Product description: how to write one, with examples and AI

In short: a product description is the text on a product page that says who the product is for, what problem it solves and which facts prove it. A good one has six parts: a first sentence with the main benefit, 3-5 benefits, specifications, use and care, answers to common questions and a meta description. An AI agent can write 1,000 of them from one template, with 5 samples for your approval before anything is saved.

What a product description is and why it decides the sale

A product description does the job of the sales assistant: it answers what the product is made of, how big it is and who it is for. When the answer is missing, the customer closes the tab.

The description also has a legal role everywhere in the EU. Under Article 6(1)(a) of the Consumer Rights Directive 2011/83/EU, an online seller must give the consumer "the main characteristics of the goods" before the contract is concluded. Directive (EU) 2024/825 amends the Unfair Commercial Practices Directive: misleading about environmental or social characteristics, durability or reparability becomes an unfair practice, and generic environmental claims you cannot prove are banned. Member states apply these rules from 27 September 2026. In practice, every claim needs a source: the label, the manufacturer or your own measurement.

The third role is visibility. Google and AI assistants such as ChatGPT, Gemini and Perplexity read the text to decide which searches the page answers and whether to quote it, in every language you sell in.

Product description structure: 6 parts

A product description works best in 6 parts, each answering one customer question, which keeps it easy to read on a phone.

Part	Question it answers	Length	Example (travel mug)
1. First sentence	What is it and who is it for?	Up to 25 words	"A 450 ml double-walled travel mug for coffee on the way to work."
2. Benefits	What do I get?	3-5 bullets	"The lid closes with one twist and does not leak in your bag."
3. Specifications	What is it made of, what size?	Table or list	Capacity, material, height, weight, colour
4. Use and care	How do I use and look after it?	2-4 sentences	"The mug goes in the dishwasher, the lid is hand wash only."
5. Common questions	Will it fit, will it work for me?	2-4 questions	"Does it fit a car cup holder?"
6. Meta description	Why should I click in Google?	About 150 characters	Product type, main benefit, one distinctive fact

A benefit and a feature are two different things. The feature is the fact ("double stainless steel walls"), and the benefit is what it gives the customer ("your coffee is still hot an hour later"). A good bullet holds both, with the feature as proof.

The short description is a separate field: the teaser next to the add-to-cart button or in product lists. In CloudCart you add it with "Add short description" in the product editor, below the main description. Write 1-2 sentences with the benefit and the key fact, without repeating the first sentence word for word.

How to write a product description: 7 steps

A product description is quick to write once the facts are collected.

1. Collect the facts. Label, spec sheet, your own measurements, care instructions. Write down only what you could show a customer who asks.
2. Describe the buyer in one sentence. "A commuter aged 30-45 who drinks coffee on the go." The buyer sets the tone and which benefits come first.
3. Pick the main search phrase. "travel mug 450 ml", "linen midi dress". Put it in the product name, the first sentence and the meta description, once each.
4. Write the first sentence. Product type, who it is for, main benefit. Skip "Introducing" and superlatives.
5. Turn features into benefits. For every fact, ask "so what?". The answer is the benefit.
6. Add the questions customers ask. Take them from chat, emails, reviews and returns. A return reason such as "runs small" is a question the description did not answer.
7. Check it on a phone. If the first 3 lines do not say what the product is and who it is for, rewrite the first sentence.

Product description examples by niche

The examples below follow one structure in 2 niches; the travel mug from the table above is the third. The products and figures are illustrations: in your store, use the facts from the label and the manufacturer.

Beauty: face serum

In beauty, the description gives ingredients with percentages, the skin type and how to apply it, and never promises a cure.

Write "hypoallergenic" or "dermatologically tested" only when the manufacturer gives them with a document.

Fashion: linen dress

In fashion, the customer cannot try the garment on, so give measurements in centimetres and the model's height.

Keep the size chart separate and the same for the whole category; size charts and variants are covered in the guide to the product catalog and variants. When you sell in several countries, give EU sizes and centimetres.

Product descriptions in several languages

A product description in a second language works when it is written for the local buyer, in the words that buyer types into Google. A German customer searches "Thermobecher 450 ml", a Polish one "kubek termiczny", and a literal translation often misses both the search phrase and the tone.

In CloudCart, each language gets its own language store through the Multilang app: a complete store with its own domain, currency, prices and couriers, which receives the catalog from your main store. Multilang copies the products and machine-translates names, descriptions, categories and meta fields, with no review queue before a translation goes live. So gaps in the main description carry over into every language.

1. Finish the main description first. All 6 parts, with measurements and care, before the first language store.
2. Find the local search phrase. For each country, put the phrase buyers use in the name, first sentence and meta description.
3. Review the top 20 products by hand. Units and sizes stay the same, while idioms and tone usually need a native touch.

Language stores, hreflang and prices per market are explained in the guide to the multilingual online store and on the multistore page.

SEO product descriptions: what Google looks at

An SEO product description is a unique text with the main search phrase in the product name, the first sentence and the meta description, while the rest answers the customer's questions.

