Ecommerce Keyword Research

Ecommerce keyword research runs the standard process through a commerce lens, and the lens changes everything downstream: queries sort by where they land (category, product, or guide — the mapping decision that prevents most store cannibalization), buying modifiers carry the margin, and the demand data has a second source nobody else gets — your own search box and sales history. Here's the store-specific method.

The three demand pools, and where each lands

1. Category-shaped demand — "[product type]", "best [type]", "[type] for [use]": the volume core, landing on category pages (or comparison pages where the SERP says listicle — read every money term's SERP per the intent method before assigning).
2. Product-shaped demand — "[brand] [model]", "[exact product] review/price/vs": lower volume, highest intent, landing on product pages — and largely winnable on template quality alone, since competitors' product pages are boilerplate.
3. Question-shaped demand — "how to choose", "what size", "is [x] worth it", "how to [task the product serves]": the research-phase pool landing on guide content that routes buyers (and links) toward the first two pools. Stores that skip this pool cede the entire consideration phase to publishers and competitors.

The commerce-specific expansion sources

Beyond the standard tooling: your site-search logs — the highest-signal keyword source in retail (buyers typing exactly what they want, in their words: misspellings, synonyms, attribute language — feed it into naming, categories and content); autocomplete on Amazon and marketplace search (commerce-intent completions that web tools underweight); your sales and support data (what people call things on the phone becomes what they type); competitor taxonomies (their category trees are keyword maps someone already validated — the export workflow plus reading their nav); and attribute × category matrices — systematically checking which combinations ("waterproof + [category]", "[material] + [type]") carry real volume, which is how subcategory decisions and facet-indexing exceptions get made on evidence.

Prioritising by money, not volume

The commerce twist on the standard triad: weight demand by margin and conversion likelihood, not raw volume — "buy [specific product]" at 200 searches out-earns "[generic category] ideas" at 5,000; modifier language grades intent (buy/price/deal/near-me high; ideas/DIY/free low — sometimes negative for a premium store); and seasonality gets planned on SEO's lead time (peak-period pages built quarters early). The output is the store's map with a landing-type column — every keyword-group assigned to category, product, or guide before anything gets built, which is the single discipline that prevents the taxonomy-vs-blog cannibalization every store audit finds.

Frequently asked questions

How do I research keywords for products nobody searches by name?

Unbranded catalogue means category and question pools carry everything: rank the categories for the type-demand, own the question layer ("how to choose [type]"), and let products catch long-tail attribute queries via template quality. Product-name demand arrives later as the brand grows — it's earned, not found.

Should I target "cheap/discount" modifiers if I'm not the cheapest?

Only with a page that honestly serves the intent (a sale section, an outlet line) — ranking premium products on "cheap" queries buys bounces, per the satisfaction physics. Price-modifier demand belongs to stores whose offer matches it; yours may be "best/quality/durable" territory, which converts better at your margins anyway.

How often should store keyword research refresh?

The map annually, site-search logs quarterly (they drift with catalogue and fashion faster than external tools see), and per-launch for new lines — each new category enters through the full pool-and-assignment process. The research names the fights; winning them stays the usual pairing — pages built right and authority behind them (supplied here).