Product Page SEO: Descriptions, Product Markup and Reviews That Rank and Sell

Every store, one paragraph, and no reason to pick yours

Most product copy arrives with the product. The supplier sends a spreadsheet, the description column goes straight into the store, and the same paragraph appears on every site that carries the item. Google's spam policies list, among their scraping examples, republishing content from other sites without adding original content or value, and reproducing content feeds from other sites without some unique benefit to the user. A supplier paragraph used with permission is not spam. It is simply no reason to rank you ahead of the other copies.

The fix is not cleverness. It is information the other stores did not bother to add: the measurement a buyer keeps asking about, the honest comparison with the cheaper model, the care instruction that prevents a return. The page that answers more of the buyer's questions has the better argument with both the customer and the search engine. Which searches each page should answer is settled earlier, in e-commerce keyword research; this guide is about the page itself.

A product page is a salesman who works every shift and never gets to ask the customer a question, so it had better answer all of them.

The title, the H1 and the first line do most of the ranking work

Shoppers who reach a product page usually typed something specific: a brand, a model, a size, a material. The page title and the H1 should carry that same information in the same order a buyer would say it. "Merino Hiking Sock, Midweight, Crew Height" beats "The Trailblazer" because it matches what people search and tells them at a glance they found it.

• Brand and model first when people search by name; the product type first when they search by kind.
• One defining attribute in the title (material, size, capacity, compatibility), not five.
• The H1 can match the title or run slightly longer; it should never contradict it.
• The first sentence of the description should say what the product is and who it is for, in plain words a search snippet can quote.
• Specs as HTML text, not a picture of a spec table, so shoppers, screen readers and crawlers can all read them.

Our guide to title tags covers length and rewrites, and on-page SEO covers the rest of the page's signals. The template that holds all this, from the gallery to the add-to-cart button, belongs to e-commerce website design.

Writing a description from facts a competitor cannot copy

A good product description answers, in order, the questions a knowledgeable clerk hears most. Write it from the spec sheet, the return reasons and the customer questions, not from adjectives. This order works for most catalogs:

1. 1

   Who it is for and what it does

   One or two sentences. The buyer decides in that time whether to keep reading.

2. 2

   The proof

   Materials, measurements, weights, capacities, test results you can document. Each one replaces an adjective.

3. 3

   How it differs from its neighbors

   Why this one and not the cheaper or the larger model you also sell. Shoppers compare anyway; help them do it on your page.

4. 4

   What goes wrong and how to avoid it

   Sizing advice, compatibility, care. Your return log and support inbox write this section for you.

5. 5

   What arrives in the box

   Parts, accessories, what is sold separately. It prevents the one-star review that starts with "No batteries included."

Length follows the product. A replacement filter needs a few lines and a compatibility list; a $900 stroller needs the full argument. The craft of the sentences themselves is in our copywriting guide, and longer buying advice belongs on a guide page that links back to the product, the subject of e-commerce blogs.

Product markup: merchant listings, product snippets and the fields that count

Google's Product structured data documentation describes two kinds. Product snippets are for pages where people cannot buy the product directly, such as editorial review pages. Merchant listings are for pages where customers can buy from you, with more room for details like shipping, returns and apparel sizing. A page that adds the required merchant listing properties is eligible for product snippets as well, so a store marks up its product pages as merchant listings.

Two warnings from the merchant listing guide deserve attention. Markup generated by JavaScript can make Shopping crawls less frequent and less reliable, which hurts when prices and stock change daily, so Google recommends putting Product markup in the initial HTML. And Google may try to verify merchant listing data before showing it. Google also says enhancements are shown at the discretion of each experience, so valid markup earns eligibility, not a guaranteed star rating. Google recommends sending the same products through a Merchant Center feed as well, saying both together maximize eligibility.

Sizes and colors: one product, several URLs, one canonical

Variants are where product pages quietly multiply. Google's product variant documentation groups them with a ProductGroup carrying three properties: variesBy (the attributes that differ, such as size or color), hasVariant (the individual Product items) and productGroupID (the parent SKU). Each variant needs its own identifier, such as a SKU or GTIN.

