E-commerce SEO has a reputation for being endless: a checklist of forty items, each one urgent, most of them written by someone selling a plugin. On a real shop, the outcome is decided by far fewer things, and they are structural rather than cosmetic. How categories and products link to each other. What happens to the thousands of URLs your filters generate. How variants, stock and discontinued products are handled. Whether the product data Google reads matches the page. This guide takes those from Google’s own e-commerce documentation, sets out what Shopify, WooCommerce and PrestaShop already do for you, and names the popular advice that Google has explicitly said does nothing.
— Guide
E-commerce SEO: what actually moves an online shop
Most shop SEO advice is a list of forty things. On a real shop, about six of them decide the outcome, and several of the popular ones are myths Google has denied in writing.
What each platform does for you, and what it leaves
| Question | Shopify | WooCommerce | PrestaShop |
|---|---|---|---|
| Product URLs | A fixed /products/ path, which Shopify’s own redirects cannot touch | One of four product bases in Settings → Permalinks, /product/ by default | Friendly URLs must be switched on; the default pattern includes the category and an ID |
| Duplicate URLs | Products also reachable under /collections/…/products/…, canonicalised to /products/ | A category-in-URL option exists; keep bases short, as the docs advise | One page can have many URLs, which is why a redirect-to-canonical setting exists |
| Canonical tags | Added automatically | Added by WordPress core on single pages; plugins extend them | A 301 or 302 redirect to the canonical URL, chosen in the SEO settings |
| Sitemap | Generated automatically | Built into WordPress since 5.5, at /wp-sitemap.xml | Through PrestaShop’s own Google sitemap module |
| robots.txt | Editable through robots.txt.liquid since June 2021, which Shopify calls an unsupported customisation | A file you edit, or a plugin | Generated from the back office with one button, replacing the existing file |
| Product structured data | Depends on the theme | Output as JSON-LD by WooCommerce itself | Depends on the theme and modules |
| Several languages | Hreflang generated from your Markets and languages | Needs a multilingual plugin | Alternate-language links output by the default theme |
From each platform’s own documentation, checked September 2026, plus the default behaviour of Shopify’s Dawn theme. Themes, apps and plugins change several of these rows, which is the first thing to check on an existing shop.
A shop is a structure problem before it is a content problem
Google says it plainly in its e-commerce documentation: it "generally doesn’t look at the structure of URLs to work out the structure of a site." It works out which pages matter from how they link to each other. So the folder names in your URLs are a matter of taste, and the menu is not. Link the main menu to categories, categories to subcategories, and subcategories to products, and link to every product you want indexed, because Googlebot does not type into your search box to find the rest.
That one principle explains most of what goes wrong on shops that seem to do everything right. A product reachable only through on-site search, a filter or a "you may also like" widget loaded after the page is, as far as a crawler is concerned, barely there. A category page that lists forty products on its first page and loads the rest with a button nobody crawls is hiding most of its catalogue.
The fastest audit of any shop is to count how many clicks it takes to reach each product from the home page, following only plain links. Anything more than three or four deep, or reachable only through a script, is where the indexing problems will be.
Filters are where most shops leak
Every combination of colour, size, brand, price and sort order can become its own URL, and a catalogue of two thousand products can generate millions of them. Gary Illyes of Google called faceted navigation "by far the most common source of overcrawl issues site owners report to us," and the fix Google recommends is blunt: if you do not need those URLs in search, block them in robots.txt, or put the filters in URL fragments after a #, which Google ignores. Canonical tags and nofollow, the usual advice, are "generally less effective in the long term."
The part the blunt version misses is that some filter combinations are exactly what people search for. "Black running shoes for women" is a filter on most shoe shops and a real query with real volume. Those deserve to be proper pages: a stable URL, a heading and a paragraph of their own, and a link from the category they belong to. If you do let filter URLs be indexed, Google asks for a consistent parameter order, the & separator, and a 404 for combinations that return nothing.
So the work is a decision, not a setting. Make a short list of the filter combinations that match searches worth having, turn those into real landing pages, and keep crawlers out of everything else. On most shops that list has a few dozen entries, not thousands.
