Ecommerce keyword research fails for a boring reason: people research keywords, then go looking for somewhere to put them. On a store, the page already exists. You have a category tree, a product catalog, and maybe a blog. The job is not finding 3,000 terms. The job is deciding which page owns which term, and which pages you are missing entirely.
Get that mapping right and the rest is execution. Get it wrong and you end up with a blog post competing against your own category page for the same query, both ranking on page three, neither converting.
Start With Your Catalog, Not a Seed List
Export every category, subcategory, and product name you sell. That export is your first keyword list, and it is better than anything a tool will hand you cold, because it is grounded in inventory you can actually ship.
Run those names through a keyword explorer to attach volume, difficulty, and CPC. What you are looking for first is the vocabulary mismatch — the places where your merchandising team named something differently from how buyers search. A store selling “performance outerwear” while 40,000 people a month search “waterproof jacket” is not losing to competitors. It is losing to its own naming convention.
SEO Rocket’s keyword explorer takes multiple seeds at once and returns up to 150 ideas per search with volume, difficulty, CPC, global volume, and SERP features, against a country-specific index. For a store, that multi-seed behavior matters: you want to feed in twenty category names in one pass, not run twenty separate searches and stitch the exports together.
Sort Every Term by the Page Type It Wants
This is the step that separates keyword research for ecommerce from keyword research for a content site. Search intent on a store splits into four buckets, and each maps to exactly one page type.
- Product terms — a model number, a brand plus a specific item, “buy X.” These belong on the product page. Nothing else should target them.
- Category terms — “running shoes,” “leather sofas,” “wireless earbuds.” These belong on a category page. They are usually the highest-volume, highest-difficulty terms you will chase.
- Modified category terms — “waterproof running shoes for wide feet,” “grey leather sofa under $1000.” These belong on a filtered subcategory or facet page, and they are where most stores leave the largest amount of money on the table.
- Informational terms — “how to break in running shoes,” “leather vs fabric sofa.” These belong on the blog, and they convert badly on a first visit. Track them on assisted conversions or do not track them at all.
Check the SERP before you finalize a mapping. If Google is showing category listings for a term you assumed was informational, Google has told you what page type wins. Argue with it and you lose.
Modified Category Terms Are Where the Volume Hides
Head terms are brutal. “Running shoes” is owned by brands with eight-figure link budgets, and a new store will not rank for it in year one. The winnable volume sits in the modifier tail, and there is far more of it than the head.
Build modifier lists per category and multiply them out:
- Attribute — color, material, size, capacity, waterproof, cordless, organic.
- Audience — for men, for kids, for beginners, for small kitchens, for wide feet.
- Price — cheap, budget, under $50, luxury, premium.
- Use case — for trail running, for camping, for office use.
- Comparison — best, top rated, alternatives to.
Run the combinations through the explorer, keep anything with real volume and difficulty you can plausibly clear, and check whether your faceted navigation already generates that page. Usually it does — it is just noindexed or blocked in robots.txt because someone once panicked about crawl budget. Selectively indexing high-intent facets with clean URLs, self-referencing canonicals, and unique intro copy is often the single highest-return technical change on an established store.
Mine Competitors for the Categories You Do Not Have
Your competitors have already paid for this research. A content gap analysis across three to five rivals shows terms they rank for and you do not, with per-competitor position columns so you can see whether one outlier is skewing the list or all five are beating you.
Read the gap output as a merchandising document, not a keyword list. Clusters of related missing terms almost always mean a missing category. If four competitors rank for a dozen variations of “compression socks” and you sell socks but have no compression subcategory, that is a page to build and possibly a product line to stock.
Also pull their top pages. On ecommerce sites, the traffic distribution is lopsided in a way that is instantly diagnostic — if a rival’s top ten pages are all subcategories, they have won on facets, and you can copy the structure without copying the content.
Judge Difficulty Against the Weakest Page-One Result
Difficulty scores are estimates. Treat them as a first-pass filter, then open the actual SERP for anything you are seriously considering and look at the weakest result on page one — position eight, nine, or ten. That is your real competition, because that is the page you have to displace to get a foothold.
If position nine is a thin affiliate roundup with four referring domains, a well-built category page with genuine product depth can take it inside a quarter. If position nine is a major retailer with 900 referring domains to that exact URL, pick a different term. Benchmarking against medians makes every keyword look impossible; benchmarking against the weakest page-one competitor tells you the truth.
Be honest about the data itself. Third-party volume figures are modeled roughly twelve-month averages, which means seasonal ecommerce terms are systematically misrepresented — a Christmas term showing 8,000 a month may be 60,000 in November and 400 in June. Cross-check anything seasonal against Search Console history or Google Trends before you build a quarter’s roadmap on it.
Turn the List Into a Prioritized Build Queue
A spreadsheet of 2,000 terms is not a plan. Score each mapped page on three inputs and sort:
- Commercial value — CPC as a proxy for how much the click is worth, multiplied by your margin on that category.
- Attainability — weakest page-one competitor strength versus your current authority.
- Effort — does the page exist, does it need copy, or does it need engineering work on faceted navigation?
Work the top of that list, ten to twenty pages at a time, and re-score quarterly. Save the final term set to a project keyword pool so the same list feeds your writer and your rank tracker instead of living in three drifting spreadsheets — that is how SEO Rocket handles it, and it removes the most common failure mode in ecommerce SEO keyword research, which is that the research and the tracking slowly stop describing the same site.
Measure the Right Thing After Launch
Give a new or rewritten category page eight to twelve weeks before judging it. Watch non-brand impressions in Search Console first — they move weeks before clicks do — then top-10 keyword counts, then revenue. Daily position swings of two or three are normal noise on any competitive commercial term; if you react to them you will rewrite pages that were working.
The end state you want is simple: every commercially valuable query in your niche has exactly one page on your site that deserves to rank for it, and you know which page that is. Most stores never get there because they treat how to do keyword research for ecommerce as a discovery exercise. It is an assignment exercise. Do the assignment properly and the discovery mostly takes care of itself.
