E-commerce SEO
Keyword Research for Ecommerce That Decides Which Pages Your Store Needs
For an online store, keyword research is really site architecture research. It tells you which categories to build, which filter combinations deserve their own landing page, and which searches belong to a product rather than a collection.
- years in SEO
- 6+
- keywords researched
- 100,000+
- person doing the work
- 1
Keyword research for ecommerce is store architecture in disguise
Keyword research for ecommerce finds the words shoppers use for what you sell and turns them into a map of categories, filter pages, products and guides, each with one job. On a service site, keyword research mostly decides what to write. On a store, it decides what to build. Every category, subcategory and curated collection is a bet that enough people search for that grouping of products to justify a page, and every filter is a choice about whether a combination of attributes should be visible to search engines or not.
Most stores inherit their category structure from the warehouse or the supplier catalogue. That is why I see menus organised by brand codes or internal product lines while customers search by use, style or recipient. "Men's waterproof walking boots" is a category in the searcher's head. If your site files those boots under "Footwear > Outdoor > Range B", there is nothing for that query to land on.
Research also has to deal with volume that is spread thinly. A large catalogue can have thousands of product and model searches, each tiny on its own, alongside a handful of competitive category terms. Both matter, but they need different pages, and confusing them is how category pages and product pages end up competing.
Key takeaways
- Store searches split into category, product or model, comparison, gift and navigational intent, and each one has a different winning page type.
- Category pages usually own the most valuable commercial queries in e-commerce; the research decides how many categories and subcategories the demand supports.
- A filter combination earns its own indexable landing page only when real search demand exists for it and you stock enough products to fill it.
- Volume is weighed against margin, stock depth, seasonality and who already owns the results page, because marketplaces dominate some queries outright.
- Vocabulary changes by market: trainers vs sneakers, delivery vs shipping, click and collect vs curbside pickup. Research per market, not per language.
What's included in my keyword research for ecommerce
Six outputs, all feeding one map. The modifier work comes first because it is where most of the winnable demand on a store hides.
Modifier mining: the attribute, audience, use and price searches beneath each head term
Head terms usually belong to marketplaces and big brands; the opportunity sits in the modifiers around them. Attribute modifiers ("black leather ankle boots", "wide fit", "king size") are the filter combinations that might deserve landing pages. Audience modifiers ("women's", "kids'", "for small spaces") are often the first split your category tree is missing. Use and occasion modifiers show groupings the warehouse never thought of, and price modifiers ("under £50", "luxury") are only worth chasing if your range can honestly serve them.
- Modifier families pulled for every core range
- Demand per modifier, not just per head term
- Flags where audience or use splits are missing from the menu
- Price modifiers checked against your actual price points
Model, part number and compatibility searches for the product pages in your catalogue
Exact model names, part numbers, "compatible with" searches and replacement parts are tiny one by one and convert very well together. I size this long tail from your own Search Console data, Shopping search terms and site search logs, then check whether product pages carry the specs, codes and compatibility details these shoppers type.
- Model and part-number demand sized across the catalogue
- Compatibility and replacement queries grouped by product line
- Gaps where product pages lack the specs people search for
A seasonal and gift calendar so peak pages for your shop are indexed in time
"Gifts for him", Mother's Day, Christmas jumpers and back to school are predictable peaks. I chart the demand curve for every seasonal cluster, judge its value on the peak months rather than the annual average, and set a date by which each page has to be live so it is crawled and indexed before shoppers arrive.
- Demand curves for every seasonal and gift cluster
- Go-live date per peak-season page
- Gift collections planned by recipient and budget
A category tree rebuilt around how customers search, not how stock is filed
Most store menus mirror the supplier catalogue or the warehouse. I compare your current tree with the clusters the research found and recommend which categories to add, which to merge because they split the same demand, and which to retire because nobody searches for them. Stock depth is part of the call: a category with three products is usually a filter, not a page.
- Categories to add, merge or retire, with the reasoning
- Subcategory splits backed by search demand
- Stock depth checked before any new page is proposed
Which filter combinations earn indexable landing pages on an online store
A combination of attributes earns its own URL only when three things line up: real search demand, a results page that shows category-style pages, and enough stock to fill it. I list the combinations that pass, with the URL, title and owning cluster for each, and mark the rest as filters that stay out of the index so technical SEO for e-commerce can enforce the rule.
- Approved filter combinations with proposed URLs
- Non-indexable filters listed for the developer
- Stock count against each proposed landing page
The keyword-to-URL map that settles cannibalisation across categories, products and guides
Every cluster gets one owning URL, an intent label, a margin weighting and a seasonality flag, and the map records what happens to pages that compete for the same terms: merge, redirect, canonicalise or reword. If you sell into more than one country, each market gets its own vocabulary notes, because translated prices do not translate the words people use.
- One owner per cluster, with intent and margin weighting
- Cannibalisation actions for competing URLs
- Vocabulary notes for each market you sell into
- A build order sequenced by season and priority
Experience
Keyword Research from someone who has done it many times
You work with me from the first call to the last report. No account managers, no hand-offs to a junior team.