Categories carry the traffic
Most of a shop’s search demand is broad: "running shoes", "office chairs", "dog food". Those queries land on category pages, not products, and Google’s own advice is to give categories descriptions "to cover less specific queries", alongside product descriptions that use the words shoppers search with. A category with nothing but a grid of thumbnails is competing for its most valuable query with no text to be understood by.
Pagination is the other half. Google wants each page of a category linked in sequence with ordinary links and given its own URL, such as ?page=2, and it is explicit about two common mistakes: do not point page two onwards at page one with a canonical tag, and do not rely on rel="next" and rel="prev", which Google no longer uses. Infinite scroll and "load more" buttons need a crawlable fallback, because Google’s crawlers "don’t click buttons". An empty category should carry a noindex, or return a 404 if the category is gone for good.
If you want the wider picture of how shops of different sizes spend on this, what a €2,000 shop and a €50,000 shop actually differ in covers it, category pages included.
Variants, stock and the products you delete
Each variant a shopper might search for needs its own URL, as a path such as /t-shirt/green or a parameter such as ?color=green; Google treats everything after a # as the same page, so fragment-based variants are invisible. When variants use optional parameters, Google asks for the URL without them as the canonical. Since February 2024 there is also structured data built for this, ProductGroup, which tells Google that a set of variants is one product that varies by size or colour.
Stock is simpler than most shops make it. A product that is only out of stock should stay online and say so, with the availability in its structured data updated to match. That is Google’s own recommendation, because it keeps the page’s rankings for the day the stock returns. Deleting it, redirecting it or hiding it throws those away.
A product that is gone for good is a judgement call, and Google has never published a single rule for it. John Mueller’s advice has been consistent: redirect when there is a genuine replacement, and otherwise let the page return a 404 or show a helpful not-found page. What does not work is redirecting every discontinued product to the home page, which Google tends to treat as a soft 404 anyway.
Structured data and Merchant Center
Google distinguishes two kinds of product markup. Product snippets are for pages where people cannot buy directly, such as reviews and comparisons. Merchant listings are for pages where they can, and they unlock the richer shopping results: price, availability, shipping and returns shown right in the search results. Adding the merchant listing properties generally makes a page eligible for both.
Google’s documentation says that providing structured data on the page and a Merchant Center feed together "maximizes your eligibility", and Merchant Center is mandatory for some surfaces, the Shopping tab among them. Free listings in Shopping have existed since April 2020 and are switched on by default for most merchants, so a shop without a feed is leaving a free channel unused. The one rule that matters more than any property: the price and availability in the markup must match what the page shows, or Google stops trusting either.
What the platform decides for you
The table above is worth reading before you choose or change a platform, because several of the most argued-about SEO questions are settled by the platform and cannot be fixed by effort. Shopify’s /products/ and /collections/ paths are fixed, and Shopify’s help says you cannot redirect them. The duplicate product URLs its themes generate under collections are handled with a canonical tag back to /products/. Its robots.txt can be edited since 2021, but Shopify calls that "an unsupported customization" and warns that getting it wrong "can result in loss of all traffic."
WooCommerce is less limited than its reputation. The belief that it needs an SEO plugin to be indexable is out of date: WooCommerce outputs product structured data itself, and WordPress core has generated sitemaps since version 5.5 and canonical tags on single pages for much longer. A plugin earns its place for editable titles and descriptions, noindex controls and richer schema, not for the basics. PrestaShop’s friendly URLs are a switch in the SEO settings, and its documentation warns that turning them on for a server that does not support URL rewriting "can make your shop completely unavailable."
If you are still choosing, Magento vs Shopify for B2B goes deeper on the trade-offs, and we build on both of the most common ones: a Shopify Store or a WooCommerce Store, set up with the structure above from the first day rather than retrofitted.
