I recently held a workshop on how to optimize product pages for search on a large ecommerce website. It is an interesting area, and I still think there is a lot of untapped potential in product page SEO.
On an ecommerce website with tens of thousands of products, there is very little room for optimizing every product page individually. A lot of ecommerce SEO work naturally focuses on strong category pages and inspirational or editorial content. When it comes to product pages, the most valuable changes are usually the ones that can affect every product page at once.
That is classic large-scale SEO.
It is also increasingly relevant for AI-driven search. Product pages contain highly structured information about products, attributes, price, availability, specifications and relationships between products. If that information is technically accessible and clearly structured, both traditional search engines and AI systems have a much better foundation for understanding what you sell.
What should you optimize on a product page?
1. Page titles and meta descriptions
The scalable way to handle page titles and meta descriptions on product pages is to create templates that can be applied across the entire product catalogue.
A page title template could look something like:
- [PRODUCT NAME] – Buy at [XYZ]
A meta description template could look like:
- Buy [PRODUCT NAME] at [XYZ]. Fast delivery, easy returns and competitive prices.
The main query for most individual product pages will usually be the product name itself, so the template naturally includes it.
These templates can be managed through the ecommerce platform, a plugin, a custom solution developed specifically for the website or through product data imports.
The important point is that you are not manually editing metadata for 50,000 products. You are defining a system that generates sensible metadata for 50,000 products. You could also add AI to the process and use product attributes to generate more specific page titles. However, without manual checks this can become risky, so I would initially use it on priority pages where someone can verify the output.
2. On-page structure
Most modern ecommerce platforms produce reasonably structured HTML, and headings themselves should not be treated as some kind of magical ranking factor.
Still, the product page template should make the page easy to understand.
I normally want the product name to be the main H1, with descriptive sections underneath using logical subheadings. Product images should have useful alt text that describes what is actually shown. Where relevant, this will naturally include the product name.
For example:
- Product name as the main H1
- Product description sections using logical H2 headings
- Descriptive alt text on meaningful product images
- Important product attributes available as text, not hidden inside images
These may look like small details on one page, but a template-level change can improve thousands of pages at once.
That is what makes them interesting in large-scale SEO.
3. JavaScript and hidden product content
Large ecommerce websites often rely heavily on JavaScript or are built on JavaScript frameworks. Product variants, prices, availability, specifications, reviews and related products may all be loaded dynamically after the initial HTML has been delivered.
JavaScript itself is not an SEO problem. The problem begins when important product information depends on JavaScript being executed correctly before it becomes available to a crawler.
Google can render JavaScript, but you should not assume that AI crawlers and AI search systems handle JavaScript in the same way. In my own testing, I have seen AI-generated answers miss content that was only available after client-side rendering, even in cases where Google Search could render the page correctly.
Rendering adds another step between discovering a URL and understanding its content, and problems with scripts, APIs or rendering can result in important information being missed.
I would therefore try to make the core information of a product page available directly in the HTML whenever possible. This includes things such as:
- Product name
- Product description
- Price
- Availability
- Important product attributes and specifications
- Internal links
- Structured product data
Be particularly careful with content that is only loaded after a user interaction. If specifications, descriptions or internal links only appear after someone clicks a tab, changes a product variant or scrolls to a certain point on the page, you should verify that crawlers can actually access that information.
On large websites, this becomes another large-scale SEO problem. A JavaScript implementation issue in a single product template can affect tens or hundreds of thousands of URLs at once. It is therefore useful to compare the raw HTML with the rendered page and regularly test important product templates using crawling and rendering tools. It is also valuable to compare important product attributes in the raw HTML with the rendered HTML and identify anything that only becomes available after JavaScript execution.
This matters beyond traditional Google Search as well. Not every crawler or AI system can be assumed to execute JavaScript in the same way a modern browser does. If the information that defines a product only exists after complex client-side rendering, you are making it unnecessarily difficult for machines to understand what the page contains.
4. Internal links from product pages
Product pages are excellent places to build internal linking.
I have been involved in projects where there was discussion about bringing in a mathematician to design an algorithm for internal linking. You can go that far, but in many cases it is overkill relative to the effect you are trying to achieve. From a consultant’s perspective, though, I guess it would be a wonderful project to work on.
Internal links help crawlers understand the structure of an ecommerce website and discover related content. More importantly, they are also useful to actual visitors.
A product page should typically include some combination of:
- Breadcrumbs linking back through the hierarchy
- A link to the main category the product belongs to
- Links to relevant or similar products
- Products frequently bought together
- A link from the brand name to a dedicated brand page
- Links to the product support section
Large ecommerce sites should generally have useful brand pages that collect products from the same manufacturer. These pages can become valuable landing pages in their own right.
There is more than one way to work with internal links. You can also use internal linking strategically by identifying strong pages and using them to support weaker pages or segments you want to improve.
But the important large-scale principle is the same: if useful internal linking functionality is built into the product template, it automatically affects the entire product catalogue.
5. Product structured data
Product structured data should be a priority on a large ecommerce website.
