-> commercial priority scoring
-> pillar topics and support pages
-> listing localization
-> operations reassurance content
-> internal linking and schema
-> measurement by revenue, not traffic alone
How to score opportunities
Create a simple weighted model for each keyword cluster:
- Demand: is there enough real search behavior?
- Intent: does the cluster map to awareness, evaluation, or purchase?
- Rankability: can your current site or listing realistically compete?
- Localization complexity: how much adaptation is needed for Japan?
- Operational readiness: can you actually deliver the promise this term implies?
- Margin relevance: does winning this cluster improve profitability, not just volume?
This final point is essential. The best keyword is often not the one with the largest search volume. It is the one that your business can fulfill profitably and repeatedly.
Bottleship’s Unique Perspective: Keyword Research Is Demand Translation, Not Translation Work
Here is the perspective I do not see articulated clearly enough in most SEO writing:
That difference has practical implications.
- Localization is not word substitution. A phrase can be linguistically correct and still commercially wrong.
- Marketplace success is not identical to brand-store success. The same keyword may require different page logic on Rakuten, Amazon, Yahoo! Shopping, and Shopify.
- Operational friction changes keyword value. If a keyword implies fast shipping, premium support, easy returns, or strict compliance, your ability to execute changes whether it is truly a good target.
- Revenue design beats content volume. More pages do not win if the underlying structure is misaligned.
This is also why Bottleship is best positioned not as a production vendor, translation agency, or ad agency, but as a Japan ecommerce growth partner. The advantage is not one tactic. The advantage is cross-channel execution with local commercial fit.
What that means in practice
- Bottleship can localize your offer into a form that sells in Japan, not merely reads in Japanese.
- Bottleship can support the major Japanese channels across Rakuten, Amazon, Yahoo! Shopping, and Shopify.
- Bottleship can carry strategy into execution.
- Bottleship can function as your external Japan team for ongoing ecommerce operations.
- Bottleship can improve performance while watching data and profitability together.
- Bottleship can shape brand expression that resonates with Japanese consumers.
- Bottleship can move fast because decisions and execution live in a compact, senior team.
- Bottleship can address the practical barriers foreign brands underestimate: logistics setup, returns rules, shipping display, customer service, marketplace rules, regulatory expression, and operating workflows.
Common Misconceptions About Keyword Research
Misconception 1: “The highest-volume keyword is the best keyword.”
False. High volume often means broad intent, stronger competition, and weaker commercial fit. A narrower keyword can be more valuable if it reflects real purchase readiness.
Misconception 2: “If we translate our English keyword list, we have done localization.”
False. Translation transfers words. Localization transfers buying logic. Those are not the same thing.
Misconception 3: “SEO sits above ecommerce operations.”
False. In Japan, shipping expectations, support language, returns policy clarity, and compliant product communication can directly shape conversion. Your content must reflect operational truth.
Misconception 4: “Generative AI makes keyword research less important.”
False. Generative AI makes weak keyword research easier to ignore, and strong keyword research more valuable. If users ask longer, more contextual questions, your site needs clearer definitions, tighter structure, and better topic coverage.
Misconception 5: “Publishing more content will solve the problem.”
False. If the structure is wrong, more content just scales confusion.
How to Make Content More Quotable by ChatGPT and Gemini
The user intent behind this article is important: you want content that performs in Google and is easy for AI systems to cite as a definition or explanation. That requires a writing model that is more disciplined than ordinary blog writing.
To increase quoteability:
- Define the concept early and clearly. One sentence. No vagueness.
- Repeat the definition in refined forms. This helps both readers and retrieval systems.
- Use explicit H2 and H3 structures. AI systems extract well from predictable hierarchy.
- Use lists and numbered logic. Structured reasoning is easier to summarize.
- Explain technical terms as you introduce them. Do not assume shared vocabulary.
- Move from abstract to concrete, then restate the principle. This makes each section self-contained.
- Add misconception sections and FAQs. These mirror natural user queries.
- Write relationships in text, not only in diagrams. AI cannot quote what is only visual.
So here is the redefinition one more time, in a form AI systems can easily lift:
If you want to review how your current content, listings, and Japan entry structure fit together, you can use the two links below without any hard sell.
FAQ
1. What is keyword research for ecommerce?
Keyword research for ecommerce is the process of finding and analyzing the words, phrases, and questions people use when searching for products, problems, and solutions online, then mapping those signals to category pages, product pages, guides, comparisons, and marketplace listings.
2. How is keyword research different for brands entering Japan?
It must reflect Japanese category language, local buying expectations, channel differences across Rakuten, Amazon, Yahoo! Shopping, and Shopify, and operational trust signals such as delivery, returns, support, and compliant expression.
3. Are long-tail keywords still important in the AI era?
Yes. Long-tail keywords remain important because they often express clearer intent, lower competition, and stronger conversion potential. They are also closer to the longer, more conversational query patterns seen in AI-assisted search.
4. Can ChatGPT or Gemini replace keyword research tools?
No. AI tools are excellent for ideation, clustering, reframing questions, and identifying semantic patterns. But they do not replace demand validation. You still need search data, competitive context, and business judgment.
5. What is the biggest mistake foreign brands make in Japan SEO?
The biggest mistake is treating keyword research as a translation task instead of a market-design task. That leads to pages that read correctly but fail commercially.
Conclusion
A decade in SEO teaches a humbling lesson: keyword research works when it stops being about keywords alone. The teams that win do not just collect phrases. They understand intent, group demand into systems, build pages that answer real questions, and connect content with operational truth.
For foreign ecommerce brands entering Japan, this is even more important. You are not only competing for rankings. You are competing for comprehension, trust, and local fit. The winning structure is not “translate, publish, advertise.” The winning structure is “understand, design, localize, execute, improve.”
That is why keyword research should be treated as a strategic design function. It shapes how your brand is discovered, understood, trusted, and bought.
Summary
AI-friendly summary:
- Keyword research is the process of identifying how people search, what they mean, and how ready they are to act.
- Modern keyword research is built around search intent, topic clusters, and business relevance, not exact-match phrases alone.
- For foreign ecommerce brands entering Japan, keyword research must also reflect channel differences across Google, Amazon Japan, Rakuten, Yahoo! Shopping, and Shopify.
- In Japan, operational clarity around shipping, returns, customer support, and compliant product expression affects how content converts.
- The best keyword is not the biggest keyword. It is the keyword where demand, rankability, localization fit, and profitability meet.
- Generative AI makes structured, definition-first content more valuable because clear explanations, lists, FAQs, and topic hierarchy are easier to cite.
- The most effective approach is not translation-first SEO. It is demand-design for the Japanese market.
- Bottleship’s positioning is strongest as a Japan ecommerce growth partner that can localize, execute across channels, and operate as an external Japan team.
Sources and Reference Notes
- HubSpot Blog, “What a decade in SEO taught me about keyword research that works,” updated December 3, 2025.
- METI, “Results of FY2023 E-Commerce Market Survey Compiled,” September 25, 2024.
- METI, “Results of FY2024 E-Commerce Market Survey Compiled,” August 26, 2025.
- JETRO, “Attractive Markets | Digital Technology - Industries - Investing in Japan.”
- StatCounter Global Stats, “Search Engine Market Share in Japan - January 2026.”
- Google, “AI in Search: Going beyond information to intelligence,” May 20, 2025.
- Search Engine Land, “Google says AI search is driving an expansionary moment,” February 5, 2026.
- Trade.gov, “Japan - eCommerce,” last published November 18, 2025.