Descriptions, reviews and the myths
The most repeated e-commerce SEO warning is that copying the manufacturer’s description gets a shop penalised. It does not, and Google has said so for years: there is no duplicate content penalty, and duplicate content "is not a violation of our spam policies." What Google does is pick one of the identical pages and show only that one. Copied manufacturer text is not penalised, it is filtered, and for every shop that is not the one chosen, the result is almost the same. Write your own descriptions for the products that earn money, and do not lose sleep over the long tail.
Reviews work the same way. Google’s guidance on high-quality reviews explicitly covers "a merchant who guides people between competing products", and asks for evidence of real use, measurements and differences "beyond what the manufacturer says." Its reviews system judges the reviews you write yourself, such as buying guides and comparisons, not the star ratings customers leave under a product. It also runs only in certain languages, which include English and German but not, so far, Romanian.
Three pieces of advice can be dropped entirely. Google has "no notion of optimal keyword density." John Mueller put it as bluntly as it can be put: "there is no such thing as LSI keywords." And there is no word count a category or product page has to reach; Google’s Starter Guide says there is "no magical word count target".
Speed matters, but not the way it is sold
Google confirms that Core Web Vitals are used by its ranking systems, and in the same breath says it will still show the most relevant page "even if the page experience is sub-par." Speed is a tiebreaker in ranking. In conversion, it is more than that. Deloitte’s study for Google across 37 brands found that a 0.1-second improvement in mobile speed went with 8.4% more conversions for retail sites, a correlation rather than a controlled test. Vodafone’s controlled test is the cleaner number: a 31% better Largest Contentful Paint produced 8% more sales.
For a shop, that makes speed a revenue project that happens to help SEO, rather than the other way round. Core Web Vitals, explained covers what each metric measures, and our Speed Optimisation package is the fix when a theme, a stack of apps or an unoptimised image library is what is slowing the product pages down.
How much of a shop’s revenue search really brings
It is less than the SEO industry implies and more than the ads industry would like. BrightEdge, measuring thousands of domains with direct traffic excluded, put organic search at 41% of retail and e-commerce traffic and 36.4% of revenue, against 30.2% for paid search. Wolfgang Digital’s 2020 report, across 130 million European, UK and US sessions, recorded the first year that paid search overtook organic in revenue share, 34% to 33%. Contentsquare’s 2026 benchmark, across 99 billion sessions, puts paid channels at 42% of all traffic.
Germany adds a complication of its own: in a 2023 survey of 1,000 people for an Amazon agency, 74% of respondents said they start a product search on Amazon rather than Google. Romania adds another. Its e-commerce market reached an estimated €12.9 billion in 2025, and eMAG alone reported sales of 10.6 billion lei, which means many shops compete in search with a marketplace that sells their own products. Neither changes the work above. Both change how much of a shop’s revenue it is reasonable to expect from it, and why a marketplace presence often sits alongside the shop rather than instead of it.
What AI search changes
Less than the vendors of AI-optimisation products suggest. Google’s own documentation says there are "no additional requirements to appear in AI Overviews or AI Mode", and that you do not need new machine-readable files, AI text files or special markup. What it asks for is what this guide has already covered: product data kept current in Merchant Center, and structured data that matches the visible page. AI Overviews and AI Mode are both available in Germany and in Romania.
The traffic is real and still small. Shopify reported AI-referred sessions to its merchants growing 197% in the second quarter of 2026 against 12% for organic search, while organic search still referred more sessions than all the tracked AI platforms combined. Agentic checkout, where an assistant completes the purchase, has so far launched only in the US. How AI agents actually buy things covers that side in detail. The shop that is easiest for an AI to recommend is the same shop that is easiest for Google to rank: fast, crawlable, and honest in its product data.
Sources
Checked September 2026. Google’s guidance is quoted from its own documentation, platform behaviour from each platform’s documentation, and the traffic figures from the studies named, with their samples. Where a figure is a correlation, it is described as one.
- Google Search Central — Help Google understand your e-commerce site structure ↗
Importance inferred from internal links rather than URL paths, linking every product you want indexed, and Googlebot not using on-site search.
- Google Search Central — Designing a URL structure for e-commerce sites ↗
A URL per variant, fragments treated as one page, the canonical without optional parameters, and noindex or a 404 for empty categories.
