Guide · Ecommerce SEO
Ecommerce SEO: Which Page in Your Shop Should Rank for Which Search
Every page type in a shop gets exactly one search job – that is the core of ecommerce SEO
Ecommerce SEO means giving each page type in your shop its own search job. Category pages carry the buying searches (for example “women’s running shoes under 100 euros”), product pages carry the specific items and model numbers, and guide pages carry the questions people ask before they buy. The mistake I keep seeing in shop projects is not a technical one. It is a gap: for the most important buying search, no page is responsible. The blog ranks, the category does not. Technical SEO comes after that. For filters, variants and pagination, Google has its own documentation for online shops, and it is clearer than most guides on the subject. What it does not answer: which filter combination deserves a page of its own – and how to treat a sold-out product, for which there are only Google’s general rules on 404 and 301. You decide that, not your shop system.
Five sentences before you change anything in your shop
- One category page per buying search, one product page per item, one guide per question before the purchase. If a mapping is missing, either the wrong page ranks or none does.
- In a Shopify shop I evaluate in Search Console, 8 of the 12 pages with the most clicks over 16 months are category pages (collections), two are product pages. Categories bring the volume; individual bestsellers top the list.
- Filter URLs that should not rank are blocked in robots.txt. Google recommends this in its faceted navigation documentation; canonical tags and nofollow are, according to Google, usually less effective in the long run.
- Every variant and every paginated page needs its own URL with an
<a href>link. Google does not follow buttons and generally does not read#fragments. - Structured data on the product page and a Merchant Center feed complement each other. Google recommends both; for the Shopping tab, markup alone is not enough.
Which page for which search? The mapping that comes before any technical work
Before you change a title or block a filter, you need an answer to one question for every search that brings revenue: which page in the shop is responsible for it? Without that mapping, you are optimising pages that are responsible for nothing.
Four search types cover almost every shop. Each has a page type that should carry it, and a mistake I see again and again. The examples are made up and come from the running-shoe range of a fictional sports retailer – not a shop I work for:
| Search type | Example | Responsible page type | Typical mistake |
|---|---|---|---|
| Buying search for a product group or an attribute | “women’s running shoes under 100 euros”, “buy merino shirt” | Category or subcategory | There is only the top-level category “Running shoes”; the price question is answered by a blog post that cannot sell |
| Search for a specific item | Model number, brand plus model, EAN | Product page | Manufacturer text with nothing of your own; three variant URLs compete with each other |
| Brand search within the range | Manufacturer name plus product group, such as “[brand] running shoes” | Brand category (manufacturer page with a listing) | The brand is only a filter value and therefore blocked or without text |
| Question before the purchase | “how often to replace running shoes”, “running shoes or trail shoes” | Guide that links to the matching category | The guide ranks and ends without a link to the category – or the category tries to answer the question |
Two observations from my Search Console show why the mapping sets the order of work. Both are snapshots, not before-and-after comparisons, and I am not naming the shops.
Clicks come mainly through categories. In a Shopify shop I look after, 8 of the 12 pages with the most clicks since Search Console data began (June 2025 to October 2026) are collection pages, in other words categories. Two are product pages, two are variants of the homepage. Categories carry the breadth, individual bestsellers the peak: in the shorter period from August to October 2026 (3,694 clicks, average position 10.2), a product page is in first place. Among the twelve pages with the most clicks, it is followed by six collections (one of them in the French language version), two more product pages and three variants of the homepage. Anyone who puts all their time into product descriptions in this shop before the categories are in place has the wrong priority.
Buying pages and guide pages share the traffic – and both need their own page. A shop with a guide section gets most of its clicks after the homepage, between April and October 2026, through a buying page with the word “kaufen” (German for “buy”) in the URL: 17,915 clicks at position 4.2. The strongest guide page, a question about how to use a product, brings 9,580 clicks. Both searches have their own page. Neither has to do the other’s job. That is what the mapping from the table looks like when it works.
What I cannot derive from this: that a new category page in your shop will bring a particular number of clicks. For that I would need a project in which I changed the mapping on a set date and measured the effect in isolation afterwards. I am writing this down because many of the guides I have read on the subject claim an effect at this point without backing it up.
