8 ChatGPT Prompts for Amazon Sellers (Copy and Paste)

"So what do I ask it?" That's the question we hear most after a seller connects ChatGPT or Claude to their Amazon seller account. Not because the data is missing, but because an empty text box gives you nowhere to start. The result is vague questions ("how are sales going?") and vague answers. This guide gives you 8 ChatGPT prompts for Amazon sellers, grouped by area, plus the rules for writing your own prompts that actually work.

Why most Amazon prompts don't help

The "100 ChatGPT prompts for Amazon" lists out there are mostly about writing titles and descriptions. Useful, but that's a small slice of a seller's week. The real work is operational: figuring out what changed, finding problems, deciding what to fix first. For that, the model needs to see your data.

Two things to be clear on:

Without a connection, ChatGPT and Claude only work on what you paste in. The prompts below still work, but you'll need to export and paste the right reports.
With your seller account connected through an MCP connector (such as Vendooly, which uses official Amazon APIs), the model reads orders, stock, Buy Box, listings and campaigns itself, and the prompts become direct questions.

5 rules for a good Amazon prompt

Before the examples, the rules. A prompt that works almost always includes:

Marketplace: amazon.com, amazon.co.uk... Leave it out and the model guesses or mixes them.
Time window: "last 7 days vs the previous 7" beats "recently" every time.
Threshold: "at least 20 clicks and no orders" turns an opinion into a filter.
Output format: "a table with these columns" saves you the follow-up.
Intent: say whether you want analysis only or a prepared change.

The last rule matters most. We split prompts into two families: analysis (read-only) and operational (they prepare the change, show a before → after preview and wait for your approval).

8 ChatGPT prompts for Amazon sellers, area by area

1. Seller account overview

Summarize my week on amazon.com: revenue and units for the last 7 days vs the previous 7, the 3 SKUs that grew most and the 3 that declined most. Close with the strengths and the issues of the week.

This is your Monday opener. The closing request ("strengths and issues") forces the model to take a position instead of just listing numbers.

2. Catalog and listings: find the problems

List my active listings on amazon.com that have issues flagged by Amazon or are suppressed. For each, show SKU, ASIN, issue type and whether it sold in the last 30 days. Sort by revenue.

Sorting by revenue is the trick: fix the problems that cost you money first.

3. Catalog and listings: prepare a fix

For ASIN B0XXXXXXXX (use your own), read the current title and bullet points and suggest an improved version, without inventing any features that aren't already in the listing. Show me before and after, and don't publish anything until I confirm.

This one is operational. "Without inventing features" is essential: a listing should only claim what's true about the product.

4. Pricing and Buy Box

Which ASINs lost the Buy Box in the last 7 days on amazon.com? For each, show my price, the Buy Box price and the percentage gap. Don't change any prices, I just want the picture.

Before touching a price, you need to know how big the gap is and whether your margin can take it.

5. Inventory and FBA

For every FBA SKU, calculate average daily sales over the last 30 days and days of cover based on available stock. List those with fewer than 21 days of cover, most urgent first.

The 21-day threshold is just an example. Set it to match your replenishment lead time.

6. Advertising

Find Sponsored Products keywords that spent money without generating sales in the last 14 days, with at least 15 clicks. Show spend, clicks and campaign. Then prepare a proposal (lower bid or negative keyword) and wait for my approval.

This prompt moves from analysis to action, but the final call stays with you.

7. GPSR and EPR

Check my listings on amazon.de and tell me which ones are missing the product safety information required under GPSR (manufacturer, EU responsible person, warnings). Group them by the type of information missing.

GPSR/EPR compliance is the maintenance nobody does until a notice arrives. If you sell into the EU, it belongs in your routine.

8. Orders and sales

Compare revenue by marketplace for the last 30 days against the previous 30. Where am I growing and where am I declining? For the marketplace declining most, tell me which SKUs explain the gap.

The value is in the second question: not just "where", but "because of which products".

From 8 examples to a method: 27 prompts in 7 areas

These examples are a starting point. Inside the Vendooly app there's a library of 27 ready-made prompts in 5 languages, organized into 7 areas: seller account overview (3), catalog and listings (6), pricing and Buy Box (4), inventory and FBA (3), advertising (4), GPSR/EPR (4), and orders and sales (3). 15 are analysis prompts and 12 are operational. You open one, run it in your AI, and start from a result instead of a blank page.

And every Monday, if your seller account is connected and you had sales in the last 14 days, the email recap suggests "what to ask your AI this week", with prompts picked from your own data.

Using these prompts safely

Whatever the prompt, with Vendooly no write runs on its own: prices, stock, listings, bids, budgets and campaign status always show a before → after preview and are only sent to Amazon after your explicit confirmation. The details are in Action approvals, and there are more starter examples in Your first analysis and on the Vendooly MCP examples page.

Want to run these prompts on your own data? The 14-day free trial includes the full Seller + Ads plan: sign up here. To browse guides by area, start with the wiki.