How I Use Claude to Write 50 Product Descriptions in 30 Minutes

I used to spend three hours every Monday morning writing product descriptions. Fifty SKUs, each needing a unique title, bullet points, and a paragraph of body copy. By description number twenty I was recycling the same phrases, my brain was mush, and the quality was visibly dropping. The last ten descriptions always read like they were written by someone who hated the product.

Then I built a workflow in Claude that gets all fifty done in about thirty minutes. Not rough drafts that need heavy editing -- polished, platform-ready descriptions that only need a quick human pass. This guide is the exact process I use every week, with the actual prompt templates you can copy and adapt for your own catalogue.

Why Claude for product descriptions

I have tested every major AI writing tool for product descriptions over the past eighteen months. ChatGPT, Jasper, Copy.ai, Writesonic -- all of them. Here is why I keep coming back to Claude for this specific job.

Context window size matters more than you think. Claude's 200K token context window means I can paste in fifty product briefs, my brand voice guidelines, formatting rules, and example descriptions all in a single conversation. ChatGPT-4o has a 128K window but tends to lose coherence around description thirty. Claude stays consistent through to the end of the batch.

Tone consistency across a batch. When you give Claude a brand voice sample at the top of the conversation, it maintains that voice remarkably well across fifty outputs. I have found ChatGPT drifts more noticeably, especially on longer runs. It starts adding filler phrases and generic marketing language around the halfway mark.

Formatting compliance. Amazon bullet points need specific formatting. Shopify descriptions need HTML. Etsy listings need tags. Claude follows structural instructions more reliably than any other model I have tested. When I say "five bullet points, each starting with a capitalised benefit keyword, no bullet exceeding 200 characters", Claude delivers exactly that almost every time.

Projects feature for persistent context. Claude Projects let me store my brand voice guide, formatting rules, and example descriptions as persistent context. I do not have to re-paste everything each session. This alone saves me ten minutes per batch and improves consistency across weeks.

That said, Claude is not perfect for every situation. I will cover its limitations later in this guide.

The setup: what you need before you start

Before you open Claude, you need three things prepared. Skipping this step is the number one reason people get mediocre AI-generated descriptions.

1. Product data spreadsheet

Create a simple spreadsheet with one row per product. The columns I use are product name, key features (three to five per product), target customer, price point, and any unique selling points. Export this as a CSV or just copy and paste the text.

The more specific your input data, the better your output. "Stainless steel water bottle" gives you generic copy. "750ml double-wall vacuum-insulated stainless steel water bottle, keeps drinks cold 24 hours, fits standard car cup holders, powder-coated matte black finish" gives you copy that actually sells.

2. Brand voice document

Write a short brand voice guide -- two hundred words is plenty. Cover your tone (casual, professional, playful, technical), words you always use, words you never use, and your target audience. Here is a stripped-down example:

BRAND VOICE GUIDE - [Brand Name]

Tone: Confident but not aggressive. Friendly but not cutesy. Expert but accessible.
Audience: Health-conscious millennials, 28-40, willing to pay premium for quality.

Always use: "crafted", "designed for", "built to last"
Never use: "cheap", "best on the market", "game-changer", "revolutionary"
Never use: emojis in product titles or bullet points

Style: Short sentences. Active voice. Benefits before features.
Example line: "Keeps your coffee hot for 12 hours -- so your 3pm cup tastes as good as your 7am brew."

3. Platform formatting rules

Each marketplace has different requirements. Write these out explicitly so Claude can follow them. I keep a separate document for each platform.

The workflow: step by step

Here is my exact process from start to finish. The whole thing takes twenty-five to thirty-five minutes depending on how many products I am processing.

Step 1: Set up the conversation (2 minutes)

If you are not using Claude Projects, start by pasting your brand voice guide and platform rules. Then send this framing prompt:

You are an expert e-commerce copywriter. I am going to give you product data for
50 products. For each product, write a complete product listing following the
formatting rules and brand voice guide above.

Process all 50 products in order. Number each output so I can match it to my
spreadsheet. Do not skip any product. If a product's data seems incomplete, write
the best listing you can with what is available and flag it with [REVIEW NEEDED].

Maintain a consistent brand voice across all 50 listings. Do not use filler
phrases like "elevate your experience" or "take it to the next level". Every
sentence should communicate a specific benefit or feature.

Step 2: Feed in product data (3 minutes)

Paste your product data. I typically send it in batches of ten to fifteen products per message, not all fifty at once. This gives Claude a natural breakpoint and lets me spot-check quality before continuing.

Here are products 1-15. Write complete listings for each following the rules above.

