Ecommerce SEO starts below the surface: a clear site structure, complete product data, fast pages and URLs that search engines can crawl without getting lost. I fix that technical foundation first on every online store, because Google rankings and AI shopping answers both read from it.
The short answer
- Ecommerce SEO is the work that makes category and product pages easy for search engines and AI assistants to find, understand and trust.
- Fix the structure first: clear product and category URLs, and a deliberate plan for filter combinations that should or should not be crawlable.
- Keep product data consistent across the visible page, structured data and feeds. That supports discovery and shopping eligibility; it does not decide every ranking or AI recommendation.
- Measure speed with Core Web Vitals and real visitor data. A slow store creates friction, but no speed score alone predicts its rankings or sales.
- Keywords and backlinks matter, but they cannot rescue a store that search engines struggle to crawl.
What is ecommerce SEO, and what still holds from the old advice?
Ecommerce SEO is the practice of making an online store’s category and product pages discoverable, understandable and worth ranking. The classic advice from ten years ago still mostly holds: research specific product terms, write unique descriptions, describe images, work well on mobile and collect reviews. What changed is how much the technical foundation now decides.
This site first published ecommerce SEO advice in 2015. One useful principle survives: match the page to what a buyer is trying to choose. A specific product page can answer “brown leather A5 notebook”, while a notebook category helps someone compare options. Neither format wins every query; check the intent before forcing it onto the wrong template.
Some old advice has aged badly. AMP is no longer a priority for most stores, and caching needs to suit the host and checkout. For Google’s AI features, I would use its current eligibility guidance: pages need normal Search eligibility and the site’s effective Search generative AI setting must include them. Include is the default. A clean store can still be omitted from an answer.
A quick note on how I work. I don’t sell SEO campaigns. My team at WP Creative builds and maintains the technical foundation, and today we refer ongoing SEO campaigns to specialist partners. This post is about the part we build.
Why does site structure come first in ecommerce SEO?
Site structure comes first because search engines can only rank pages they can reach and tell apart. An online store with clean categories, one URL per product and controlled filter pages gets crawled efficiently. A store that spawns thousands of near-duplicate URLs wastes the crawler’s time and splits ranking signals across copies.
The usual culprit is faceted navigation: filters for colour, size, price and brand that each create a new URL. Google’s own guidance says filtered URLs can cause overcrawling and slower discovery of new pages, and recommends blocking filter parameters in robots.txt unless you genuinely need those pages indexed.
What I check on every store:
- A category tree no deeper than it needs to be, with every product reachable in a few clicks from the home page.
- One canonical URL per product, even when the product sits in several categories.
- Filter, sort and internal search URLs kept out of the index unless a filtered page targets a real search term.
- Duplicate or overlapping pages merged, with 301 redirects from the old URLs.
A useful example is Tile Factory Outlet, a tile importer selling in-store and online. When it came to my team in 2024, indexed pages had fallen while paid campaigns carried significant traffic. Month one focused on quick wins, unused plugins, speed and maintenance, followed by on-page work, content briefs and a fortnightly care plan. Our case-study record reports more quote requests after the programme. Traffic rose from 20.9K in March 2024 to an average 26.3K a month across September and October, with paid visits included. The record does not separate the effect of the website from marketing spend. I take it as a lesson in sequence, not a promise that one fix produces the same result elsewhere.
If you are building on WooCommerce, my guide to creating an ecommerce store with WordPress covers the store setup that this structure sits on.
How should product pages be written and marked up?
Product pages should carry unique, specific copy and structured data that matches what the shopper sees. Each product needs its own description, real image descriptions, price, availability, shipping and returns details, and genuine reviews. Search engines and AI assistants read that data directly, so gaps and contradictions cost visibility.
Write what the buyer needs to decide: size, material, fit and who the product suits. A manufacturer’s description may be accurate, but it rarely answers every question your customer has. Describe informative images in context; decorative images need an empty alt attribute. Show genuine reviews where you have them, without inventing ratings to complete a template.
Google’s product documentation distinguishes product snippets, including pages where people cannot buy directly, from merchant listings for pages that sell the product. These features overlap. Use the markup suited to the page, and keep it consistent with the product feed and visible details.
| Product data | Where it is used | What goes wrong when it is missing |
|---|---|---|
| Unique title and description | Search results, AI answers and feeds | Buyers and systems have less information to distinguish the item |
| Price and availability | Merchant listings, Google Merchant Center, ChatGPT product feeds | Out-of-date prices, disapproved listings, lost trust |
| Shipping and returns | Merchant listings | Shoppers cannot see the full cost before they click |
| Reviews and ratings | Product snippets, on-page trust | Fewer clicks and weaker purchase confidence |
| Brand, assigned GTIN and stable item or variant ID | Product matching in supported feeds | Missing or inconsistent identifiers can prevent correct matching; do not invent a GTIN |
My practical rule is to choose one master source for each product fact. Record where price, stock, variants and shipping are maintained, then trace which feeds and page templates use them. A corrected description achieves little if another system republishes yesterday’s price. Test one real product through every destination before changing the whole catalogue.
Check variants separately from duplicate filters. Open the URL for a specific size or colour and confirm that it selects the right option, image, price and stock. Google’s product-variant guidance distinguishes a single product page with selectable variants from variants spread across separate pages. Agree the URL, canonical and markup approach for that model, then test the same variant in the page and feed. A stable item ID does not fix a link that opens the wrong option.
