Ecommerce SEO: Categories, Facets and AI Shopping

Ecommerce SEO is the work of getting an online store’s category and product pages crawled, indexed and chosen: in Google’s organic results, in its shopping surfaces, and now in the product carousels that AI assistants like ChatGPT and Google AI Mode build. The difference from ordinary SEO is scale. A store is a catalog rendered through a handful of templates, so one bad decision about filters or variants repeats itself ten thousand times.

This guide is for store owners, ecommerce managers and developers who already know the basics of titles and keywords. It covers the page jobs, faceted navigation, product data, and how AI shopping answers choose products and merchants. If your store runs on Shopify, the platform-specific quirks are in our Shopify SEO guide.

On this page

• Category pages vs product pages: two different jobs
• How should faceted navigation be handled?
• Variants, duplicates and canonical URLs
• Product and Offer data: merchant listings vs product snippets
• How do AI shopping answers choose products?
• Content that earns links and citations for a store
• Where ecommerce SEO goes next

Category pages vs product pages: two different jobs

Category pages win the broad, comparison-stage queries and product pages win the specific ones. Most store SEO problems start when those jobs blur.

Someone searching “trail running shoes women” wants to see a range, filter it and compare prices. A category page answers that. Someone searching “Hoka Speedgoat 6 women’s 8.5” wants one item, in stock, at a price. That is a product page. When a store writes long keyword copy on every product page to chase the broad term, or leaves category pages as a bare grid with no text, the two page types either compete or neither wins.

Give each template a clear brief:

• Category pages: a unique H1 that names the product type, a short intro that helps a shopper choose (sizing, use cases, the two or three attributes that matter), crawlable links to every product and to child categories, and a title that matches how people search for the range.
• Product pages: the exact model name and key attribute in the title, a description you wrote rather than the manufacturer’s copy that other stores also use, specifications in HTML text (not only images), real reviews, and accurate price and availability.

The diagram below shows the architecture most catalogs should aim for: a shallow tree where every product is reachable through category links, and filters hang off categories rather than creating a parallel site.

Scroll the diagram sideways to see all of it.

Google’s own ecommerce guides frame this as helping Google “understand what is most important” through navigation. In practice that means category links in the main navigation and breadcrumbs, not products reachable only through the internal search box.

How should faceted navigation be handled?

Faceted navigation should be crawlable only for the filter combinations people actually search for, and blocked or collapsed for everything else. Filters are the single largest source of wasted crawling on most stores.

A category with 8 colors, 12 sizes, 5 brands and 4 price bands can produce thousands of URL combinations, almost all of them near-duplicates. Google’s faceted navigation documentation, last updated in December 2025, describes the cost plainly: crawlers “will typically access a very large number of faceted navigation URLs before the crawlers’ processes determine the URLs are in fact useless.” The time spent there is time not spent on new products.

Google lists the options in rough order of strength:

If you do let some filter pages be indexed, Google asks for standard & separators, a consistent parameter order, and an HTTP 404 when a combination returns no products. A “0 results” page served with a 200 status is a soft 404 waiting to happen. The broader budget question is covered in our crawl budget guide.

The decision tree below is the test we apply to each filter type before deciding how it is exposed.

Scroll the diagram sideways to see all of it.

Variants, duplicates and canonical URLs

Each product should have one canonical URL, and variants should either point to it or have their own URL only when the variant is something people search for by name.

Google’s URL structure guidance for ecommerce accepts either a path segment (/t-shirt/green) or a parameter (/t-shirt?color=green) for variants. It is specific about what doesn’t work: “Google does not use fragment identifiers in indexing,” so /t-shirt#black and /t-shirt#white are the same page to Google. It also asks for ?key=value parameters rather than bare values, no session IDs or timestamps in URLs, and a self-referencing canonical on every indexable page.

The common duplicate patterns on stores:

1. The same product reachable under several category paths.
2. Tracking and sort parameters appended to internal links.
3. Paginated category pages that canonicalize to page 1. Google’s pagination guidance says not to: give each page its own canonical URL.
4. HTTP and HTTPS, or trailing slash variants, both resolving with a 200.

Canonical tags are hints, not commands, so the fix is to link internally to the canonical version everywhere, not only to declare it. Our canonical tag guide covers the mechanics, and duplicate content SEO explains why duplication wastes crawling rather than triggering a penalty.

Product and Offer data: merchant listings vs product snippets

Product structured data comes in two flavors, and a store selling directly should aim for merchant listing eligibility, which also makes pages eligible for product snippets.

Google’s Product structured data documentation separates them clearly. Product snippets are for pages where “people can’t directly purchase the product,” such as editorial reviews. Merchant listings are for “pages where customers can purchase products from you,” and support details like apparel sizing, shipping and returns. Adding the required merchant listing properties “means that your product pages can also be eligible for product snippets.”