- Google Search Central — Pagination and incremental page loading ↗
Sequential links with a unique URL per page, no canonical from later pages to the first, rel=next/prev no longer used, and crawlers that do not click buttons.
- Google Search Central Blog — Crawling December: Faceted navigation ↗
December 2024: faceted navigation as the most common source of overcrawl issues, and robots.txt or URL fragments as the fix for filters that do not need indexing.
- Google Search Central — Managing crawling of faceted navigation URLs ↗
Canonical and nofollow described as less effective in the long term, and the rules for indexable filters: & separators, consistent order, a 404 for empty combinations.
- Google Search Central Blog — Product variants structured data ↗
February 2024: ProductGroup, hasVariant and variesBy for describing a set of variants as one product.
- Google Search Central — Product structured data ↗
Product snippets against merchant listings, and structured data plus a Merchant Center feed as the way to maximise eligibility.
- Google Search Central — Pause your business online ↗
Keeping pages online and marking products out of stock, with availability updated in structured data, as the approach that minimises negative effects.
- Search Engine Journal — John Mueller on expired product pages and soft 404s ↗
Redirecting only to a genuine replacement, and why pages for products that are gone completely can be treated as soft 404s.
- Google Search Central — SEO Starter Guide ↗
Duplicate content not a violation of the spam policies, and no magical word count target.
- Google Search Central — Write high quality reviews ↗
Guidance that explicitly covers merchants guiding people between products, asking for evidence and detail beyond the manufacturer’s description.
- Google Search Central — Reviews system ↗
First-party review content evaluated, customer reviews on product pages not, and the list of supported languages, which does not include Romanian.
- Google Search Central — Understanding page experience ↗
Core Web Vitals used by the ranking systems, with the most relevant content still shown even when the page experience is sub-par.
- Google Search Central — AI features and your website ↗
No additional requirements or special optimisation for AI Overviews and AI Mode, and no need for new machine-readable files or markup.
- Shopify Help Center — SEO overview ↗
Automatic canonical tags, sitemap.xml and robots.txt on every Shopify store.
- Shopify Help Center — Editing robots.txt.liquid ↗
Editing robots.txt described as an unsupported customisation that can, done wrong, result in the loss of all traffic.
- WooCommerce — Structured data for products ↗
WooCommerce itself outputs product structured data as JSON-LD.
- PrestaShop — SEO and URLs ↗
The friendly URL switch and its warning, the redirect to the canonical URL, and robots.txt generation from the back office.
- BrightEdge — Channel Report 2019 ↗
Thousands of domains, direct traffic excluded: organic search 41% of retail and e-commerce traffic and 36.4% of revenue, paid search 30.2% of revenue.
- Wolfgang Digital — KPI Report 2020 ↗
Over 130 million sessions across Europe, the UK and the US: paid search 34% of revenue against organic 33%, the first year paid overtook organic.
- Deloitte for Google — Milliseconds Make Millions ↗
37 brands and 30 million sessions: a 0.1-second mobile speed improvement associated with 8.4% more retail conversions. A correlation, not a controlled test.
- web.dev — Vodafone case study ↗
A controlled test: a 31% improvement in Largest Contentful Paint led to 8% more sales.
- Shopify — How AI search is changing category behaviour ↗
Q2 2026: AI-referred sessions up 197% year on year against 12% for organic search, with organic still referring more sessions than all tracked AI platforms combined.
- OMR — 79 percent of Amazon shoppers start their product search on Amazon ↗
A 2023 survey of 1,000 people in Germany by Remazing and Appinio: 74% of all respondents start a product search on Amazon rather than Google. Commissioned by an Amazon agency.
- Economedia — Romania, the largest e-commerce sector in Eastern Europe ↗
The European E-Commerce Report 2026 estimate of €12.9 billion in Romanian B2C e-commerce for 2025.
— FAQ
Frequently asked questions
Want to know which of these your shop gets wrong?
Send us the address. We will check the structure, the filters, the product data and the speed, and tell you which two or three fixes will actually move anything.