How to Do Keywords Research for Ecommerce: A Category-First Method That Drives Revenue

Ecommerce keyword research is the process of identifying the search queries real buyers use, then mapping each query to the single page type best able to satisfy it — a category page, a filtered subcategory, a product detail page, or supporting content. That mapping step is what separates it from blog keyword research, where almost every keyword resolves to an article. Get the mapping wrong and you produce the most common failure in retail SEO: a product page competing for a broad plural query it can never win, and a blog post ranking for a buying query it cannot monetise. The volume numbers matter far less than the page-type decision, and it is the part most keyword guides skip entirely.

Quick Answer: Start from your category structure, not a keyword tool. List how customers describe each category, expand those terms using Google autocomplete, People Also Ask, Search Console and Keyword Planner, then assign every query to exactly one page type based on intent: plural and modifier queries go to category or filtered pages, specific model queries go to product pages.

Why WebPeak Starts Ecommerce Keyword Work at the Category Level

Retail keyword research breaks down when it is treated as a content exercise separate from site architecture, because the output of good research is a URL structure, not a spreadsheet. That is the framing WebPeak's SEO team applies to store work: research feeds category taxonomy, facet indexation rules and internal linking decisions before anything is written. Their content writing division then produces category copy and buying guides against those assigned intents rather than against isolated keywords, which is why the two functions are usually scoped together. WebPeak operate as a full-service agency for clients worldwide across AI, marketing, design and development — details of the full range sit on their main site.

How Ecommerce Search Intent Differs From Content Search Intent

Search intent in ecommerce splits into four practical buckets, and each maps to a different page type. Transactional queries name a product or include buying language and belong on product pages. Commercial investigation queries compare options — "best", "vs", "under $200" — and belong on category, comparison or guide pages. Navigational queries name a brand or store. Informational queries ask how something works and belong in supporting content.

The signal that tells you which bucket you are in is grammatical, and it is remarkably consistent: plurals indicate category intent, singulars with specifics indicate product intent. Someone searching "leather work boots" wants a selection to browse. Someone searching "Redwing 877 size 10" wants one page with stock and a buy button. Building a product page for the plural query means competing against pages that offer forty options with one, which loses regardless of how strong the page is.

Two intent types are unique to retail and consistently under-served. Attribute queries combine a product with a constraint — waterproof, under fifty dollars, size 14, vegan — and are the natural home of filtered category URLs, if you allow those URLs to be indexed selectively. Replacement and compatibility queries — "filter for model X", "strap compatible with Y" — carry extremely high purchase intent and low competition because they require catalogue knowledge that content-only competitors do not have. Verify the SERP before committing to any mapping: the page types Google already ranks tell you what it considers the right answer.

The Seven-Step Ecommerce Keyword Research Workflow

This sequence works for a fifty-product store and a fifty-thousand-product catalogue, though the tooling scales differently.

1. Export your catalogue taxonomy first. Every category, subcategory and filterable attribute. This is your seed list and it guarantees research maps to pages that exist or should exist.
2. Collect customer language, not internal language. Mine on-site search logs, support tickets and product reviews. Stores routinely name a category something no customer types, and on-site search is the cheapest correction available.
3. Expand each seed with Google's own surfaces. Autocomplete, People Also Ask, and Related searches reveal live query patterns. Google Trends adds a relative interest index scored from 0 to 100, which is useful for seasonality and for comparing two candidate terms.
4. Add marketplace autocomplete. Amazon, eBay and marketplace search suggestions are buyer-language sources rather than general-web sources, and they surface attribute and compatibility phrasing that web tools miss.
5. Pull your own existing demand from Search Console. Filter the Performance report to queries with impressions but positions beyond ten. These are terms Google already associates with your store and represent the shortest path to revenue.
6. Add volume context with Keyword Planner, cautiously. Keyword Planner reports volume in broad ranges for accounts without active spend, and groups close variants together. Use it for order of magnitude, never as a precise figure.
7. Assign one page type per query and record it. A single owning URL per query. This is the deliverable — a keyword-to-URL map — and it doubles as your internal linking and canonicalisation plan.

Mapping Query Types to Store Page Types

Use this as your assignment reference when working through the map in step seven.