• A unique text for every product. Keep five colours of one model as variants on one page with one text.
• The meta description sells the click. In CloudCart you write it in the product's "SEO Configuration" card, with a preview of how it looks in Google.
• The same facts everywhere. Price, availability and specifications in the text match the Product structured data the store sends to Google.
• The Google Shopping description is stricter. Under the Google Merchant Center requirements, the field holds up to 5,000 characters, the key details belong in the first 160-500, and price, sale dates, shipping, links and your company name stay out. Google also asks for AI-written descriptions to go in the separate structured description attribute, marked as AI-generated.

Category titles, internal links and technical settings are in the guide to online store SEO.

How a description ends up in ChatGPT and Gemini answers

AI assistants quote product pages where the answer sits in one self-contained sentence with a fact. "The travel mug is 18 cm tall and fits a car cup holder" gets quoted; "Perfect for every day!" gives the assistant nothing to quote.

1. Write sentences that stand on their own. Use the product name as the subject, without "it" that depends on the line before.
2. Give numbers with units. 450 ml, 115 cm, 30 °C, 12 months after opening.
3. Answer comparison questions. "For oily or dry skin?", "Warm enough for winter?". Assistants get exactly these.

More in the guide to GEO for online stores.

AI product description generator in the editor: Aura Studio

An AI product description in CloudCart is written by Aura Studio, inside the product editor, from the data already in your store: the description, the short description and the meta description. Generation uses tokens from your plan; on the pricing page the Cloudio AI row shows 10,000 for Baby Pack and 20,000 for Starter Pack, CC Pro and CC Master.

1. Open the product. Products → Products → the product. Below "Description" and "Short description" in "Details", and below the meta description in "SEO Configuration", is the AI bar (ShopperPen by CloudIO).
2. Choose length and style. Short, medium or long, and "Descriptive" (prose) or "Option list" (bullets).
3. Choose the source data. Name and category are always on; add properties, variant values, vendor, tags and the short description. A description over 3,000 characters cannot be used as source.
4. Add a photo or keywords. VisionSense AI picks up colour, material and finish from the image; RankMaster AI takes up to 5 keywords and adds links to the category and the brand.
5. Generate and review. The token estimate shows before you start. Each result has "Keep" and "Copy text", and earlier results stay in the product's history.
6. Save the product. "Keep" fills the field, and the product is stored only when you click Save. Read the text first: AI writes from the data it has.

For whole lists, Aura Studio runs many products as one task (ShopperPen Advanced, which needs at least 3,000 tokens in the balance). See also product upload and import.

How an AI agent writes 1,000 product descriptions

An AI agent writes 1,000 descriptions by reading each product's data from your store, writing from your template and saving only what you approved. The agent is Claude Code, ChatGPT Codex, Gemini CLI, Cursor or VS Code, connected with a personal access token (PAT) you create in the admin, as explained on the store run by an AI assistant page.

1. Create the token. Settings → PAT Tokens → a new token with only the products scope and an expiry date.
2. Ask for a list first. "Show me the products in Mugs with a description under 50 words or none. Only the count and the first 10."
3. Give a template per category. The six parts above, the buyer, the tone and the rules: "no superlatives, nothing that is not in the product data".
4. Approve 5 samples. "Write 5 descriptions and show them to me. Do not save." Fix the template until 5 out of 5 need no edits.
5. Run it in batches. "Save 50 at a time and after each batch list the names and which ones you skipped for missing data."
6. Check in the store. Open 5 random products per category on a phone.
7. Measure after 4 weeks. In Google Search Console, compare clicks and impressions with the period before.

The same template works for a language store: each language store is a store of its own, so an agent connected with a token from that store can rewrite the machine translations of your best sellers in the local search phrases. Backups to CSV before a bulk change and how to undo one are in the guide to bulk catalog changes with AI.

Important: according to Google's guidance on generative AI content, generating many pages without adding value for users may violate the spam policy on scaled content abuse. Google also asks you to fact-check AI text, meta descriptions included, before publishing.

Manual, Aura Studio or an AI agent: which one when

Write your top products by hand, use Aura Studio for products in the admin, and an AI agent for the whole catalog.

Method	When	Source data	Control
By hand	Top 20 products, new brands, expensive products	Your experience, customer questions	Full
Aura Studio in the editor	A new product, a refreshed description	Name, category, properties, variants, photo	You review each text before Save
Aura Studio for a list	A new category, a supplier import	The data of the listed products	Task and results in the CloudIO history
AI agent with a PAT token	The whole catalog, language stores, meta descriptions	Store data plus your template and files	Samples, batches, a report per batch

7 product description mistakes

• The supplier's text, copied as is. Use the manufacturer's facts and rewrite the text: a copy sits in every other store and gives Google no reason to pick yours.
• Properties the product does not have. AI invents them when it gets no data; put "only from the product data" in every task.
• Superlatives without a fact. "The best serum" answers nothing. "Niacinamide 10%" does.
• Unproven green claims. "Eco-friendly" without proof is banned across the EU from 27 September 2026. Name the fact instead, such as "100% recycled polyester, certified by the supplier".
• No measurements. A customer without a size in centimetres orders at random and returns within the 14-day withdrawal period.
• Price and shipping inside the text. They change and the description starts telling untruths; in Google Shopping they are not allowed there.
• The same meta description on every product. Google shows identical snippets and the customer sees no difference.

Google Shopping and the Facebook and Instagram catalog read the description from your store's catalog, so one good text works in every channel; see the sales channels page.

Frequently asked questions

Questions? Write to us at support@cloudcart.com.