Book a free call- years in SEO
- 6+
- client websites worked on
- 60+
- keywords researched
- 100,000+
- documented case studies
- 35
More E-commerce SEO
The other disciplines, written for online stores.
On-Page SEO for e-commerce
Title and H1 patterns that scale across thousands of products, category pages with copy that helps people choose, unique product detail, and Product, Offer and BreadcrumbList markup that matches what the page shows.
Read moreTechnical SEO for e-commerce
Faceted navigation, sort and filter parameters, pagination, internal search pages, variants and out-of-stock handling. Plus Shopify collection-path duplicates, app script weight and redirect plans for replatforming.
Read moreContent Strategy for e-commerce
Buying guides, comparisons, gift guides and category copy planned around the seasonal calendar and built to feed revenue pages. Manufacturer descriptions replaced where it pays, and dead blog posts pruned.
Read moreLocal SEO for e-commerce
For retailers with shops: store pages that are not doorway pages, Google Business Profile categories and attributes, local inventory in Merchant Center, and click and collect searches that online-only rivals cannot win.
Read moreOff-Page SEO for e-commerce
Product-led PR, gift guides, stockist pages, product and store reviews within the fake review rules, and disclosed creator content. Authority routed to category pages, where it moves revenue.
Read moreShopper intent classes and the page each one needs
I label every cluster with one of these intents. Transactional splits in two for stores, because a category searcher and a model-number searcher want very different pages.
| Intent | Example queries | What the shopper needs | Page type that wins | Conversion element |
|---|---|---|---|---|
| Informational | how to clean suede trainers; what size road bike do I need | A straight answer, often with a product as the fix | Care guide, size guide, how-to | Links into the exact products and categories that solve the problem |
| Commercial investigation | best robot vacuum for pet hair; [model A] vs [model B]; [model] review | Help choosing between options | Buying guide, comparison page, product page with reviews | Clear recommendation, comparison table, links to in-stock products |
| Transactional - category | women's trail running shoes; oak dining table; linen duvet cover king size | A range to browse and compare | Category or subcategory page | Strong first row of products, useful filters, delivery promise near the top |
| Transactional - product | [brand] [model] size 10; [model number] replacement filter | One item, a price and a delivery date | Product page with the variant preselected | Price, stock, delivery date and add to basket in the first screen |
| Navigational | [store name] returns; [store name] discount code | Your store specifically | Help pages, returns page, homepage | Fast answer; a genuine offers page so voucher sites do not take the click |
Gift and occasion searches cut across these classes. "Gifts for gardeners" behaves like commercial investigation for most of the year and like a transactional category query in the weeks before Christmas, so it needs its own row in the map.
Scoring clusters on margin, stock and the shape of the results page
Search volume tells you how many people type a phrase. It does not tell you whether you can sell to them profitably, so it is one input among several.
Margin and basket value. A cluster around your highest-margin range is worth more than a bigger cluster around loss-leading products. Where I can, I join keyword clusters to product margin or at least to average order value by category.
Stock depth. A category page needs enough products to be useful. If you stock three items that fit "vegan walking boots", a landing page is likely to look thin next to retailers carrying thirty. Sometimes the right answer is to fix the range before the page.
Seasonality. I check the demand curve for every important cluster. A query with a sharp November peak needs a page that is indexed by early autumn, and its value has to be judged on the peak months, not the annual average.
Who owns the results page. Some queries are dominated by marketplaces, brand sites or comparison publishers. If the first page is all Amazon, eBay and the manufacturer, a small store may get a better return from the long-tail variants of that query than from the head term.
Free review
Get a free Keyword Research review
Send me your site. Before the call I look at your rankings and pages, then tell you what I would fix first and what it is likely worth.
UK and US shoppers search with different words
Stores selling in both markets often translate prices and forget vocabulary. These differences change which phrases each market's category pages should target.
UK shoppers
- "trainers", "jumper", "trousers", "nappies", "duvet"
- "delivery" and "free delivery", "next day delivery"
- "click and collect" for picking up in store
- "sale" and "clearance" peak around Boxing Day and January
- Prices are shown including VAT, so price modifiers match the shelf price
US shoppers
- "sneakers", "sweater", "pants", "diapers", "comforter"
- "shipping" and "free shipping", "2-day shipping"
- "curbside pickup" or "buy online pick up in store"
- Black Friday and Cyber Monday dominate the seasonal calendar
- Sales tax is usually added at checkout, which affects how price modifiers behave
My keyword research process for an online store
Two to four weeks depending on catalogue size. The output is a map your developer, merchandiser and copywriter can all work from.
- 1
Agree commercial priorities
Which ranges carry margin, which you are expanding, which are being phased out, and when your peaks are. This stops research from recommending pages for products you are about to drop.
- 2
Mine your own data
Search Console queries, internal site search terms, Shopping ad search terms and customer service questions. Site search logs are especially useful: they show the words customers use once they are already on your store.
