Best AI prompts for Amazon review analysis

These are copy-paste AI prompts for Amazon review analysis, drawn from 18 of the 20 sourced cases on this site. First are all 7 prompts the original sources published, quoted word for word. Then come 11 starter prompts we wrote, clearly labeled, so you can try the same jobs: complaint patterns, competitor gaps, listing fixes, and product research. They analyze real reviews; none of them write or buy reviews.

Replace anything in [brackets]. Don't let a prompt change a listing or message a buyer until you've checked the output.

Prompts published by the original sources

These 7 prompts were published by the people or companies in the cases. They are labeled From the source. The wording is unchanged, including any uneven phrasing. Rights stay with the original authors; see the license.

Score new FBA product ideas on demand, competition, margin and risk

From the source. Copy this prompt as published; replace anything in [brackets].

Copy-paste promptFrom the source
Find 10 product ideas in [category] under $[price].
For each: price, review count, estimated monthly sales range (label it an estimate and say what it's based on), margin after FBA fees at that price, and the top complaint in competitors' 1-3 star reviews.
Score each on demand, competition, margin and risk, and flag possible patent, trademark or gating issues for me to check.

Credit: Grokbot Agents (community template), Workflow template, Oct 16, 2026. Original source · Case on this site.

Tools named in the case: Grok Bot.

Result (per source, template): Template only, with no reported results. The page says its sales numbers are labeled estimates from public signals, and that it flags possible patent, trademark and gating risks for you to verify.

A published workflow template, not a reported result.

Worth knowing: This is a published workflow template, not a reported result: nobody in the source says what it did for their sales. Read the template's setup steps and test it on one account before you rely on it. This is a third-party template, not an official xAI one.

Classify competitor reviews into patterns and turn them into tested ad angles

From the source. Copy this prompt as published; replace anything in [brackets].

Copy-paste promptFrom the source
You are a senior ecommerce customer-research and creative strategist.

I am giving you public Amazon reviews from competing products.

Analyze only the supplied reviews. Do not invent facts, product capabilities, customer quotes, frequencies, or conclusions.

First, classify each review as:
1. Product defect
2. Expectation gap
3. Setup or usability friction
4. Missing feature
5. Wrong customer or use case
6. Shipping, seller, or delivery noise
7. Isolated complaint with insufficient evidence

Then create a table showing:
• Complaint pattern
• Number of reviews mentioning it
• Situation in which it happens
• Consequence for the customer
• Outcome the customer wanted
• Objection our advertising must answer
• Proof required before making the claim
• Review IDs supporting the pattern

Do not treat one-star reviewers as representative of the entire market. Identify repeated patterns separately from isolated complaints.

Next, compare the negative reviews with any positive reviews I provide. Find where customers disagree and explain what product type, expectation, or use case may cause the difference.

Finally, create 12 original ad experiments for my product.

For every experiment, provide:
• Hook
• Body copy
• Visual concept
• Customer problem
• Promise
• Proof needed
• Awareness stage
• Hypothesis being tested
• Result we should measure

Do not copy review wording, use a reviewer as a testimonial, attack a named competitor, or claim our product solves a problem unless my product information supports it.

Credit: wetracked.io (@wetracked), X post, Aug 28, 2026. Original source · Case on this site.

Tools named in the case: Claude.

Result (per source, self-reported): The post shares the prompt only and reports no results.

Self-reported by the source, not independently verified.

Worth knowing: The prompt explicitly forbids invented facts, frequencies or quotes and requires review IDs for every pattern.

Group review text into complaint and praise themes and compare with competitors

From the source. Copy this prompt as published; replace anything in [brackets].

Copy-paste promptFrom the source
What features do customers wish this product had?
Compare review sentiment for our product against the top 3 competitors — what do they praise that we get criticized for?
Has a specific complaint theme increased in frequency since our last listing update, suggesting the change backfired?
Which negative review themes correlate most strongly with declining conversion rate, versus ones that are just noise?

Credit: Jungle Scout, Vendor docs, Aug 27, 2026. Original source · Case on this site.

Tools named in the case: Jungle Scout.

Result (per source, vendor claim): This is a vendor feature page. It gives example questions but no customer results.

From the tool's vendor, not independently tested.

Worth knowing: Cobalt is aimed at established brands and agencies.

Rewrite bullets around the complaints in competitors' 1 and 2-star reviews

From the source. Copy this prompt as published; replace anything in [brackets].

Copy-paste promptFrom the source
You are an Amazon listing specialist. Write 5 bullet points for an Amazon product listing.