1. Product: Bamboo Cutting Board Set (3-piece)
   Features: Organic bamboo, juice grooves, non-slip silicone feet, three sizes
   (small/medium/large), food-safe mineral oil finish
   Target customer: Home cooks upgrading from plastic boards
   Price: $34.99
   USP: Only set on market with built-in juice grooves on all three sizes

2. Product: Stainless Steel Mixing Bowls (5-piece)
   Features: 18/10 stainless steel, nested for storage, non-slip base,
   measurement markings inside, 1L to 5L range
   Target customer: Home bakers and meal preppers
   Price: $29.99
   USP: Measurement markings eliminate need for separate measuring cups

[...continue for all products in the batch]

Step 3: Review and iterate (10-15 minutes)

This is where the time goes, and it should. Read through the first batch of outputs. Look for three things: accuracy (did Claude get the features right), tone (does it match your brand voice), and formatting (does it meet platform requirements).

If something is off, correct it immediately before sending the next batch. For example:

Good output on products 1-15. Two adjustments for the remaining products:
- The bullet points are averaging 220 characters. Keep them under 200.
- Tone is slightly too formal. Use more contractions (you'll, it's, won't).

Now process products 16-30.

Step 4: Export and format (5 minutes)

Once all fifty descriptions are generated, I copy the output into my spreadsheet. If you need HTML formatting for Shopify, add that instruction to your original prompt and Claude will wrap everything in the right tags.

My prompt templates

These are the actual templates I use every week. Copy them, adjust the brand voice sections, and they work out of the box.

Amazon listing template

Write an Amazon product listing for the following product. Follow this exact format:

TITLE: [Product name] - [Primary benefit] - [Key feature] - [Secondary benefit]
(Keep under 200 characters. Front-load the main keyword.)

BULLET POINTS (5 total):
- Each bullet starts with a CAPITALISED BENEFIT KEYWORD followed by a colon
- Each bullet is under 200 characters
- Bullet 1: Primary benefit/USP
- Bullet 2: Key material/quality feature
- Bullet 3: Practical use case
- Bullet 4: What is included/dimensions/specs
- Bullet 5: Guarantee or trust signal

PRODUCT DESCRIPTION (150-200 words):
- Opening line states the core problem this product solves
- Second paragraph covers features with benefits
- Third paragraph paints a picture of the product in use
- Final line is a soft call to action
- No HTML tags
- No superlatives like "best" or "perfect"

SEARCH TERMS (5 comma-separated backend keywords not already in the title):
- No brand names, no competitor names, no subjective claims

Product data:
[paste product details here]

Brand voice reference:
[paste your brand voice guide here]

Shopify product description template

Write a Shopify product description for the following product. Output valid HTML
that I can paste directly into the Shopify rich text editor.

FORMAT:
<h2>[Benefit-driven subheading, not just the product name]</h2>
<p>[Opening hook: 1-2 sentences addressing the customer's problem or desire]</p>
<p>[Feature-benefit paragraph: 3-4 sentences weaving features into benefits]</p>
<ul>
  <li>[Spec 1: dimension/material/quantity]</li>
  <li>[Spec 2]</li>
  <li>[Spec 3]</li>
  <li>[Spec 4]</li>
</ul>
<p>[Closing line: reinforce the USP and include a subtle urgency element]</p>

RULES:
- Total word count: 120-180 words
- Tone: conversational, second-person ("you" and "your")
- No generic phrases like "look no further" or "whether you're a beginner or pro"
- Include the primary keyword naturally in the first paragraph
- Every feature must be paired with a benefit

Product data:
[paste product details here]

Etsy listing template

Write an Etsy product listing for the following handmade/artisan product.

TITLE (max 140 characters):
[Primary keyword] - [Descriptive phrase] - [Use case or occasion] - [Material]
(Front-load searchable terms. Etsy search prioritises early words.)

DESCRIPTION:
Paragraph 1 (2-3 sentences): What makes this product special. Speak directly to
the buyer. Mention it is handmade/made to order if applicable.

Paragraph 2 (2-3 sentences): Materials, process, or story behind the product.
Etsy buyers care about craftsmanship.

Paragraph 3: Specifications as a simple list
- Dimensions:
- Materials:
- Care instructions:
- Processing time:

Paragraph 4 (1-2 sentences): Shipping info and a warm closing line.

TAGS (13 tags, comma-separated):
- Mix of broad and specific keywords
- Include material, style, occasion, and recipient tags
- No tags that repeat words already in the title

Product data:
[paste product details here]

Bulk processing wrapper prompt

This is the wrapper I use when processing a large batch across any platform:

I am going to give you [NUMBER] products to process. For each product, follow the
[PLATFORM] template I provided above.

BATCH RULES:
1. Number each output to match my input numbering.
2. Maintain consistent tone across all products -- no drifting into generic
   marketing language as the batch progresses.
3. Vary your sentence structures. If product 5 opens with "Crafted from...",
   product 6 should not. Aim for at least 5 different opening patterns across
   every 10 products.
4. Flag any product where the input data seems insufficient with [REVIEW NEEDED].
5. If you reach your output limit, stop cleanly at the last complete product and
   tell me which product number to continue from.