- 3
Expand and check the results pages
Tools for breadth, then a manual review of each important cluster: product grids, free listings, image packs, marketplaces, video, AI answers. The results page tells me which page type Google expects.
- 4
Cluster, then decide categories vs filters
Queries that share results become one cluster with one owning URL. Attribute clusters with real demand and enough stock become indexable landing pages; the rest stay as filters that search engines are told to ignore.
- 5
Score and sequence by season
Clusters are ranked by margin, stock, realistic difficulty and seasonal timing, so the build order puts peak-season pages early enough to be indexed in time.
Where SEO meets CRO
The query sets the first row of products
Intent research should change how a category page is merchandised, not only how it is titled. If people search "waterproof walking boots", the first row of products on that page should be waterproof walking boots, sorted so the best sellers in stock appear first, with the waterproof filter already applied or highlighted. A model-number search should land on the exact variant, selected and in stock, not the default colour. I write these merchandising requirements into the keyword map so the page that ranks is also the page that sells.
A sample keyword-to-URL map for an outdoor footwear store
A simplified extract. The real map covers every category, the attribute pages that earned a URL, and the guides that support them.
| Cluster | Intent | Owning URL | Status | Priority |
|---|---|---|---|---|
| women's walking boots | Transactional - category | /collections/womens-walking-boots | Exists, ranks page 2, thin copy | High - rewrite intro, fix internal links |
| waterproof walking boots women's | Transactional - category | /collections/womens-waterproof-walking-boots | Only exists as a filter URL | High - create indexable landing page |
| walking boots wide fit | Transactional - category | /collections/wide-fit-walking-boots | Missing, eight suitable products | Medium - build once range is confirmed |
| [model A] vs [model B] | Commercial investigation | /blogs/guides/model-a-vs-model-b | Missing | Medium - comparison with own photos |
| how to waterproof leather boots | Informational | /blogs/guides/waterproof-leather-boots | Outdated post, competes with care product page | Low - refresh and link to care products |
| gifts for hikers | Gift / seasonal | /collections/gifts-for-hikers | Missing | High before Q4 - curated collection |
Cannibalisation on store sites
Stores create competing pages without anyone meaning to. A category, a near-identical tag page, an indexed filter URL, a brand collection and a blog post can all target "black leather ankle boots" at once. Google picks one, often not the one you would choose, and the ranking flips between them.
The map resolves this by giving each cluster one owner and recording what happens to the rest: merge, redirect, canonicalise to the owner, or reword to target a different cluster. On Shopify, tag-filtered collection URLs and duplicated collections are common causes. On WooCommerce, product tags and categories with the same names compete. Both are visible in Search Console once you look at which URLs are taking impressions for each query.
Where the keyword map goes next
The map is the brief for everything that follows. It sets the title and H1 patterns in on-page SEO for e-commerce and the filter rules that technical SEO for online stores has to enforce. Comparison and gift clusters become the plan for content strategy for online stores.
Stock-checking and "near me" clusters feed local SEO for retailers if you have shops. For how the disciplines fit together, see SEO for online stores.
Growth from real projects
Learn more about keyword research
Keyword Research in other industries
Keyword Research for Ecommerce FAQs
Should we target category keywords or product keywords first?
Usually category keywords, because they carry the most commercial demand and a category page can rank for many related phrases at once. Product and model keywords matter too, but they are mostly won by getting product pages technically sound and specific. The exception is a store with a small number of hero products that people search for by name.
How do you decide which filters get their own landing page?
Three tests. There must be real search demand for the combination, the results page must show category-style pages ranking for it, and you must stock enough products to make the page useful. Combinations that pass get a clean URL, unique title and copy. Everything else stays a filter that search engines are steered away from.
Is internal site search data useful in keyword research for ecommerce?
Very. It shows the words customers use once they are already on your store, including product types and attributes you do not have categories for. Searches that return no results are particularly useful: they show missing ranges or missing synonyms on existing pages.
Can we compete with Amazon for product keywords?
For broad head terms, often not directly. For specific categories, attribute combinations, compatibility searches and comparison queries, independent stores win regularly, especially with better category pages, original photography and real expertise. The research identifies where the results page is realistically open to you.
How far ahead should seasonal keyword research be done?
Several months before the peak. Pages need to be built, crawled, indexed and linked internally before demand rises, and gift guide outreach to journalists happens even earlier. For Christmas, I aim to have the research done by late summer. Spring and summer peaks need the same lead time.
Do we need separate keyword research for each country we sell to?
Yes, if the vocabulary or the competition differs, which it usually does between the UK and the US. Product names, sizing terms, delivery language and seasonal events all change. Translating one market's keyword list into another market's currency misses most of that.
Book keyword research for ecommerce and see which categories you lack
Send me your store URL and your main ranges. On a free 30-minute call I will show you where category demand is going to competitors, which filter pages are worth indexing, and what I would build first.