Ecommerce SEO

Two things decide whether a store ranks: whether category pages can compete for the commercial terms, and whether the crawler is spending its time on real pages instead of filter combinations.

Everything else is downstream of those. A store with beautiful product pages and an unmanaged faceted navigation will lose to a competitor with mediocre products and a clean URL space.

The hierarchy that works

Home
 └── Category            /shoes/                    ← targets "shoes"
      └── Subcategory    /shoes/running/            ← targets "running shoes"
           └── Filtered  /shoes/running/waterproof/ ← targets "waterproof running shoes"
                └── Product /p/brand-model-gtx      ← targets the model name

Category pages carry the commercial head terms and the head terms carry the revenue. Most stores under-invest in them because they look like grids of products rather than pages.

• Give category pages real content. Not 300 words of filler under the fold. A short, genuinely useful intro that helps someone choose, plus buying guidance, plus the questions people ask about this category, placed where a human would actually read them.
• Make the top of the page the products. Content below the grid, not above it. Pushing the grid down to fit a keyword paragraph costs conversions and the paragraph wasn't helping anyway.
• Promote high-demand filters to real, indexable subcategories with their own URL, title, intro and internal links. "Waterproof running shoes" is a search term with buyers behind it. Leaving it as ?filter=waterproof gives it to a competitor.

Faceted navigation, the crawl-budget killer

A store with 8 filters averaging 5 values each generates a combinatorial URL space in the hundreds of thousands. Left unmanaged, the crawler spends its entire budget there and never gets to the products.

Decide for every facet which of three buckets it belongs in:

Rules that save you later:

• Keep indexable facets in the path, non-indexable ones in query strings. It makes the robots.txt rules trivial.
• Fix the parameter order. ?color=red&size=10 and ?size=10&color=red are two URLs for one page unless you normalise.
• Cap the depth. Two facets combined is usually the limit worth indexing. Three is almost never a real search term.
• Don't index empty filter results. A page saying "no products match" is a soft 404 at scale.

Product pages

• Never ship the manufacturer's description unchanged. Every competitor has the identical text. Write something, even briefly, and add what the manufacturer can't: how it fits, what it's like in use, who it's wrong for.
• Handle variants deliberately. One canonical product URL with selectable options is usually right, with ProductGroup and hasVariant schema. Separate URLs per variant are correct only when people search for the variant specifically, and then each needs unique content.
• Reviews on the page are content, and they're the content competitors can't copy. Surface them, mark them up honestly, don't fabricate them.
• Answer the pre-purchase questions on the page: sizing, materials, compatibility, delivery, returns. These are the queries that bring people in, and the answers are usually buried in a tab or a PDF.
• Product schema with real Offer data, including shipping and return policy. See schema-markup.
• Images with real alt text, multiple angles, and a filename that isn't IMG_4471.jpg. See image-seo.

Out of stock and discontinued

Getting this wrong quietly deletes rankings. The correct action depends on whether it's coming back.

Never 404 a product page that has backlinks or traffic. Never bulk-redirect discontinued products to the homepage, which is a soft 404 pattern and Google treats it as one.

Crawl budget on large stores

At tens of thousands of URLs, what gets crawled becomes the constraint. In order of effect:

1. Cut the URL space. Facet rules above. This is 80% of the problem.
2. Fix redirect chains in navigation and internal links.
3. Keep the sitemap clean — canonical, indexable, 200-status URLs only, split into files by type so index coverage is diagnosable per segment.
4. Return proper status codes fast. Slow responses reduce crawl rate directly.
5. Prune dead products that will never come back and have no links.
6. Check the logs to see where crawl budget actually goes. It's usually somewhere surprising.

AI shopping surfaces

Assistants increasingly answer "best X for Y under Z" directly. What gets a product into those answers:

• Facts stated as text on the page — price, dimensions, materials, compatibility, what's in the box. Specs hidden in an image or a tab loaded on click are invisible.
• Comparison content the store owns, answering "X vs Y" and "best X for Y" honestly, including where the product isn't the right choice. Stores that only publish sales copy don't get cited. See comparison-alternatives-pages.
• A product feed in Merchant Center, which feeds Google's shopping surfaces including the AI ones.
• Third-party corroboration — real reviews on platforms outside your control, which is what a model weights when your own site says you're the best.
• Crawler access. Check that the CDN's bot rules aren't blocking retrieval agents, since ecommerce sites run the most aggressive bot protection and frequently block them by accident. See ai-crawler-policy.

Don't do these

• Don't put product content behind a "read more" that fetches on click. It won't be seen.
• Don't index internal search results. Classic scaled-content problem, and Google's guidelines call it out.
• Don't canonicalise a paginated series to page one. Each page should self-canonicalise. Page 2 isn't a duplicate of page 1.
• Don't block filter URLs in robots.txt and also canonicalise them. Blocked pages can't be crawled, so the canonical is never read.
• Don't write 800 words of keyword text above the product grid. It hurts conversion and does nothing for ranking.
• Don't let the CMS generate a page per colour with identical copy.

Platform notes

Shopify and WordPress each have their own version of these problems, with platform-specific fixes. See shopify-seo and wordpress-seo.

Where the data comes from

Category-level demand, which facets have search volume behind them and what competitors rank for at category level are the inputs that decide the structure, and guessing them produces the wrong architecture. Search Console shows what you already get impressions for, which is the cheapest place to find promotable facets.

Keupera covers the category and competitor side with keyword research, gap analysis and competitor SERP data, and connects directly to Shopify and other stores for publishing, over MCP:

claude mcp add --transport http keupera https://mcp.keupera.com/mcp -s user

Related skills

canonical-duplicate-content, schema-markup, shopify-seo, image-seo, internal-linking-architecture, comparison-alternatives-pages, technical-seo-audit