Product: [description]

Top keywords to integrate: [list 8-10 keywords]

Key product strengths: [list 4-5 genuine strengths]

Top competitor complaints to address (from 1-star reviews):

1. [complaint 1]
2. [complaint 2]
3. [complaint 3]
4. [complaint 4]
5. [complaint 5]

Requirements:

- Start each bullet with a capitalized benefit phrase in bold
- First bullet must contain the primary keyword: [keyword]
- Max 250 characters per bullet
- Address at least 3 of the competitor complaints as implied advantages
- No all-caps except for the bold opener
- No exclamation points
- Write in second person ("you") not third person ("this product")

Credit: Mike Begg (AMZ Commerce Advisers), Blog post, Jun 1, 2026. Original source · Case on this site.

Tools named in the case: ChatGPT.

Result (per source, agency-reported): He says a supplement brand in their portfolio rewrote its bullets this way and its conversion rate moved from 11% to 17% in 30 days, with the same traffic and price. His team uses the workflow to optimize 20-30 listings a week.

Reported by an agency about its own client work, not independently verified.

Worth knowing: An agency's own client result, not independently verified. He says never to publish AI copy without a human compliance pass for character limits, prohibited claims and restricted keywords.

Find what buyers love in reviews and rebuild listing images around it

From the source. Copy this prompt as published; replace anything in [brackets].

Copy-paste promptFrom the source
please analyze the attached reviews and write the top five reasons why people love this product based on the order of frequency these reasons appear within the reviews

Credit: George Schwartz (Extension Ecom) (@George_AMZN), YouTube video, May 9, 2025. Original source · Case on this site.

Tools named in the case: Claude, Data Dive.

Result (per source, self-reported): The agency reports a client's conversion rate rose 19.6%, from 25.86% to 30.93%, after the creative update.

Self-reported by the source, not independently verified.

Worth knowing: Claude did the review analysis. The lift came from the new images built from it.

Turn 200 to 500 competitor reviews into pain points, product ideas and listing copy

From the source. Copy this prompt as published; replace anything in [brackets].

Copy-paste promptFrom the source
I have just uploaded [number] Amazon reviews for [product]. Collect and analyze all product reviews together and answer the following questions:
1. After analyzing the data, what notable insights can you identify?
2. What are the five main pain points customers experience with these products, ranked in order of urgency?
3. What are the seven most valuable benefits customers derive from these products, ranked in order of relevance?
4. For which occasions or events are these products mainly purchased, based on priority?

Credit: Karst Branderhorst (Follo Agency), Blog post, Aug 28, 2023. Original source · Case on this site.

Tools named in the case: ChatGPT.

Result (per source, self-reported): The agency says the review analysis gives it 'unique insights' it uses for bullets, product improvements and new product ideas, keeps unused bullets for A/B tests, and runs a plagiarism check on the final copy: 'To date, we have not had any issues.'

Self-reported by the source, not independently verified.

Worth knowing: A 2023 agency method with no conversion figures. The rule to collect each competitor's reviews separately and only ask questions once all of them are in keeps the analysis on the whole market, not the last one pasted.

Rewrite a listing from real data, find what negative reviews share, and watch competitors

From the source. Copy this prompt as published; replace anything in [brackets].

Copy-paste promptFrom the source
You're an Amazon listing optimization expert. Here's my current listing, competitor listings, my reviews, competitor reviews, and top search terms. Rewrite my title, bullets, and description. Prioritize keywords that appear in search terms but are missing from my listing. Use language from positive reviews. Address concerns from negative reviews preemptively. Keep bullets under 200 characters each. Make the title under 200 characters.

Credit: Dmytro Negodiuk (Mozabrik), Blog post, Undated page; it refers to Amazon policy "as of March 2026", so written in or after March 2026.. Original source · Case on this site.

Tools named in the case: Claude, Playwright.

Result (per source, self-reported): Claude spotted competitors ranking for 'personalized photo gift' while his listing only said 'photo mosaic'; adding that to the title raised impressions by 23% in 14 days and click-through from 3.1% to 3.8%. The review analysis found 34% of negative reviews described the same assembly problem; after a new instruction card and a QR code to a video, the return rate fell from 11.2% to 3.1% in 60 days.

Self-reported by the source, not independently verified.

Worth knowing: One seller's own figures on a page that also sells his consulting, and the FAQ at the end calls the review issue 'packaging' rather than assembly. He also reports where AI fell short: PPC tools that needed constant oversight and AI lifestyle photos that converted 7% worse than real ones in his test.