The search for missing categories starts in Search Console: the “Queries” tab, filtered by buying words such as “buy”, “cheap”, “under”, “for”, sorted by impressions. Every query with many impressions and a position beyond 20 is a candidate for a page that does not exist yet. How to turn that into a clean list is covered under keyword research.
Category pages: one search target per attribute, not per product group
One category page per product group is not enough, because buying searches rarely ask for product groups. They ask for price, brand, target group, features or size. Each of these attributes that has demand of its own gets its own subcategory with its own URL, its own title and its own text.
This is what it looks like in a shop whose 2026 editorial plan I manage. Instead of new blog posts, the plan lists subcategories month after month, each split by one attribute for which Search Console or Keyword Planner shows separate demand:
- Price segment: several “under €X” tiers, each with its own search demand
- Target group: men, women, beginners, for example
- Brand: the manufacturers people search for in combination with the product group
- Features: the technical attributes buyers type into the search box
- Weight and size: weight classes and individual sizes with demand of their own
By September 2026, all planned pages are approved and implemented. Whether they rank is something I cannot show with figures here, because I do not have Search Console access for this shop – the method is documented, the effect is not. I mention it anyway because it shows the logic: every row in the plan is a buying search that previously had no responsible page.
The order in which you create such pages depends on two things: impressions in Search Console for the attribute and the number of products behind it. A subcategory with two products is, for Google, an almost empty page. For empty categories, Google’s guide to URL structure for ecommerce sites explicitly recommends noindex – or, if the system removes empty categories automatically, considering a 404.
How the categories link to each other also plays a part in how important Google considers them. The documentation on shop structure describes the path menu → category → subcategory → product pages and strongly advises linking every product that should be indexed. It also says that Google reads the number of internal links to a page as a hint of its relative importance. And that Googlebot generally does not type search terms into your shop’s search box. Google usually won’t find anything that can only be reached through the search box.
For the text on the category page, there is no word count from Google. Word counts such as “around 350 words”, which appear in guides and AI-generated summaries on shop SEO, do not come from a Google source. What the page needs is the part a filter cannot deliver: what the products in this selection are for, how they differ, what goes wrong at the point of purchase. Two paragraphs that do that beat 350 words that do not.
Filters, variants, pagination: what Google specifies and what you decide
For filter URLs, variants and pagination, the best source is Google’s own documentation, not a shop forum from 2014. Most of it has been online in an updated version since December 2025. What Google specifies there, I separate clearly here from what is my own assessment.
Filter URLs that should not rank are kept away from crawling. In the documentation on faceted navigation, Google describes how crawling filter URLs tends to cost sites a lot of computing resources. If filtered URLs should not appear in search, Google recommends blocking their crawling in robots.txt. Canonical tags and nofollow are, according to the documentation, generally less effective in the long run. For filter combinations without a single product: HTTP status 404 at the same URL, no redirect to a general error page. Google generally does not read fragments after # when crawling. Filters that only live in the fragment are therefore invisible – in both directions.
Which filter combination deserves its own page is something Google does not tell you. This is my rule, not a Google specification: a combination becomes a crawlable landing page if it meets three conditions. First, there is a separate search query for it with impressions in Search Console. Second, there are enough products behind it that the page is not an empty list. Third, you maintain a separate title and separate text for it – otherwise it is a duplicate of the parent category with a different URL. If it meets all three, it is no longer a filter URL but a subcategory as in the previous section. If it fails one, it stays blocked.
Decision path for a filter URL:
- Does the combination have zero results? → 404, no redirect.
- Is there a separate search query and enough products, and do you maintain a separate text? → Own subcategory with a clean URL, linked from the menu or the parent category.
- Otherwise → Block crawling in robots.txt. Do not work with noindex on every single filter page, because Google has to load each URL first to see the noindex – that is my trade-off; Google lists noindex as an option for such variants.
