Product page SEO: descriptions, reviews, Product schema and variants

A product page has an awkward job. It has to rank for the product's name and attributes against every other shop selling the same item, often with identical photos and the same manufacturer description. It has to convince a buyer who may have arrived from a search, a comparison site or an AI assistant. And it has to give search engines clean, structured information about price and availability. The pages that win usually do so on details: things only the seller who actually handles the product can provide.

What product pages rank for

Product pages typically attract three kinds of queries:

• Exact product searches: the model name, sometimes with a code or colour. Here you compete directly with other retailers and the manufacturer.
• Attribute searches: "merino hiking sock cushioned crew". A product page can rank if its content covers those attributes clearly.
• Problem and comparison searches: "is X waterproof", "X vs Y". Product pages rarely rank first for these, but a page that answers them clearly can, and it can be cited in AI answers.

Category pages usually rank better for broad queries; see the category pages guide. The product page's job is to own its product and close the sale.

Descriptions: add what the manufacturer did not

Manufacturer descriptions are written once and copied by every retailer. Search engines group such pages and pick a few to show, usually the strongest sites. A smaller shop needs a reason to be chosen, and the description is the most obvious place to give one.

A practical structure for a rewritten description:

1. One or two sentences on who it is for and what it is best at. "A cushioned crew sock for long days in boots, warm enough for autumn hikes but thin enough for summer trail shoes."
2. Specific details from handling the product: how the sizing runs, what the material feels like, how it compares with the previous model or a popular alternative.
3. Specifications in a table: material, weight, dimensions, care. Tables are easy to scan and easy for machines to read.
4. Answers to the questions customer service gets: "Does it shrink?", "Is the left/right shaped?", "Will it fit a wide foot?"

You do not have to rewrite thousands of descriptions at once. Start with the products that bring the most revenue or have the most impressions in Search Console, then work down. For long-tail products, adding a sizing note and a specification table to the standard text is already an improvement.

Images and video

Original photos are one of the clearest signals of a real retailer. Shots of the product in use, close-ups of details, size comparisons and photos taken by customers all help buyers and give image search something unique to index. The image SEO guide covers alt text, formats and sizes.

Mark the main product image as eager-loading with high priority; it is usually the page's Largest Contentful Paint element. Gallery images further down can be lazy-loaded. A short product video helps conversion; if you add one, follow the video markup rules so it can appear in video results.

Reviews

Customer reviews do three things: they add fresh, unique text in the language buyers use, they answer questions you did not think of, and they make the page eligible for review stars in search results when marked up correctly. A few practical points:

• Render reviews in the HTML, not only in a script-loaded widget. Many review apps load reviews from their own servers after the page renders; check that at least the first few reviews and the rating are in the page source or the rendered HTML Google sees.
• Ask for reviews at the right moment: after the product has been used, not the day it ships.
• Respond to negative reviews on the page. It shows a real business and answers the concern for future buyers.
• Never invent reviews or mark up ratings that are not shown on the page. It breaks consumer protection rules in many countries and Google's structured data policies.

Product structured data

Product markup tells search engines the product's name, image, price, currency, availability and rating in a machine-readable form. It powers price and availability snippets, review stars, and eligibility for shopping-related features. A minimal example in JSON-LD:

{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Merino Cushioned Crew Hiking Sock",
  "image": ["https://example.com/img/sock-crew-1200.jpg"],
  "description": "Cushioned merino crew sock for hiking boots.",
  "sku": "MCS-CREW-GRY",
  "brand": {"@type": "Brand", "name": "ExampleBrand"},
  "offers": {
    "@type": "Offer",
    "url": "https://example.com/products/merino-crew-sock",
    "priceCurrency": "USD",
    "price": "24.00",
    "availability": "https://schema.org/InStock"
  }
}

Keep the markup consistent with the visible page: the same price, the same availability, the same rating. Mismatches are a common reason for rich results disappearing. Validate with the Rich Results Test, and check the product-related reports in Search Console for errors across the site. Also see structured data and Open Graph for the general rules.

Variants: one URL or many?

Situation	Recommended
Sizes of the same item	One URL; size chosen on the page
Colours that people do not search for by name	One URL; colour chosen on the page, optional ?colour= parameter with canonical to the main URL
Colourways or editions searched for by name ("X in forest green")	Separate URLs can work, each with unique images and a self canonical
Materially different products sold under one name	Separate products

Whichever you choose, make sure variant selection does not create crawlable URLs that all say the same thing with different canonicals. If your platform uses a variant parameter, the canonical should normally point to the main product URL. Google also supports markup for product variants (product groups), which can help it understand how variants relate.

Price, stock and trust details

Buyers look for shipping costs, delivery times and return policies before they buy, and search engines increasingly show this information in results. Show them clearly near the price, not only on a separate policy page. If you use Merchant Center, keep feed data consistent with the page. Availability changes are covered in the out-of-stock guide.

Questions and answers on the page

A short questions section on the product page is one of the cheapest improvements available. Pull the real questions from customer service emails, chat logs, returns reasons and reviews. "Does this run small?", "Can it go in the dryer?", "Will it fit a 15-inch laptop?" Each answer removes a reason not to buy, adds specific text that other retailers do not have, and gives search engines and AI assistants a clear statement they can reuse when someone asks the same thing.

Keep answers short and factual, and update them when the product changes. If the same question appears across a whole category ("how should hiking boots fit?"), answer it once on the category page or in a guide and link to it, rather than pasting the same paragraph onto two hundred product pages.

Returns data is especially honest. If many people return a jacket because the sleeves are long, saying so on the page will cost you a few sales from people who would have returned it anyway, and earn you trust and fewer returns from everyone else.

Internal links from and to the product

• Breadcrumbs back to the category, with BreadcrumbList markup.
• Links to related accessories and alternatives, chosen by a person or by sensible rules rather than random "you may also like".
• Links from guides and comparison articles that mention the product.

A product page checklist

1. Title: product name plus the key attribute, brand and shop name.
2. Description rewritten for top products; specification table for all.
3. Original images, main image not lazy-loaded, descriptive alt text.
4. Reviews rendered in HTML and marked up only if shown.
5. Product and Offer markup valid and matching the page.
6. Variants on one URL unless separately searched for.
7. Shipping, returns and delivery time visible near the price.
8. A quick technical check of the product template for canonical, structured data and response time.