E-commerce Copy Prompt Builder

Structure is what turns a product into persuasive copy

Writing e-commerce copy that actually sells is less about a clever turn of phrase and more about giving the model the right raw material in the right shape. A product detail page on Taobao, Tmall, JD or Pinduoduo has to do several jobs at once: catch the eye in the main image, win the search query in the title, build desire through the selling points, and close the doubt in the short description. This builder keeps that structure for you. You fill in the product name, the concrete selling points and specs, the target shopper, the platform, the copy type you need, the tone you want, and the SEO keywords to weave in — and it assembles them into a single clean prompt, led by a clear instruction line and followed by clearly headed sections the model can read at a glance, ready to paste into any LLM.

The highest-leverage fields are the selling points and the shopper. Concrete selling points — "304 stainless steel", "72-hour battery", "machine-washable", "ships from a bonded warehouse" — give the model facts it can turn into benefits, instead of forcing it to invent vague praise. Naming the shopper ("value-conscious parents aged 25–35", "students buying their first laptop") tells the model which benefits to lead with and what tone lands. Pick the platform and the model adjusts instinctively: Pinduoduo copy leans into value and urgency, Tmall copy leans into brand and quality, and a Douyin hook is shorter and punchier than a JD detail block. The clearer each field, the less the model guesses and the more the copy reads like it was written by someone who knows the product.

Platform, shopper and keywords separate a draft from a converting listing

The fields sellers skip and regret are tone, copy type and keywords. Tone keeps the writing on-brand — warm and recommending for a lifestyle product, calm and technical for an appliance, young and playful for a trend item. Copy type focuses the output: ask for a title and you get a search-friendly title under the platform's character limit; ask for main-image copy and you get a six-to-twelve-character hook; ask for detail-page selling points and you get scannable bullets, each tied to a pain point. Keywords are what connect the listing to real demand — naming the words shoppers actually search lets the model weave them in naturally, so the title earns impressions and the detail page reinforces relevance, without keyword-stuffing that reads like spam.

Because the output is structured plain text, the same prompt is portable across every major Chinese model and works just as well on ChatGPT, Claude or Gemini, and you can reuse one set of selling points to spin out a title, an image hook, a detail block and a short description simply by changing the copy-type field. A word of caution worth building into the tone field: many marketplaces and advertising laws prohibit superlatives and unverifiable claims ("lowest price anywhere", "number one"), so ask the model to keep claims truthful and verifiable, then proofread before you publish. And because the whole tool runs locally in your browser, you can iterate freely — tighten a selling point, switch the platform, copy again — without anything you type ever leaving your device, being sent to a model, or being stored. Treat the first prompt as a draft, see where the copy drifts, and sharpen the matching field; two or three rounds usually turn a flat description into a listing that reads like it was written to sell.