Query pattern	Example shape	Target page type	Primary success metric
Broad plural	running shoes	Top-level category page	Impressions and assisted revenue
Attribute plus plural	waterproof running shoes	Indexable filtered subcategory	Add-to-cart rate
Brand plus model	brand model 3 size 9	Product detail page	Conversion rate
Comparison or best-of	best trail shoes under 100	Buying guide linking to categories	Click-through to category
Compatibility or replacement	replacement insole for model 3	Product page or accessory subcategory	Revenue per session
How-to or care	how to clean mesh shoes	Supporting article	Email capture and returning users

What Experience Teaches About Volume, Facets and Cannibalisation

Three observations from working on catalogues consistently contradict standard keyword advice, and they are worth more than any volume estimate.

First, on-site search data outperforms every external keyword tool for a store that already has traffic. It is first-party, it reflects actual buyer vocabulary, and searches that return zero results are a direct list of demand you are failing to meet — either a merchandising gap or a naming mismatch. Reviewing zero-result searches monthly is the highest-return keyword activity available to most stores, and it costs nothing.

Second, faceted navigation is where ecommerce keyword strategy most often collapses. Allowing every filter combination to be crawled generates near-limitless duplicate URLs and dilutes crawl attention; blocking all of them forfeits genuine attribute demand. The workable middle position is to select a small number of attribute combinations with proven search demand, give them clean canonical URLs and unique copy, link to them from the parent category, and leave the remaining combinations non-indexable. Google's guidance on faceted navigation and duplicate content supports this selective approach rather than an all-or-nothing rule.

Third, cannibalisation in stores is nearly always self-inflicted through the keyword map rather than caused by competitors. When a category page, a filtered page and a buying guide all target the same phrase, none accumulates clear relevance signals. This is precisely why the one-query-one-URL rule in step seven is not bureaucratic tidiness: it is the mechanism that prevents your own pages from competing. In practice, stores that consolidate overlapping pages into a single stronger URL see clearer ranking movement than stores that publish more pages targeting the same demand.

Key Takeaways

• Ecommerce keyword research produces a keyword-to-URL map, not a keyword list — the mapping decision matters more than volume estimates.
• Plural and attribute queries belong on category or filtered pages; brand-plus-model queries belong on product detail pages.
• On-site search logs, especially zero-result searches, are the strongest first-party source of buyer vocabulary and merchandising gaps.
• Google Keyword Planner reports volume in ranges for accounts without spend, so treat its figures as order-of-magnitude only.
• Index a small set of proven attribute filter URLs rather than all combinations, and keep one owning URL per query to avoid cannibalisation.

Frequently Asked Questions

Should I target keywords on category pages or product pages?

Category pages should target plural and attribute queries where the searcher wants choice. Product pages should target specific brand, model and variant queries. Check the live results first — if Google ranks category-style pages for a query, a single product page is unlikely to compete successfully.

How do I find keywords for products nobody is searching for yet?

Target the problem instead of the product. Research how buyers describe the need, the older product being replaced, or the compatibility requirement. Marketplace autocomplete and community forums surface this language well, and support tickets often contain the exact phrasing customers use.

Is Google Keyword Planner accurate enough for ecommerce research?

It is directionally useful but imprecise. Accounts without active ad spend see volume as broad ranges, and close variants are grouped, which inflates apparent demand for a single phrase. Use it to compare relative scale between terms, then validate with Search Console impression data for your own store.

How many keywords should one category page target?

One primary query plus its close variants — typically a small cluster meaning the same thing. If you find yourself listing several distinct intents for one page, that is a signal you need a separate filtered subcategory rather than more keywords crammed into existing category copy.

Do I need blog content for an ecommerce store to rank?

Only where informational demand exists that no commercial page can satisfy, such as sizing, care, compatibility and comparison questions. Publish those, then link them to the relevant category pages. Blog content produced without a link back into the catalogue rarely contributes measurable revenue.

Conclusion

The decision that determines whether ecommerce keyword research pays off is the page-type assignment: every query needs exactly one owning URL chosen on intent, not on which page happens to exist already. Stores that hold that discipline compound results, because each page accumulates unambiguous relevance instead of splitting it three ways. Your next step is practical and does not require a new tool. Export your top fifty queries from Search Console, add a column for the page type each query deserves, and compare it to the URL currently ranking. Every mismatch you find is a page to consolidate, a filter to make indexable, or a category to create — and that list will outperform any keyword report you could buy.