ChatGPT and Claude prompts for Shopify product descriptions that AI assistants can read
More shoppers now ask an AI assistant what to buy. Nobody outside those companies knows exactly how their assistants pick products, but what reliably helps is a page whose facts are complete, consistent and machine-readable. These 7 prompts get you there without inventing anything.

One rule for every prompt: the AI must never invent specs, materials, certifications or claims. Anything it doesn't know stays as a [FILL] placeholder for you.
The 7 prompts
1. The first two sentences
OpeningSay what it is and who it's for. Slogans like "our best yet!" don't help a shopper or an agent.
Rewrite the first two sentences of this product description so they say what the product is, its main use and who it's for. No slogans. [paste description]2. A specs table
SpecsAgents quote specs: size and fit, weight, materials, capacity, compatibility, what's in the box.
Turn this description into a specs table for [category]. Use [FILL: …] for any fact that isn't in the text. Don't guess. [paste]3. Best for / not for
PositioningHonest positioning helps match the product to the right shopper.
Write one "Best for…" and one "Not for…" line for this product, using only facts from the page. [paste]4. 4 to 8 real FAQs
FAQSizing, care, compatibility, warranty, shipping, returns: answered in 1 to 3 sentences each.
Write 6 FAQs a shopper would ask before buying this. Answer only from the page; mark answers I must confirm as [CONFIRM]. [paste]5. SEO title and meta description
SEOTitle tag up to 70 characters, meta description up to 160.
Write an SEO title (max 70 characters: name + product type + one key attribute) and a meta description (max 160 characters). Show the character counts.6. Image alt text
Images"Black trail shoe outsole with 5 mm lugs", not "IMG_4432.jpg".
Write descriptive alt text for each of these product photos: [describe each photo]. Max 125 characters each.7. The structured data you're missing
JSON-LDPrice, currency, availability, GTIN or MPN, shipping and returns, in machine-readable form.
Here is my product page HTML [upload]. List which Product JSON-LD properties are missing (price, priceCurrency, availability, gtin/mpn, shippingDetails, hasMerchantReturnPolicy) and write only the missing ones, with [FILL] placeholders.Then run a 5-question agent test
Act as an AI shopping assistant. Using ONLY the facts on this product page [paste or upload], answer 5 questions a real shopper would ask about this product (fit or size, materials, delivery time to [country], returns, and how it compares for [use]). For each, say Answerable / Partly / Can't tell, and what's missing.Where to fix it on Shopify
- Barcodes and SKUs: on each variant.
- SEO title and meta description: the product's Search engine listing.
- Alt text: click each image in Media.
- Specs: metafields shown by your theme.
- Shipping and returns: Settings → Shipping and delivery, and Settings → Policies.
- JSON-LD: the theme's product section (ask your developer) or a structured-data app. Validate it with Google's Rich Results Test first, and don't add a second Product block that contradicts the first.
Menu names can vary by theme and version. None of this guarantees how any assistant ranks products; it makes your facts readable.
Everything here writes drafts: you review and send. No income or ranking promises. Not affiliated with or endorsed by Anthropic, OpenAI, xAI or Google.