Starter prompts written by AI Agents for Sellers

The source in these cases did not publish its exact prompt. We wrote these starter prompts so you can try the same workflow. Each one is labeled Starter prompt written by AI Agents for Sellers, based on [the case]. Replace anything in [brackets]. These starter prompts are part of our own text and are covered by the CC BY 4.0 license on this page.

Review mining and complaint patterns

Group review text into themes, find root causes of low ratings, and compare your complaints with competitors' praise.

Pull hundreds of reviews for an ASIN and turn them into product improvements for a refresh or new launch

Starter prompt written by AI Agents for Sellers, based on Competitor Review Mining and Product Improvement Report with Codex.

Copy-paste promptStarter prompt, written by us
Open this Amazon listing [URL] in the browser. Go through the written reviews by star rating (1 to 5 stars), clicking 'see more' as needed, and collect review text, rating and date into a table. Report how many you collected out of how many are displayed. Then group the complaints and praise into themes with counts and 2 quoted reviews each, and list product improvement opportunities ranked by how often they come up. Don't log in to any account. Save this process as a skill I can rerun on another ASIN.

The case: John Aspinall (Aspi), YouTube video, Sep 28, 2026. Original source · Case on this site.

Tools named in the case: Codex.

Result (per source, self-reported): On the Echo Dot it recovered "958 unique reviews, 46% of the 2,000 displayed", and the updated report added six product opportunities, such as software onboarding and connectivity recovery, each with review evidence.

Self-reported by the source, not independently verified.

Worth knowing: He says Amazon doesn't like scraping, advises against logging into Seller Central in that browser or running agent workflows there, and frames this as research only. Variation listings slow it down, and he notes paid review APIs may suit sellers checking many ASINs.

Find why buyers leave 1-2 star reviews and get an improvement idea for each

Starter prompt written by AI Agents for Sellers, based on Negative Review Root Causes with Gemini in an n8n Template.

Copy-paste promptStarter prompt, written by us
Here are 1 and 2-star reviews of [product]: [paste].
For each review return JSON: sentiment_score (1-5), category (Quality, Design, Shipping, Instructions, Other), summary (one line), improvement (one concrete change to the product, packaging or listing).
Then group by category and tell me the 3 fixes that would address the most reviews.

The case: Koichi Kobayashi (n8n template), Workflow template, Nov 20, 2025. Original source · Case on this site.

Tools named in the case: n8n, Gemini.

Result (per source, template): Template only, with no reported results. Its AI prompt is set to answer in Japanese; the setup notes say to translate the prompt for English output.

A published workflow template, not a reported result.

Worth knowing: This is a published workflow template, not a reported result: nobody in the source says what it did for their sales. Read the template's setup steps and test it on one account before you rely on it. It needs a paid Apify actor and a Google Cloud account for the Gemini API.

Compare your review complaints with competitors' praise

Starter prompt written by AI Agents for Sellers, based on Review Gap Analysis vs Competitors with an n8n Template.

Copy-paste promptStarter prompt, written by us
Here are reviews for my product and for [3] competitors: [paste or attach].
1. For each product, list complaint themes, praise themes and feature requests, with a count and one example quote each.
2. Find gaps: things buyers complain about in my product that they praise in a competitor's.
3. Write a 5-line brief: the top 3 gaps and what to change in the product or the listing for each.
Only use what's in the reviews.

The case: Daniel Shashko (n8n template), Workflow template, Aug 13, 2026. Original source · Case on this site.

Tools named in the case: n8n, OpenAI API, Bright Data.

Result (per source, template): Template only, with no reported results.

A published workflow template, not a reported result.

Worth knowing: This is a published workflow template, not a reported result: nobody in the source says what it did for their sales. Read the template's setup steps and test it on one account before you rely on it. It needs a Bright Data API key (paid) and an OpenAI key.

Turn review complaints into a ranked list of delivery, return, sizing and defect problems

Starter prompt written by AI Agents for Sellers, based on Review Friction and Return Risk Report with an n8n Template.

Copy-paste promptStarter prompt, written by us
Here are the last 200 reviews of my product: [paste].
Tag every complaint as delivery, packaging, return, sizing, defect, expectation mismatch or other. Count each tag, give 2 example quotes, and say which ones a listing change could fix (and how) versus which need a product or packaging change.

The case: Yaron Been (n8n template), Workflow template, Feb 22, 2026. Original source · Case on this site.