Variants get their own URL, and it must be clearly defined which of them is canonical. In the guide to URL structure, Google recommends making every product variant such as size or colour reachable via its own URL. Whether as a path segment or as a parameter makes no difference to Google. With optional parameters, the URL without the parameter counts as canonical, and the variant URL points to it via rel="canonical". Temporary parameters such as session IDs and tracking codes should, according to the same documentation, not be linked internally. If every variant has its own full page, then according to Google’s documentation on product variants there is no single canonical URL for the whole product group – each variant then needs its own distinct URL.
Paginated pages each get their own URL and remain indexable. The documentation on pagination and incremental page loading says three things that contradict the old forum debates. The first page should not serve as the canonical for all subsequent pages. Google no longer uses rel="next" and rel="prev". And the next page must be linked via <a href>. Sorted versions of the same list (?order=price) are kept out with noindex or robots.txt. For infinite scroll and “load more” buttons, the same documentation applies: Google generally only crawls URLs from href attributes and does not click buttons. If you use infinite scroll or a load-more button anyway, you need parallel paginated URLs – or you supply the products via sitemap and Merchant Center feed. The details with examples are in the guide on pagination.
Google itself says in the URL documentation that users of a shop platform can usually skip the URL section because the platform covers it. That is true for the variant canonicals and the pagination. It is not true for the question of which filter combination becomes a subcategory – Shopify, Shopware and WooCommerce all leave that decision to you. Where the systems differ in detail is covered on the pages for Shopify SEO, WooCommerce SEO and Shopware SEO.
Product pages: your own text, variant URL, structured data, Merchant Center
The product page ranks for the specific item. For that it needs four things: a text that does not come from the manufacturer, one canonical URL per product, structured data, and a Merchant Center feed that matches the page data.
The manufacturer text is the text every other retailer has too. Google’s guide to writing high-quality reviews is aimed at review articles, not at shop product pages. What it demands there can be carried over, though – that is my derivation, not a Google specification for retailers: first-hand experience, measurements, how the product differs from alternatives, recommendations with reasons, and quality and originality before length. Translated for the product page: what you yourself know about the item belongs in the text. Which shoe size fits which foot length. Which accessory is missing from the box. What has changed compared with the previous model. Two sentences of your own with a measurement beat ten paragraphs of manufacturer prose, because the ten paragraphs appear on forty other pages.
Structured data decides how the product page is displayed in search. In the documentation on product markup, Google distinguishes two cases: product snippets for pages without a direct purchase and merchant listings for pages where you can buy. For merchant listings, according to the merchant listing documentation, name, image and offers are required, and the offer must be an Offer, not an AggregateOffer. If you implement the required fields for merchant listings, Google says you may also be eligible for product snippets. A price drop display is calculated by Google itself from observed price changes and is not guaranteed; there is no separate markup property for it. Which further types make sense in a shop – BreadcrumbList, Organization with return information, ProductGroup for variants, Review – is listed in Google’s overview of ecommerce structured data. How to check markup and find typical errors is in the guide on structured data.
One thing I often read differently on agency service pages: structured data is there for rich results, not for visibility in AI answers. According to Google’s guide to optimising for generative AI features in Search, it is not required for that, and there is no special markup for it. Google continues to recommend it because it qualifies pages for rich results.
Merchant Center and markup complement each other but do not replace each other. In the documentation on sharing product data, Google recommends combining both. The Shopping tab can only be reached through Merchant Center; markup alone does not get you there. And there is a trap Google describes in the same documentation: the feed deviates from the website. On the page the product is sold out, in the feed it still says “in stock”. For this, Merchant Center has the setting that updates product data automatically from the website data. Without it, Google shows prices and availability that are wrong on your page. In my experience that costs trust faster than any ranking.
Product images are ranking targets in their own right. Alongside web search, Google lists image search, Lens, the Shopping tab, the Business Profile and Maps in its overview of surfaces for shop data. What a product image needs for that is in the guide on image SEO.
Sold out, discontinued, no longer available: three cases, three status codes
You treat an unavailable product according to whether it is coming back, has a successor, or is gone for good. Each answer leads to a different status code, and the wrong answer costs either rankings or trust.
