Product detail page prompts solve a very specific problem: the specs are already on your desk, yet the listing copy never gets written, or it comes out as filler like "premium quality you can trust" that could describe anything. Copy the prompt below, fill in the product name, core selling points, specs and target buyer, and the AI will return a title, selling-point paragraphs, a use scenario and a spec list in the order shoppers actually read them. Check a few numbers and it is ready to publish.
When this prompt works best
It fits products that already have real specs and selling points and only lack someone to turn them into copy, such as home goods, small appliances, beauty and personal care, and food. Shoppers on a detail page want to confirm three things: what it is, what problem it solves for them, and whether the specs are right. The prompt keeps those three apart so the AI does not stir selling points and parameters into one rambling paragraph. If the selling points themselves are still unclear, do not force the copy yet. Send the spec sheet first and ask the AI to tell you what information is missing; that works far better than writing blind.
The prompt, ready to copy
You are an e-commerce copy editor. Based on the information I give you, write product detail page copy for one product.
Product name: {{product_name}}
Core selling points: {{selling_point_1}}; {{selling_point_2}}; {{selling_point_3}}
Key specs: {{material_or_ingredients}}, {{size_or_capacity}}, {{scope_of_use}}
Target buyer: {{target_buyer}}
Never write: {{banned_claims}}
Output requirements:
1. Start with a one-line product title, no more than 30 words, containing the product name and the single strongest selling point, without keyword stuffing.
2. Then write 3 selling-point paragraphs. In each one, state the benefit to the buyer first, then support it with a spec. Keep each paragraph under 60 words.
3. Next, write one use-scenario paragraph saying who uses it, in what situation, and how, without empty praise.
4. Finish with a separate spec list. List the key specs one by one, with numbers and units exactly as I gave them. Do not change or invent any figure.
5. Never use absolute claims such as "best", "number one" or "100%", and never invent test data, sales figures or customer reviews.How to fill in the variables
- Product name: use the full name with its category word, for example "collapsible silicone food container", not just the brand.
- The three selling points: order them by how much buyers care. Put the most convincing one first; the AI naturally gives later points less weight.
- Key specs: enter only figures you have verified. Leave an efficacy spec blank if no test report supports it, instead of letting the AI guess.
- Target buyer: name a specific person and moment, for example "a working mother packing lunch for her child". The more specific this is, the less generic the scenario paragraph becomes.
- Banned claims: list the words your platform rules and brand guidelines forbid, such as absolute claims and unverified efficacy. Listing them up front is much easier than deleting them afterwards.
Fix these three things before you publish
First, check every number. An AI will occasionally turn "500 ml" into "a generous 600 ml", so compare the spec list against your sheet word by word. Second, align the selling-point order with your images: whatever the main image shows, the first paragraph should say, so shoppers do not have to match text and pictures themselves. Third, if the title contains a word that appears nowhere in your specs, delete it. Do not keep a promise the product cannot fulfil just for search traffic.
When the same product moves to another platform, you do not need to rewrite the whole prompt. Only change the length and paragraph counts in the output requirements: keep the full spec list for marketplace listings, and on content platforms move the use scenario up and shrink the specs to one line. To keep social posts and the detail page consistent, see Xiaohongshu Note Prompt: Fill In the Product and Real Experience for a Post That Does Not Read Like an Ad, then bring the same set of selling-point variables back into this prompt so the two pages never contradict each other.
