What is a product description in an online store?

The product description is the text read by someone who has already reached your product page and hasn't decided yet. It is not the meta description, which summarizes the page for search results, nor the manufacturer's PDF spec sheet, nor the list of product attributes, though it draws on them. It is the place where the page says what an attribute table can't: who the product is for, when it helps and where it falls short. And today it has three readers: the shopper, Google and the AI assistants that answer questions about products.

What it includes

On the page and in the markup. In Schema.org, description is a property of any thing, defined as a description of the item, and it takes text.⁠^{[1]Schema.org. description} In Product markup it is the field that carries that same description, and the sensible thing is for it to say what the page's visible text says.

In the Google feed. Google Merchant Center requires a description for every product, from 1 to 5,000 characters, and advises putting the most important details in the first 160 to 500 characters. It suggests including size, material, the age group the product is for, special features, technical specifications, shape, pattern, texture and design, and asks you to be specific and accurate.⁠^{[2]Google Merchant Center Help. Description [description] and structured description [structured_description]} It also says what the description can't carry: comparisons with or details of other products, links to your store or other sites, promotional text such as price, sales or shipping, mentions of compatible products or accessories, capital letters used for emphasis, and references to category systems.⁠^{[2]Google Merchant Center Help. Description [description] and structured description [structured_description]}

The generative AI case. Merchant Center separates descriptions by who writes them: every description created with generative AI has to be submitted in the structured_description attribute, not in description.⁠^{[2]Google Merchant Center Help. Description [description] and structured description [structured_description]} If your store generates or rewrites descriptions with a language model, that rule affects how you send them to Google, not just how you publish them.

For AI assistants. In its recommendations for appearing in AI search answers, Microsoft gives an example that holds for any description: a page should say "42 dB dishwasher built for open-plan kitchens", not just "quiet dishwasher". It also recommends lists and tables, which break details into pieces that are easy to reuse, and writing for the question the shopper asks, not only for the keyword. And it warns of three places where information gets lost: tabs and accordions, because AI systems may not read hidden content; PDFs, which lack the structure signals of HTML; and images, because pulling text out of an image is harder and less accurate.⁠^{[3]Microsoft Advertising (Krishna Madhavan, October 8, 2025). Optimizing Your Content for Inclusion in AI Search Answers}

In a store

A fictional example of before and after, for a desk lamp. The starting description:

Modern, elegant LED lamp design, perfect for any space. FREE SHIPPING in 24 hours. Also discover our light bulbs.

What fails, sentence by sentence:

The corrected version says what it is (a desk lamp with an articulated arm and an adjustable head), what the light is like (dimmable, from warm to cool), what desk it suits (with the arm's dimensions in the text, not just the photo), what keeps it stable (a weighted base) and what it doesn't include (the clamp that fixes it to the edge of the desk). Every sentence answers a question a shopper could ask, and none depends on an image to be understood. The Merchant Center rules belong to the feed field, not the page, but if the same description feeds both, writing it once with no promotional text saves you from maintaining two versions.

How to check it

1. Read the first 500 characters of your description and ask whether they say what the product is and who it is for. It is the part Merchant Center asks you to reserve for the most important details.⁠^{[2]Google Merchant Center Help. Description [description] and structured description [structured_description]}
2. Look for adjectives with no fact behind them ("durable", "quiet", "high quality") and swap them for the number or fact that backs them up.⁠^{[3]Microsoft Advertising (Krishna Madhavan, October 8, 2025). Optimizing Your Content for Inclusion in AI Search Answers}
3. Open the product page without clicking anything and check that the key information is in view, not inside a tab, an accordion, a PDF or an image.⁠^{[3]Microsoft Advertising (Krishna Madhavan, October 8, 2025). Optimizing Your Content for Inclusion in AI Search Answers}
4. Compare the page description with the feed one: no links, no price, no shipping, no capitals for emphasis and no other products.⁠^{[2]Google Merchant Center Help. Description [description] and structured description [structured_description]}
5. If an AI generated the description, confirm it reaches Merchant Center in structured_description and not in description.⁠^{[2]Google Merchant Center Help. Description [description] and structured description [structured_description]}

Related concept: meta description

The meta description is a tag in the code that summarizes the page for search results. Google asks for it to be unique on every page and, on a product page, to include concrete details such as price, manufacturer or availability; it has no length limit, though results trim it to the width of the screen.⁠^{[5]Google Search Central. Control your snippets in search results} The product description, by contrast, is page text, read by everyone who lands there, and it is the one that feeds the markup and the feed. The meta description tries to get someone to click; the description tries to get them to buy once they are in. A good description gives you material to write the meta description from; it doesn't work the other way round.

To keep going, the catalog and entities reading path puts the other terms in order, and the catalog enrichment entry explains how the data a good description is built from gets filled in.