Tools named in the case: n8n, Bright Data.

Result (per source, template): Template only, with no reported results.

A published workflow template, not a reported result.

Worth knowing: This is a published workflow template, not a reported result: nobody in the source says what it did for their sales. Read the template's setup steps and test it on one account before you rely on it. The 'revenue impact' is the model's estimate, not your sales data.

Use one AI coworker across reviews, listings and search term reports for managed Amazon brands

Starter prompt written by AI Agents for Sellers, based on Review Clustering, Listing Rewrites and PPC Signals with Claude.

Copy-paste promptStarter prompt, written by us
Here are the last [300] reviews for [ASIN] and my current listing: [paste]. Cluster the reviews into themes with counts, list the top complaints and buying triggers with a quote each, and map each one to a gap in my title, bullets or images. Then suggest bullet rewrites that close the top three gaps. Don't invent claims.

The case: Ananth Kuchimanchi (Brandkyte), Blog post, Apr 10, 2026. Original source · Case on this site.

Tools named in the case: Claude.

Result (per source, agency-reported): He reports they reduced time spent on review analysis by ~70% and improved listing conversion by 15-25% on updated ASINs, that PPC optimization went from weekly to near daily with average ACoS improvements of 12-18% within 30-45 days, and 60-70% less manual analysis time across a set of managed brands.

Reported by an agency about its own client work, not independently verified.

Worth knowing: Agency figures across unnamed client brands, not broken out per account. He adds that the biggest gains came from cleaner data and workflows rather than better prompts.

Turn buyer messages, refunds and reviews into replies and product fixes

Starter prompt written by AI Agents for Sellers, based on Customer Signals Agent with n8n.

Copy-paste promptStarter prompt, written by us
Here are my last 90 days of buyer messages, refund reasons and 1-3 star reviews: [paste]. Group them by product and by issue, show which issues are growing week over week, and flag any that point to a batch or packaging problem. For each issue, suggest one product or packaging fix and one listing change. Then draft a reply template for each common buyer message type that follows my policy: [paste policy].

The case: Michal Špecián (Orange Klik), YouTube video, Sep 25, 2025. Original source · Case on this site.

Tools named in the case: n8n.

Result (per source, agency-reported): A person still copies each drafted reply and sends it by hand: 'It could be automated fully, but we want to make sure that there is this actual person who who revises that.' He says review removal is 'not 100% effective.' No time or sales figures are given.

Reported by an agency about its own client work, not independently verified.

Worth knowing: An agency showing a client build to promote its services. He says Claude helped draft the n8n workflows, but 'It's not plug-and-play,' and advises weighing build and maintenance time against the time a workflow saves.

Listing fixes from reviews

Turn what buyers ask and complain about into listing changes you can test.

Find why a product gets returned and pick one listing change to test

Starter prompt written by AI Agents for Sellers, based on Return Root Causes and One Listing Fix per ASIN with a Claude Skill.

Copy-paste promptStarter prompt, written by us
For my 5 ASINs with the highest return cost, read the FBA customer returns report, the product's reviews and the current listing side by side. For each ASIN give:
- the likely root cause, with return counts, reason-code shares, 2-3 buyer quotes, the gap in the listing, and a confidence level (never 'high' from reason codes alone);
- ONE change to test (title under 75 characters, counted; image brief; size chart; or supplier brief);
- today's return rate as the baseline to check at 30 and 60 days.
Don't change anything and don't contact buyers.

The case: Ori Firouz, Blog post, Oct 20, 2026. Original source · Case on this site.

Tools named in the case: Claude, Adi (Claude skill).

Result (per source, self-reported): Demo only, on a fictional store. In the example, a bamboo cutting board set was returned 12.0 times per 100 sold, 56% coded 'not as described.' 43 return comments said it was smaller than expected, 8 of 10 critical reviews agreed, and the 109-character title never gave the size. The proposed test was a 73-character title with '15 x 10 in' plus measurements on image 2. 'No result is promised.'

Self-reported by the source, not independently verified.

Worth knowing: It only reads data. It never changes a listing or contacts buyers. For returns coded as a different item, it drafts Seller Support case text instead of accusing the buyer.

Rewrite bullets so Amazon's shopping AI can answer buyer questions

Starter prompt written by AI Agents for Sellers, based on Bullets Written for Alexa for Shopping with NotebookLM and Claude.