Pick the setup your platform supports and apply it everywhere. The common failure is a hybrid: variant URLs that exist but canonicalize to the wrong place, or selectors that change the photo without changing the URL. Our guide to canonical tags explains the signals, and WooCommerce stores with missing options can start with the fix for variations that are not showing.

Reviews: the copy you are not allowed to write

Customer reviews add what no store copy can: other buyers' experience, in their words, including the complaints. Google's review snippet guidelines support ratings on Product markup, but the reviews must be readily available on the marked-up page, and it must be obvious to visitors that the page has review content. Ratings must come from users.

The law is stricter than the markup rules. The Federal Trade Commission's rule on consumer reviews and testimonials, in effect since October 21, 2024, bans these practices among others:

• Creating, buying or spreading fake reviews, including AI-generated ones, or reviews from people with no real experience of the product.
• Offering a discount or gift in exchange for reviews that express a particular sentiment, positive or negative.
• Claiming the reviews on display represent all or most of the reviews submitted when negative ones have been suppressed.
• Using baseless legal threats or intimidation to stop or remove a negative review.

DataCram is not a law firm; have counsel check any review incentive before you launch it. The practical upshot is simple: ask every buyer, show every review, and reply to the bad ones. The online review statistics page collects what surveys say about how shoppers read them, and AI assistants weigh them too, as reviews and AI recommendations explains.

Photos that pass Google's checks and the shopper's squint

The photo is the closest a shopper gets to picking the product up. Google's merchant listing guidelines ask for image URLs that are crawlable and indexable, images that represent the product marked up, and multiple high-resolution photos of at least 50,000 pixels (width times height) in 16x9, 4x3 and 1x1 aspect ratios.

File names, alt text and formats are covered in our image SEO guide. Large photos are also the usual reason product pages load slowly on phones, which Core Web Vitals work fixes at the template level.

Using AI for product descriptions without a catalog of twins

Generative AI can draft a thousand descriptions before lunch. The danger is that they all sound alike and some of them lie. Google's guidance on using generative AI content says to fact-check and review all AI-generated content before publishing, and warns that generating many pages without adding value for users may violate its policy on scaled content abuse. Shopify's help page on generating descriptions adds that the merchant is responsible for accuracy, and that generated text can include product benefits you never listed.

1. 1

   Feed it the facts, not the product name

   Give the model the spec sheet, the label, the sizing notes and the top customer questions. A prompt with a name and nothing else invents the rest.

2. 2

   Forbid what you cannot prove

   No claims about health, safety, durability or certifications unless they are in the source data. Delete any that appear anyway.

3. 3

   Vary the structure by category

   A sock, a stove and a stroller need different sections. One template for the whole catalog produces the twin problem at scale.

4. 4

   Have a person check every page

   Read each draft against the product. Catch the invented feature before a customer does.

5. 5

   Label it in the feed

   Merchant Center says AI-generated titles and descriptions must go in the structured_title and structured_description attributes, marked trained_algorithmic_media.

Used that way, AI does the typing and the store supplies the knowledge. Our guide to AI writing covers the editorial side in more depth.

When the shopper asks a chatbot instead of a search box

Shoppers increasingly ask ChatGPT, Gemini or Google's AI Mode what to buy. Google's AI optimization guide says AI responses can include product listings and that Merchant Center feeds help products appear in them. OpenAI's commerce documentation says product feed onboarding for ChatGPT is open only to approved partners who apply. Either way, the machine reads the same facts this page has been asking for: clear names, real specs, accurate prices and honest reviews. The specifics are in ChatGPT shopping optimization and structured data for AI.

Rewriting a catalog is slow work, and it compounds. Start with the twenty products that earn the most, then the twenty with the most impressions and fewest clicks in Search Console. If the platform cannot render markup in the HTML or give variants their own URLs, the template is the problem, and our e-commerce development team builds custom stores at a fixed price starting at $3,500. For the rest of the store, the e-commerce SEO guide and the coming e-commerce SEO audit checklist pick up where this page stops, and a free site audit checks speed, SEO, mobile and security today.