The same page recommends sending data two ways: “Providing both structured data on web pages and a Merchant Center feed maximizes your eligibility to experiences.” It also asks for shipping costs and return policy details, and to provide “as much rich product information as available, without concern for the exact experiences that will use it.” That last line is good advice for AI surfaces too, since you can’t know which attribute a shopper’s question will hinge on.

What to put in the Offer, at minimum: price, currency, availability, condition, and a URL that matches the canonical product page. The most common failure we see is not missing markup but stale markup: a price in JSON-LD that no longer matches the price on the page or in the feed. Our product schema guide has full JSON-LD examples, and review schema covers the rules for marking up ratings.

On the Google side there is also a leak-level detail worth knowing. The 2024 Content Warehouse leak, analysed by iPullRank’s Mike King, lists an attribute QualityShoppingShoppingAttachment.shoppingSiteScoreShopfab, described as a shopping site classifier score, and CompressedQualitySignals.productReviewPUhqPage, “the possibility of a page being a high quality review page.” Google stores these attributes; the leak does not show how much they are weighted. They do confirm that Google classifies whole sites as shopping sites and scores review pages separately, which is a reason to keep editorial buying guides distinct from product templates.

How do AI shopping answers choose products?

AI shopping answers pick products from structured product data and third-party content, then rank merchants on availability, price and whether the seller is the maker. Clean feeds decide far more than clever copy.

OpenAI’s help article Shopping with ChatGPT Search states that product results “are not ads” and that ChatGPT considers “structured metadata from first-party and third-party providers (e.g., price, product description) and other third-party content.” When a shopper clicks a product, merchants are “ranked based on factors like availability, price, quality, and whether they are the maker or primary seller of that item.” Shopify merchants are already integrated through Shopify Catalog, and other merchants can apply to send OpenAI a direct product feed.

Google’s position for AI Overviews and AI Mode is that there are “no additional requirements” beyond normal indexing. Its AI features documentation does single out one action: make sure “your Merchant Center and Business Profile information is up-to-date.” Behind AI Mode shopping sits Google’s Shopping Graph, which Google says holds more than 50 billion product listings, with more than 2 billion of them refreshed every hour. Your feed is one of those listings. AI Mode also uses query fan-out, running many related searches for sub-questions, which is why well-structured category pages with real buying advice get pulled in for queries you never targeted.

The layered diagram below shows the inputs you control and the surfaces they feed.

Scroll the diagram sideways to see all of it.

Two practical consequences. First, price and availability have to agree across the page, the markup and every feed, because any engine comparing them will trust the store that is consistent. Second, being the maker or primary seller is an explicit ChatGPT merchant factor, so brands selling direct should make that obvious in their Organization data and product copy. Our guide to ChatGPT shopping goes deeper on feeds and the checkout side.

Content that earns links and citations for a store

Stores earn links and AI citations with buying guides, comparisons and sizing or fit data that no product page can hold. Blog posts about company news don’t do that job, and most content gets nothing at all: in Ahrefs’ study of around 14 billion pages, 96.55% got zero traffic from Google.

Google’s guidance on writing high quality reviews is the best brief for this content even if you aren’t writing reviews. It asks for evidence of your own experience, explanation of “what sets something apart from its competitors,” benefits and drawbacks from original research, and “quantitative measurements about how something measures up.” A running shoe store that publishes its own measured stack heights and weights by model has something a competitor can’t copy from the manufacturer.

Useful formats for stores:

• Buying guides by use case: “best boots for wet trails” that link to the category and the three products you’d actually recommend.
• Comparison tables: model versus model, with measured specs.
• Sizing and fit data: returns data turned into advice (“runs half a size small”).
• Care and repair guides: they attract links and reduce returns.

Link each guide to the matching category page with a descriptive anchor. That internal link is how the guide’s authority reaches the page that sells.

Where ecommerce SEO goes next

Ecommerce SEO is moving from ranking pages toward supplying accurate product data to every engine that shops on a buyer’s behalf. The fundamentals above still decide Google rankings, and they are also what AI assistants read.

A sensible order of work for most stores:

1. Audit crawl stats and indexing for filter and parameter URLs; block or collapse the waste.
2. Fix canonicals and internal links so every product has one URL.
3. Rewrite category intros and titles around real product-type queries.
4. Validate Product and Offer markup against live prices, and connect Merchant Center.
5. Apply for AI shopping feeds where available, and keep them in sync.
6. Publish two or three buying guides with original data per major category.

Speed still matters for conversion and crawling, especially on image-heavy templates. Google’s Web Vitals guidance calls a page good when its largest content element loads within 2.5 seconds and it responds to interactions in 200 ms or less, measured at the 75th percentile of page loads; Core Web Vitals explained covers what moves each metric. If you want help connecting the technical layer, the product data and AI visibility into one plan, that is what our AI SEO service is built around.