How to write product descriptions with AI - a step-by-step workflow for digital sellers

Quick answer: For your 5 most important products, write descriptions one at a time with Claude using a detailed prompt with buyer context. For a catalogue of 30+ products, use Hypotenuse AI with a brand-voice guide to batch-generate at scale. Never publish AI copy without reading it aloud once.

Product descriptions are the highest-volume writing task most digital sellers face. A new Etsy shop with 50 listings has 50 descriptions to write. A Shopify store doing seasonal refreshes rewrites them quarterly. AI has made this tractable.

The difference between AI-written copy that converts and AI-written copy that does not comes down to what you put into the prompt.

What goes wrong with default AI product copy

Most sellers open ChatGPT or Claude, type "write a product description for [product name]", and publish whatever comes back. The result is generic, reads like every other listing, and converts poorly.

The problem is not the AI — it is the input. AI generates copy that reflects the specificity of what it is given. Vague prompt → vague copy.

Specifically, default AI product copy fails because:

• It describes features, not outcomes ("includes 20 pages" vs "gives you a week of meal plans in one sit-down")
• It uses safe, brand-neutral language rather than your actual voice
• It does not address the buyer's specific concern or context
• It uses filler phrases ("perfect for...", "ideal for...") that signal low-effort copy to experienced buyers

The information to gather before you prompt

For each product, before you open any AI tool, write down:

1. What the buyer is trying to solve or achieve — not what the product is, but what problem it solves or what outcome it enables
2. Who the buyer is — specifically. Not "women" but "Etsy shoppers buying a birthday gift for a friend, probably in the week before the birthday"
3. What makes this product different from the 10 alternatives they saw before finding yours — one specific thing
4. The most common objection to buying — what would make someone add it to cart but then close the tab?
5. Any specifics that matter — format (digital download, PDF, PNG), dimensions, compatibility, what software is needed

This takes 5 minutes per product. It is not optional.

The Claude prompt for one product

I am writing a product description for an Etsy listing. Here is the context:

Product: [name and a two-sentence description of what it is]
What it does for the buyer: [the outcome or problem solved]
Who buys it: [specific buyer description]
What makes it different: [one specific differentiator]
Main objection: [what might stop someone from buying]
Key specs: [format, dimensions, compatibility, delivery method]
Brand voice: [casual/warm/professional - add 1-2 adjectives]

Please write:
1. A title (max 140 characters for Etsy, or 70 for Shopify product title)
2. A short description (2-3 sentences, leads with the outcome, uses the buyer's language)
3. A full description (200-350 words, structured with a benefit paragraph, a features list, and a short note on what they receive and how)

Do not use: "perfect for", "ideal for", "elevate your", "transform your", or any phrase that could apply to any product.

Read the output aloud. If any sentence would sound strange said to a customer face-to-face, rewrite it.

The Hypotenuse AI workflow for bulk descriptions

For catalogues of 30+ products, Hypotenuse AI is faster than Claude because it has a batch import feature — you provide a CSV of product names and attributes, it generates all descriptions in one job.

Setup for bulk generation:

1. Create a brand-voice guide in Hypotenuse AI (paste 3-5 examples of your best existing copy — this calibrates tone)
2. Export your product data from Etsy or Shopify as a CSV
3. Map your CSV columns to Hypotenuse AI's input fields (product name, key features, category)
4. Run the batch job — Hypotenuse AI generates one description per row
5. Review the output CSV before importing — spot-check 10 descriptions from different parts of the catalogue

Batch descriptions will be more consistent in structure but less nuanced than individual Claude prompts. Use Hypotenuse AI for the catalogue; use Claude for your top 10 products where copy quality directly affects your conversion rate on high-ticket or high-competition items.

Platform-specific differences

Etsy listings: Title character limit is 140. The first 40 characters of the title carry more weight in Etsy's search algorithm — put your primary keyword early. Descriptions have no hard length limit but buyers on mobile rarely read past 100 words.

Shopify: Product titles should be under 70 characters for Google Shopping display. Descriptions are often shown in a truncated preview; put the most important benefit in the first sentence. HTML formatting is supported — bold your key benefit statement.

Gumroad: Gumroad product pages are minimal. Short, direct copy (under 150 words) outperforms long descriptions. The first line is your hook; make it about the outcome, not the format.

The editing pass that separates converts from non-converts

After AI generates your copy, do these three checks:

1. Read it aloud — catch robotic phrasing that reads fine but sounds wrong
2. Cut the opener — AI descriptions almost always start with a weak sentence. Delete the first sentence and see if the second works better as the opening
3. Make the CTA specific — "Add to cart" is the default. If you can add a deadline, quantity note, or outcome ("get your planner today and start next week"), do it

Jasper and ChatGPT both work for this task. Claude's advantage is following multi-part instructions more reliably across a full description — other tools are more likely to ignore the "do not use" list or miss the structure requirements. Test one product with your preferred tool before committing to a batch.