How fast does an online store need to be?
An online store needs to pass Google’s Core Web Vitals on mobile: the main content loads within 2.5 seconds, the page responds to a tap within 200 milliseconds, and the layout stays stable while it loads. Those pass marks are measured on real visitors, so a fast office connection proves nothing.
The thresholds come from Google’s Core Web Vitals guidance: Largest Contentful Paint of 2.5 seconds or less, Interaction to Next Paint of 200 milliseconds or less, and Cumulative Layout Shift of 0.1 or less, each measured at the 75th percentile of page loads.
On the stores we rebuild, the slow parts are rarely a mystery:
- Oversized product images served at full resolution.
- Too many plugins or apps loading scripts on every page, including checkout.
- Cheap shared hosting that buckles under a sale or a bot attack.
- Page caching that is missing or misconfigured. Cache pages for anonymous visitors, but never serve the cart, checkout, account pages or anything personalised from a shared page cache. Logged-in shoppers need object caching and a fast server instead.
Start with a baseline from free website speed testing tools, then work through my steps to optimise WooCommerce performance.
How do AI answers choose which product pages to show?
AI shopping products use different discovery systems, so I would not assume they share a ranking formula. Google’s AI features use its Search systems; supported ChatGPT integrations can submit merchant product feeds. Accurate pages and feeds help those systems understand an offer. They do not guarantee that it will be recommended, or that a shopper can complete checkout through an assistant.
Google’s guidance for AI Overviews and AI Mode is plain. A page must be indexed and eligible to show a snippet. There is no special schema to add, but structured data should match the visible text, and Google advises keeping Merchant Center information up to date.
OpenAI’s current file-upload specification requires nine discovery fields: item ID, title, description, product URL, brand, seller name, image URL, availability and price. The ID must stay stable when stock or price changes. Product variants need their own rows and identifiers. Eligibility does not guarantee display, and checkout requires a separately enabled integration.
The stores that win in AI shopping answers will not be the ones chasing a new trick. They will be the ones whose product data is complete, consistent and identical across the page, the structured data and every feed. That is unglamorous build work.
How do I know if the foundation is working?
Check three different things: whether Search Console indexes the intended product and category URLs, whether real visitors get a usable experience, and whether analytics records sales accurately. Search Console does not report organic sessions turning into sales; that needs ecommerce tracking in your analytics system. Keep those definitions separate before deciding the foundation has improved.
The original version of this post said it bluntly: SEO without sales equals nothing. That still holds. Set up ecommerce tracking in Google Analytics 4 so you can see revenue by landing page, not just traffic. Then review three things each month:
- The Page indexing report in Search Console, looking for product pages excluded as duplicates or “crawled, currently not indexed”.
- The Core Web Vitals report, split by mobile and desktop.
- Revenue and conversion rate by landing page type: category, product and content.
If indexed pages fall suddenly, as they did for Tile Factory Outlet, look for a structural cause first: a theme change, new filters, a migration without redirects, or a plugin that changed canonical tags. If you are still choosing a stack, my take on whether WordPress is a good choice for SEO covers the platform side.
What to do next
- Crawl your store and count how many URLs exist compared with how many products you actually sell. A large gap usually means filter or duplicate URLs.
- Pick your top 20 products and check the facts, useful image descriptions, genuine reviews where available, and valid markup appropriate to pages where people can buy.
- Check Core Web Vitals in Search Console for mobile, and fix the product page template first.
- Make one system the master for product data, and make sure your site, Merchant Center and any other feeds read from it.
- If you want a second opinion on your store’s technical foundation, book a free call with me and my team through the contact page.
Frequently asked questions
What is the difference between ecommerce SEO and regular SEO?
Ecommerce SEO deals with problems that brochure sites rarely have: thousands of product URLs, filters that create duplicate pages, products that go out of stock, and product data that must match feeds and structured data. Regular SEO principles still apply, but an online store succeeds or fails on structure and product data at scale.
Do I need product structured data for ecommerce SEO?
Product structured data is not required to rank, but it makes a store eligible for richer results such as price, availability and review stars in Google. Google recommends merchant listing markup on pages where people can buy. The markup must match what shoppers see on the page, or it can be ignored.
Should filter pages be indexed?
Keep useful filtered landing pages indexable when they meet a distinct need. For combinations you do not want crawled, Google documents robots.txt controls. Blocking crawling is not the same as removing an already indexed URL: a blocked crawler cannot read a new noindex tag. Plan deindexing before crawl blocks, and test a small set so valuable category or product URLs remain reachable.
Is there a special way to optimise for ChatGPT or Google AI Overviews?
There is no special schema required for Google’s AI answers. Check ordinary Search eligibility and the site’s effective Search generative AI inclusion setting, then keep products and Merchant Center details accurate. Supported ChatGPT feeds have their own integration requirements. Clean pages and consistent data are a sensible foundation, rather than a guarantee of AI display or checkout.
What should I do with out-of-stock or discontinued products?
Keep temporarily out-of-stock products live and marked as out of stock, so the page keeps its rankings. For permanently discontinued products, redirect the URL only to a genuine replacement or a category that closely matches what the shopper wanted. If nothing comparable exists, let the URL return a 404 or 410 so it drops out of the index; Google treats the two the same way. A redirect to a loosely related category or the home page can be treated as a soft 404 and throws away the page’s relevance. One exception: if an old product page still earns links or searches, keep it live as an information page that says it is discontinued and points to alternatives.