First, so you know what you are relying on: Google has no rule specifically for sold-out products in its shop documentation. In general, though, Google writes in the guide to fixing crawling errors: if a page is gone and there is no replacement page with similar content, return 404 or 410. If it has a clear replacement, redirect there with a 301. In addition, Google advises in the guide to site moves against redirecting many old URLs to a single unrelated page such as the homepage – that can be treated as a soft 404. Cases 2 and 3 in the matrix follow from that. Case 1 is my rule from practice.
| Case | What happens to the page | Status code | What else you do |
|---|---|---|---|
| Coming back (season, reorder) – my rule | Page stays online with full content; availability in the markup set to OutOfStock, with a foreseeable restock BackOrder, and the feed to match (PreOrder only for items not yet released) | 200 | Offer a delivery date or a notification, link to alternatives from the same category. Do not strip the page down to a “sold out” notice: Google recognises soft 404s by content, and a page that only shows “sold out” risks exactly that |
| Has a successor – Google: 301 for a clear replacement | Redirect the old URL permanently to the successor | 301 | The successor page mentions the previous model in its text, so the old search lands there |
| Gone for good, no successor – Google: 404/410 without a replacement | Remove the page | 404 or 410 | The category stays; remove internal links to the old URL; no redirect to the homepage |
The availability values for the markup are in the merchant listing documentation.
The most common mistake is the third case, solved wrongly: discontinued products are redirected wholesale to the homepage or the parent category. For the user, that is a page with nothing to do with their search; for Google, a redirect to an unrelated page. Google warns in the site move guide that many redirects to a single unrelated target such as the homepage can be treated as soft 404s. An honest 404 with links to matching products is the better answer. Which code is correct when, and how to find soft 404 errors in Search Console, is in the guide on status codes.
The second most common mistake is the first case, solved too hastily: a product that will be back in four weeks gets deleted. And with it the URL that ranked. After the reorder, a new URL starts from zero.
What Google says about shops in AI answers
For the visibility of your products in AI Overviews and AI Mode, Google names two levers you need anyway: Merchant Center and the Business Profile. In its guide to optimising for generative AI features in Search, last updated in July 2026, Google writes that product data from Merchant Center and the details in the Business Profile help products appear in AI answers and other search results. AI answers in Google Search can, according to the same guide, contain product lists and product information. Google states explicitly that you need no new machine-readable files, no AI text files, no additional markup and no Markdown for this – Google Search does not use them.
For ChatGPT, Perplexity and other systems, there is no comparable documentation, and I am not promising you anything there. What can be said beyond Google and what cannot is covered under SEO and GEO.
Self-check: nine questions for your shop
You check the nine questions with Search Console, a look at robots.txt and the source code of one product page. Each one points to the section that explains the answer.
- Is there a category page responsible for every buying search with impressions? Queries with buying words and a position beyond 20 are the candidates. → Mapping
- Does a blog post rank for a search that a category should carry? Then the category is missing, or it is not linked. → Mapping
- Are the subcategories split by attributes with demand of their own – price, brand, target group, features? → Categories
- Does Google reach every product via an
<a href>link from a category, without a search box and without a button? → Categories, Filters - Are filter combinations without demand of their own excluded from crawling in robots.txt – and do combinations with no results return a 404? → Filters
- Does every variant have its own URL – and is it clearly defined which URL is canonical? Does every paginated page have its own indexable URL? → Filters
- Is there at least one sentence on the product page that the manufacturer did not write? → Product pages
- Do price and availability match in the markup, in the Merchant Center feed and on the page? → Product pages
- Do discontinued products return a 404 or a 301 to a successor – and no redirect to the homepage? → Out of stock
If you want to answer the nine questions for your shop without opening every page individually: I pull the complete inventory with a crawler and evaluate it locally. With a few thousand product URLs, that is faster than any sample and shows the cases a sample misses. What that looks like in an audit is described under SEO audit.
When the shop system is enough and when it is not
The shop system handles the technology Google means in the URL documentation: variant canonicals, pagination, clean parameters. What it does not handle are the decisions in this text: which buying search gets a category, which filter combination becomes a subcategory, what happens to a discontinued product, which sentence on the product page comes from you.