Begin with products 1-[BATCH SIZE].

Editing and quality control

AI-generated product descriptions are first drafts, not final copy. Even with excellent prompts, you need a human editing pass. Here is what I check and how long it takes.

AI tells to hunt for. These are phrases that scream "a robot wrote this" and will erode buyer trust. Watch for "whether you're a seasoned pro or just starting out", "elevate your [noun]", "say goodbye to [problem]", "look no further", and "takes your [noun] to the next level". I search-and-destroy these across every batch.

Factual accuracy. Claude occasionally invents features or slightly misquotes specs. Cross-check dimensions, materials, and any numerical claims against your product data. This is non-negotiable -- a wrong dimension on Amazon can trigger a return and a negative review.

Keyword placement. Confirm your target keywords appear naturally in titles and first paragraphs. Claude usually handles this well if you specify keywords in the prompt, but occasionally buries them in the third paragraph instead.

Brand voice drift. Read description number one and description number forty-five back to back. If the tone has shifted, that last batch needs a rewrite. This happens less with Claude than with other models, but it still happens on very large batches.

My editing pass takes about ten to twelve minutes for fifty descriptions. I have a checklist template in a spreadsheet that I run down for each one.

Results: before and after

Here is a real comparison from my own catalogue, showing the difference between my old manual process and the Claude workflow.

The biggest win is not just speed -- it is consistency. When I was writing manually, the last fifteen descriptions in any batch were noticeably weaker. With Claude, description number fifty reads as well as description number one.

Cost breakdown: Claude vs the alternatives

Claude Pro costs $20 per month and includes generous usage for product description work. For most sellers processing under 200 descriptions per week, you will never hit the usage cap. If you need higher volume, the Claude API charges roughly $3 per million input tokens and $15 per million output tokens -- which works out to about $0.02 to $0.05 per product description.

For context, a freelance copywriter charges $5 to $25 per product description. A virtual assistant doing basic rewrites charges $2 to $5 each. Claude at scale costs pennies.

When Claude is not enough

Claude is excellent for product descriptions, but it is not the right tool for every situation. Here is when I recommend something else.

If you hate writing prompts. Claude requires you to be a decent prompt engineer. You need to write clear instructions, provide examples, and iterate on your templates. If that sounds tedious, a tool like Jasper or Copy.ai gives you pre-built templates where you just fill in the blanks.

If you need built-in SEO scoring. Claude writes descriptions but does not score them. It cannot tell you your keyword density, readability score, or how your listing compares to competitors. For that, you need a dedicated listing optimisation tool. I use a listing scorer alongside Claude to check the output.

If you need team collaboration. Claude Projects supports sharing, but it is not built for content teams with approval workflows. Jasper and Copy.ai both handle multi-user workflows better, with commenting, version history, and role-based access.

If you want one-click marketplace integration. Claude outputs text that you copy and paste. Tools like Writesonic and some Amazon-specific tools can push descriptions directly to your listings. For high-volume sellers updating hundreds of SKUs, that integration saves real time.

For a broader comparison of AI writing tools for Amazon sellers, see our best AI tools for Amazon FBA guide. Shopify sellers should check our best AI tools for Shopify roundup.

Head-to-head comparison: Claude vs Jasper vs ChatGPT vs Copy.ai

Advanced tips for better output

Once you have the basic workflow running, these tweaks will push your description quality higher.

Use few-shot examples. Instead of just describing what you want, show Claude two or three examples of perfect descriptions from your catalogue. Paste them at the top of your conversation with a note saying "Here are three examples of our ideal product descriptions. Match this style exactly." This consistently produces better output than instructions alone.

Specify what not to write. Negative instructions are surprisingly powerful. "Do not use the word 'premium'. Do not start any bullet point with 'This'. Do not use more than one exclamation mark per listing." Claude follows these constraints reliably.

Rotate your opening patterns. One of the biggest giveaways of AI-written copy is repetitive structure. Explicitly tell Claude to vary its openings: "Use at least five different sentence structures for opening lines across this batch. Track which patterns you have used and avoid repeating them within any ten-product window."

Run a second pass for keyword optimisation. After generating descriptions, send them back to Claude with your target keywords and ask it to naturally incorporate any missing terms. This two-pass approach gives you more natural-sounding keyword integration than trying to stuff keywords into the initial prompt.

Keep a swipe file of winning descriptions. When a product description leads to higher conversion rates, save it. Add it to your Claude Project as a reference example. Over time, your few-shot examples will be drawn from your actual best performers, not hypothetical ideal copy.

Frequently asked questions