New format
Keywords for an online store: research and filtering
Store semantics are built around the catalog: categories, filters, cards, and informational articles. Demand and clusters first — then copy and structure, not the other way around.
Below: where to get keywords, how to clean the list, and how to map queries to page types. General core theory is in the semantic-core article; this focuses on e-commerce.
Keyword research and landing prep take weeks. Growing a commercial query core in the results is planned over 2–6 months after promotion starts.
Why a store needs its own core
Without semantics the catalog grows “as purchasing prefers,” not as buyers search. Categories miss demand, cards lack tails, filters junk the index.
The core links assortment to URLs: what to promote first, which texts to write, where to send ads.
What a working store core gives:
- priority of categories and brands
- briefs for cards and filters
- junk-query negatives
- a shared map with ads and SEO
Where to find keywords
Markers: product names, categories, brands, synonyms, problems (“cheap,” “with installation”). Expand with Wordstat or a keyword planner, search suggestions, competitor exports as ideas — not as a final list.
On-site search and ads reports often give live customer wording.
Sources:
- Yandex Wordstat (base and refined frequency)
- suggestions and “related queries”
- Key Collector / peers for mass work
- Webmaster: search queries
- internal search and support
Filtering junk
Drop off-target: another city, a competitor brand you do not carry, informational “what is” if you have no blog for it, non-purchase queries if the goal is sales now.
Keep commercial markers and long-tail with a clear landing. Do not store in the core what you will not build a URL for.
Typical store junk:
- jobs and “download price list” without need
- competitor-brand queries outside your matrix
- kids/adult overlaps from the wrong niche
- typo duplicates with no traffic
Clusters: category, filter, card
High- and mid-frequency category queries go to listings. Modifiers (color, size, “with delivery”) go to filters or landings if demand and assortment support them. Exact models go to cards.
Do not spawn a thousand nearly empty filter URLs for every checkbox. Facet indexing policy is a separate decision.
Mapping rule:
- one main intent — one priority page
- a card does not compete with a category for the same head term
- brand + product has its own section logic
- informational queries go to the blog, not the card
How to apply the core in work
The core is a living file: priority, frequency, URL, status “page exists / needed / in progress.” From it you write Title, H1, category copy, and briefs for writers.
Ads take commercial clusters with negatives; SEO strengthens landings and snippets.
Monthly ritual:
- new products → new markers
- delisted items → redirects/cleanup
- Webmaster queries → into the core or negatives
- revisit top priorities by margin
What to avoid
Copying a competitor’s core, pages for every long-tail without stock, keyword stuffing on cards, expecting page one right after uploading Titles.
Stronger than “secret parsers”: clean structure, stock, price/availability, decent photos, and commercial factors.
Red flags:
- a 50k-row core with no clusters
- filters in the index with no content
- one text for all categories with a word swap
- bids only on head keywords with no long-tail matrix
FAQ
Where to start: products or categories?
From demand markers: what buyers search. Then decide whether you need a separate category, a filter, or cards are enough.
How is this different from a regular keyword core?
More SKUs, commercial tails (“buy,” “price,” “with delivery”), plus the risk of thin filter pages.
Can you copy a competitor’s keywords?
Studying competitors helps. Copying the whole core — no: different structure, assortment, and region.
Is Wordstat required?
For RU markets it is a handy start. Add search suggestions, Webmaster reports, ads data, and the store’s internal search. Elsewhere, planners and Search Console peers fill the same role.
Do you need Key Collector?
It speeds mass collection and cleanup. You can start manually; on large catalogs software saves time.
What about zero frequency?
Some long-tail still brings traffic. Do not spawn empty pages for every rare form.
Are Ads and SEO keywords one list?
They overlap. Ads lean harder on negatives and commercial intent; SEO on landings and content.
When should product queries start ranking?
After catalog prep — ranking growth is planned over months, not days.
Store keyword dump spawning empty filter URLs in the index?
We’ll cluster demand into category / filter / card — prep first, TOP planned over 2–6 months.
Discuss the task
