Ecommerce SEO: What Delivers Results and What Doesn't

In an online store 80% of organic traffic lands on 5% of the pages — and those are almost always categories and filtered selections, not product cards and not the blog. Understanding this proportion saves months of work: instead of writing unique descriptions for 4,000 items, you put forty landing pages in order.

Catalog structure is the main asset

Every level of nesting has to match real demand. "Tires" → "Winter tires" → "Studded winter tires R17" — three landing pages for three frequency clusters. The mistake we see constantly: categories built around internal warehouse logic ("New arrivals", "Bestsellers", "Sale") rather than around how people search. Sections like that bring no traffic at all. How to assemble clusters for a structure is covered in the article on the semantic core.

Filters: a source of traffic and a source of disaster

Filtered pages like /shiny/zimnie/r17/ deliver up to half of a store's long-tail traffic. But you cannot open every filter combination to indexing: on a mid-sized catalog that instantly spawns tens of thousands of near-identical URLs. The working rule: open by hand the combinations that have search volume above 30 a month and at least 5 products in the listing; close everything else with canonical and noindex. Google Search Central helps you verify that the crawler sees exactly what it should.

Product cards: where the reasonable limit lies

• Unique text is only needed on cards that have their own search demand for the model
• For the rest, full specifications, price, availability and Product microdata markup are enough
• Reviews give you both unique content and higher conversion — the most profitable work on a card
• Do not delete out-of-stock products: return 200, show alternatives, otherwise you lose the positions you have built up

What almost never works

Buying links to product cards in bulk, rewriting manufacturer descriptions to hit 90% uniqueness, a blog about nothing with posts like "5 tips for choosing". Twice we have seen projects where more than 400 thousand roubles went into generating product card texts and the traffic gain fell within statistical noise. The same money put into speed and structure produces a measurable result — the piece on load speed shows how the two are connected.

Technical hygiene for a store

Pagination with a correct self-referencing canonical, sorting closed from indexing, site search closed entirely, a sitemap split into files by section and updated automatically. We covered the typical errors in these files in a separate text on the sitemap and robots.txt. Check the speed of a category with 60 products separately — that is usually the slowest page of all.

How long it takes

A realistic schedule for a store with 3,000–5,000 products: month one is the audit and technical cleanup, months two and three are rebuilding the structure and opening up filters, from month four it is content and links. The first noticeable shifts on long-tail queries show up after 2–3 months, on mid-range categories after 5–7. If a contractor promises the top for "buy tires" in two months, that conversation is not about SEO.

What to do in the first week

• Export every indexed URL and count the share of junk — it is usually sobering
• Take the top 20 categories by revenue and check their positions on the main queries
• Check whether the filters that already bring traffic are open to indexing
• Set up ecommerce tracking in Metrica so you measure revenue from organic, not visits

A store is first of all a technical product, and SEO here is inseparable from development. We usually run both parts with one team: how the process works is visible in the development section and in the examples in our portfolio.