Whether Shopify, WooCommerce or Shopware ranks better is the wrong question. With the same structure, the same problem arises in all three systems: categories without a search target. The differences are in the mechanics, not in the outcome. With Shopify, the point is that the URL paths /products/ and /collections/ are fixed and robots.txt can only be changed via the theme template robots.txt.liquid. With WooCommerce, it is about plugins and filter architecture. With Shopware, about properties and listing URLs. That is what the system pages on Shopify SEO, WooCommerce SEO and Shopware SEO are for.
You need ongoing support if nobody in your shop regularly checks the mapping from the first section. Ranges change, products are discontinued, filters are added, and every one of these changes can break a mapping. What ongoing shop SEO with me covers is on SEO for online shops; whether a freelancer or an agency fits is under SEO agency or freelancer; what it costs is under SEO costs. A relaunch or a change of system is the moment when every mapping can break at once – for that there is website relaunch support.
Frequently asked questions about ecommerce SEO
Why are my category pages not showing up in Google?
Usually for one of three reasons. The category can only be reached through a search box, a filter or a JavaScript menu instead of an <a href> link. Or it has no text and no title of its own and is therefore a duplicate of the parent category. Or it is blocked by robots.txt or noindex, often as a side effect of a filter block. Check the URL with the URL Inspection tool in Search Console – it shows whether Google knows the page and why it is not indexed.
Should paginated pages be set to noindex?
No. In the documentation on pagination and incremental page loading, Google recommends giving each page its own URL and leaving it indexable. Page 1 should not serve as the canonical for all subsequent pages; Google no longer evaluates rel="next" and rel="prev". What belongs on noindex or blocked in robots.txt are variants of the same list that you do not want in the index, such as sorted by price. The next page must be linked via <a href>, otherwise Google does not find the products behind it.
Do filter pages belong in the index?
Only those for which there is a separate search query, enough products and a text of their own – you treat those as subcategories. All other filter URLs are kept away from crawling in robots.txt, according to Google’s documentation on faceted navigation; canonical and nofollow are usually less effective in the long run. Filter combinations with no results return a 404. Shop themes that output filter tags as pages of their own are a known pattern with Shopify and Shopware and, in my experience, a frequent cause of thousands of indexed URLs with no value.
Is Shopify or WooCommerce better for SEO?
Neither of them decides the ranking; the structure of your shop does. In both systems, a category with its own search target, its own text and clean linking ranks, and in both, a category without them fails. The differences are in the mechanics: Shopify fixes the URL paths /products/ and /collections/, and you can only change robots.txt via the theme template robots.txt.liquid; WooCommerce leaves everything to you, including the mistakes. Details are under Shopify SEO and WooCommerce SEO.
Does Google read infinite scroll and “load more”?
Generally not. According to the documentation on pagination and incremental page loading, Google only crawls URLs from href attributes and does not click buttons. Products that are only loaded after a click or while scrolling remain invisible to Google if there is no parallel paginated URL with an <a href> link. As a supplement, Google names the sitemap and the Merchant Center feed – they are no substitute for linking.
Do I need Merchant Center if I have structured data?
Yes, if your products should appear in the Shopping tab – markup is not enough for that. In the documentation on sharing product data, Google recommends combining both. Structured data on the product page makes it eligible for merchant listings in web search; the feed serves Shopping and – according to Google’s guide to generative AI features in Search – acts as a data source for AI answers. What matters is that price and availability match in feed and markup; if they deviate, Merchant Center has a setting that updates the feed from the website data.
Viktor Pásztor
Viktor Pásztor is an SEO freelancer in Berlin, has been in digital marketing for over 15 years and manages around 15 client projects in parallel, mostly ecommerce and B2B. He works with WordPress, Shopify, Shopware and TYPO3 – fully remote for more than five years. More about Viktor Pásztor LinkedIn
Keep reading
SEO for online shops
What ongoing shop SEO covers.
Pagination
Own URLs, canonicals, infinite scroll.
Structured data
Checking markup, finding errors.
Status codes
Using 301, 404, 410 and soft 404 correctly.
Which page in your shop is responsible for which search – settled in 30 minutes
Send me the URL of your shop. In a free initial call, we go through the nine questions of the self-check using your real Search Console data – and afterwards you know which category comes first.
→ Book an initial callContact · kontakt@viktorpasztor.de