The exact implementation depends on the business and product catalogue, but the basic idea is straightforward: make important product information explicit and machine-readable.
This can include information such as:
- Product name
- Brand
- Price
- Availability
- Ratings
- Product variants
- Shipping information
- Return information
This can make product pages eligible for richer product experiences in Google and helps search engines understand important commercial information about the product.
But I think structured product data is becoming interesting for another reason as well.
The cleaner and more consistent your product information is, the easier it becomes for machines to understand it. That applies not only to Google Search but also to AI-driven search systems trying to identify products, compare attributes and answer questions about what a company sells. With that said, there is still limited evidence showing to what extent AI systems actually use schema markup directly. That is not a reason to ignore it. Product structured data already has clear value in traditional search, and machine-readable product information may become even more important as AI-driven search develops.
Technical SEO and structured product data therefore form part of the foundation for AI visibility as well.
6. Performance and Core Web Vitals
Product pages should load quickly. That is hardly a controversial recommendation.
What makes performance particularly interesting in large-scale ecommerce SEO is that improvements to the product page template can affect every product page at once.
Product pages are often heavy. They may contain several large images, image galleries, personalization, recommendations, reviews, tracking scripts, third-party integrations and JavaScript-heavy functionality.
Improving the underlying template can therefore produce a much larger effect than optimizing a handful of individual pages.
From the user’s perspective, performance is obviously important as well. A product page is ultimately a commercial landing page, and making it faster and more stable benefits much more than SEO.
7. Product pages and E-E-A-T
How should you think about E-E-A-T on product pages?
I would not overcomplicate it at all!
A useful product description should demonstrate that the retailer actually understands the product. It should describe meaningful characteristics, specifications, benefits and limitations rather than simply repeating manufacturer copy.
If the product description is published on the brand’s own website, I would expect it to be one of the best sources of information about that product. It should cover the relevant specifications, characteristics, use cases and limitations in enough depth to demonstrate genuine first-hand product knowledge. Over time, that also contributes to the site’s authority around its own products.
Reviews are particularly valuable on large ecommerce websites because they provide genuine user experience around the product.
If possible, reviews can also include user-uploaded images and other information that makes them more useful to prospective customers.
This type of first-hand product information becomes even more valuable in an AI-heavy web. Generating another generic description of a product is easy. Producing unique information based on real customers, real product usage and proprietary product data is much harder.
How should product page copy be optimized?
For most product pages, the main query will still be the product name. I expect that to remain important for the foreseeable future. AI may increasingly capture the research phase of the customer journey, but when users are close to making a purchase, traditional search still plays an important role in finding the actual product, retailer, price and availability.
Besides the product name, there are usually many secondary queries that can also be relevant. On a small website, you could perform keyword research for each product individually. On an ecommerce website with tens of thousands of products, that is rarely realistic or commercially sensible.
Instead, I normally look at the information already available in the product data.
For example:
- Mention the product category naturally in the description.
- Use meaningful variants of the product name.
- Include relevant product attributes such as color, material, dimensions, capacity or format.
- Explain what the attributes actually mean for the customer.
The product description should ultimately be written for the visitor.
Think of it as a virtual salesperson. Its job is to explain the product, its characteristics, its benefits and why somebody might want to buy it.
This is also where AI can be genuinely useful at scale. If you have clean product data, AI can help turn structured attributes into readable product copy. But I would not let an AI model invent the product description from scratch. The source of truth should be actual product data, and the generated copy should be constrained by that data.
The scalable opportunity is not simply to “generate thousands of SEO texts with AI”. It is to build a system where reliable product data can be transformed into useful, consistent and human-readable information without inventing facts.
Finding secondary queries at scale
There is another way to approach secondary queries across a large product catalogue.
The process can look something like this:
- Crawl all product pages and extract the URL, H1 and relevant product attributes. AI can automate much of this process if it has access to the necessary crawl or product data.
- Connect the products to Search Console query data.
- Identify the queries each product page already receives impressions for.
- Compare those queries with the actual product data and attributes.
- Identify relevant queries where the correct product URL already ranks but performs outside the strongest positions.
- Prioritize opportunities based on search demand, position, impressions and commercial value.
- Use those opportunities to improve product copy or other relevant page elements.
Today, AI can make parts of this workflow considerably easier.
Instead of manually combining words from an H1 in a spreadsheet, a model can classify queries, connect them to product attributes and suggest where a query is semantically relevant to a product.
It can also help generate revised copy using the selected query and the underlying product data.
But the principle should remain the same: the data decides what should be optimized, AI helps scale the execution.
Conclusion
Product page SEO at scale is not about manually polishing tens of thousands of product pages.
It is about building a technically sound product page template, exposing reliable product information, creating useful internal relationships and using data to identify the relatively small number of products where additional manual optimization has real business value.
The same foundation increasingly matters outside traditional search as well. Search engines, shopping systems and AI-driven experiences all need to understand what a product is, how it relates to other products and what makes it useful.
AI can help scale parts of the work, but it does not remove the need for good technical SEO or good product data. If anything, it makes both more important.