Copywriter
AI Prompts for Product Descriptions
Written by PromptsByRole
Start with Step 1See an example
- Who it is for
- Freelance and in-house copywriters writing product pages for a brand's online store or catalog, who start from a spec sheet and need every line to match what the product actually is.
- Outcome
- A product description in the client's house format (opening line, benefit bullets, exact details block), a short version for marketplaces or product feeds, matching copy for other items in the same range, and a fact-check table that ties every claim to a spec line.
These prompts are plain text, so they work in whichever AI chat tool you use. Paste each step into the same chat so the AI keeps the context from earlier steps.
Step-by-step prompts
How to use it: replace the words in [brackets] with your details, paste Step 1 into your AI tool and answer any questions it asks, then paste the next steps into the same chat. Read and edit the result before you use it.
Step 1
Load the spec sheet and the rules
I'm writing product descriptions for a client's online store. Here is the product information (only what the client has cleared me to use): - Product name and range: [PRODUCT NAME, RANGE OR COLLECTION] - Specs, numbered, copied from the spec sheet: [1. MATERIAL, 2. DIMENSIONS, 3. WEIGHT, 4. CAPACITY, 5. CARE INSTRUCTIONS, 6. WHAT'S IN THE BOX, ...] - Claims the client has approved: [E.G. "HANDMADE IN SMALL BATCHES", OR "NONE GIVEN"] - Claims the client has ruled out or that need review: [LIST, OR "NONE GIVEN"] - Who buys it and why: [AUDIENCE AND THE SITUATION THEY BUY IT FOR] - Where the copy appears and its limits: [E.G. PRODUCT PAGE, MARKETPLACE LISTING UNDER 200 CHARACTERS] Use only these specs and approved claims. Never invent materials, measurements, certifications, awards, reviews, sustainability or health claims, or comparisons with other brands. Where a line would need a fact I haven't given, write [SPEC NEEDED: what's missing]. First, repeat the specs back as a table (Spec number | Spec | Value) and mark anything unclear as [CHECK].
Step 2
Turn each spec into a reason to buy
For each spec in the table, write one row: Spec number | What it is | Why this buyer would care | The question it answers for them (for example "Will it fit my bag?"). Use the audience I described. If a spec has no clear benefit for this buyer, say so instead of stretching it. Then pick the 3–5 rows that matter most to this buyer and say why in one line each.
Step 3
Draft the full description in the house format
Write the product page description in this format: - Opening line: one sentence under [25] words that says what it is and who it's for - Bullets: [3–5] benefit bullets from the rows you picked, each ending with its spec number in brackets so I can check it, e.g. [Spec 2] - Details block: dimensions, materials, care and what's in the box, copied exactly from the spec table, with units unchanged Write in the client's voice: [DESCRIBE THE VOICE OR PASTE A SAMPLE OF THEIR EXISTING PRODUCT COPY]. Use the words a shopper would type or say, such as [WORDS FROM THE CLIENT'S OWN SEARCH TERMS OR CUSTOMER REVIEWS, OR "NONE GIVEN"]. Don't convert units unless I ask.
Step 4
Write the short version and the rest of the range
Now do two things: 1. Write a short version for [MARKETPLACE OR PRODUCT FEED] under [200] characters, keeping the product name, the most important benefit and one key spec. 2. Here are the specs for other items in the same range: [PASTE NUMBERED SPECS FOR EACH ITEM]. Write a description for each in the same format. Keep shared facts (material, care, range name) worded identically across items, vary the opening lines so they don't all read the same, and list exactly how each item differs from the first one.
Step 5
Fact-check every claim before it goes to the client
Make a table of every claim across all the descriptions: Claim | Product | Spec number or approved claim, or "none". Rewrite or remove every claim marked "none". Flag words that usually need documentation, such as "eco-friendly", "sustainable", "non-toxic", "hypoallergenic", "waterproof", "organic", "made in", "lifetime", "best" or "#1", unless an approved claim covers them. Then list every [SPEC NEEDED] and [CHECK] item as a short question I can send the client.
Example of good output
Illustrative example, fictional brand and product; not real model output:
Specs given: Product: a stoneware travel mug. Buyer: commuters who take their coffee on the drive to work. 1. Stoneware body, glazed. 2. Holds 12 oz. 3. Silicone lid. 4. Dishwasher safe (mug only; wash lid by hand). 5. Not for microwave use with the lid on.
Opening line: A 12 oz stoneware travel mug for people who take their coffee from the kitchen to the car.
Bullets:
- Holds a full 12 oz pour [Spec 2]
- The mug goes in the dishwasher; wash the lid by hand [Spec 4]
- A glazed stoneware body rather than a plastic one [Spec 1]
Fact-check flag: "Keeps coffee hot for hours" was removed: no spec or approved claim covers heat retention. Question for the client: "Do you have heat-retention test results you can document? If not, we'll leave that out."
Tips and pitfalls
- Copy specs from the source document. Take measurements, materials and care instructions from the spec sheet or packaging, not from memory or an old listing. One wrong dimension leads to returns.
- Treat green and health words as claims. Words like "eco-friendly", "non-toxic" or "hypoallergenic" need documentation the client can show. If they can't, leave the word out and ask them, rather than softening it.
- Keep the range consistent. When items share a material or care routine, use the same wording on every page. Different wording for the same fact makes shoppers wonder if the products differ.
- Write for the shopper's question. "Will it fit?", "Is it easy to clean?" and "Is it a good gift?" are often more useful to a shopper than a list of features. Step 2 is where you find those questions.
- Edit in the client's voice. Read the draft against their existing pages and swap generic phrases for the words they already use. Cut any line that could describe any brand's product.
FAQ
How is this different from the landing pages playbook?
The landing pages playbook builds a whole page around one offer and one action. This one writes the descriptions that sit on individual product pages, often for many items in a range.
How is this different from the marketplace seller playbook?
That playbook is for people selling their own used or handmade items one listing at a time. This one is for copywriters writing a brand's catalog copy from a spec sheet, where accuracy and consistency across items matter most.
Can the AI pick keywords for the descriptions?
It can suggest plain phrasings a shopper might use, but it can't tell you how often people search for them. Use the client's own search terms, site search data or customer reviews in Step 3.