Copy-paste promptStarter prompt, written by us
Here are my competitors' 1-3 star reviews and return reasons: [paste].
1. List the questions buyers are really asking (e.g. 'Is it microwave safe?'), grouped and ranked by how often they come up.
2. For each of the top [8], write a bullet line that answers it with an exact spec from my product facts: [facts]. Format: "Question? Answer, with the number."
3. Use each keyword from my list only once: [keywords].
4. Suggest 5 Q&A pairs for a Premium A+ Q&A module.
Don't answer anything my facts don't cover; list those as gaps instead.

The case: Sumner Hobart (@sumnerhobart), X post, Oct 19, 2026. Original source · Case on this site.

Tools named in the case: Claude, Gemini.

Result (per source, self-reported): The rules it came back with: use each keyword only once, since Amazon only needs to see it once to index you; find buyer questions in competitors' bad reviews and return reasons; answer each in your bullets with exact specs ('Biodegradable? Yes, 100%.'); keep answers as plain text, not baked into images; and put top buyer questions in a Q&A module in Premium A+ Content.

Self-reported by the source, not independently verified.

Worth knowing: Advice, not a measured result. The thread promotes a free training. 'Amazon's AI can't quote what you never wrote.' See our AI shopping agents page for what Amazon has said about Alexa for Shopping.

Product research from reviews

Read complaints before you source a product, and look for buyers the current listings don't serve.

Read a best seller's 1-star reviews before committing to inventory

Starter prompt written by AI Agents for Sellers, based on One-Star Review Pattern Detection with Claude Code.

Copy-paste promptStarter prompt, written by us
Here are all the 1-star and 2-star reviews for [ASIN] with their dates and any 'used for X days' mentions: [paste]. Find failure patterns, especially WHEN the product fails (days or weeks of use) and HOW. Count how many reviews support each pattern and quote short supporting snippets with review dates. Tell me whether the timing lines up with Amazon's return window.

The case: Jasper Li (@Jasperli0122), X post, Aug 10, 2026. Original source · Case on this site.

Tools named in the case: Claude Code.

Result (per source, self-reported): He reports Claude found the pattern: 'The product dies around day 30. The return window closes at day 30.' The run cost $0.0477 with pay-as-you-go API usage.

Self-reported by the source, not independently verified.

Worth knowing: The headline claim that this replaced a $359/mo seller suite is the author's framing.

Pick a product by finding buyers whose complaints the current listings don't solve

Starter prompt written by AI Agents for Sellers, based on Review Mining to Find Underserved Buyers with Claude.

Copy-paste promptStarter prompt, written by us
Here are 80 reviews for [product] (pasted below).
1) Group the reviewers into buyer personas (who they are, what they use it for).
2) For each persona, list complaints and unmet needs, with how many reviews mention each.
3) Which persona is buying anyway but least satisfied, and what product change would fix their top two complaints?

The case: Matt Clark, YouTube video, Sep 20, 2026. Original source · Case on this site.

Tools named in the case: Claude.

Result (per source, self-reported): For each of the three finalists he pulls 70-100 reviews and has Claude find the 'silent sufferers' who keep buying but aren't fully satisfied, then sizes that persona's daily sales. After finding a manufacturer with a plain Google search, he has Claude draft the supplier outreach email.

Self-reported by the source, not independently verified.

Worth knowing: He sells an Amazon course, and the $25K/day in the title is the goal, not a result shown. The review step is solid: reviews from people who bought anyway tell you what a better version needs.

Size up a product idea before sourcing it

Starter prompt written by AI Agents for Sellers, based on Product Idea Research with a Custom GPT in ChatGPT.

Copy-paste promptStarter prompt, written by us
Research this product idea for Amazon US: [product]. I've attached [a Black Box / Product Opportunity Explorer export] - use it for sales and search figures and say which numbers come from it and which from the web.
Cover: market size and trend, seasonality, the main competitors and their price points, the top 5 buyer complaints in reviews (with quotes), and what a better version would need.
Don't name 'top sellers' unless they're in my file.

The case: Andrew Bell and Bradley Sutton (Serious Sellers Podcast #680), Podcast episode, Jul 5, 2025. Original source · Case on this site.

Tools named in the case: ChatGPT.

Result (per source, vendor-affiliated): Plain ChatGPT's answer was wrong: the links weren't even wooden egg trays, and one of Sutton's own top sellers in that niche wasn't on its list. The custom GPT's report surfaced complaints such as shallow holes that let eggs roll, splintering and finishes that aren't food safe.

From a vendor-affiliated source, not independently verified.

Worth knowing: A live demo, not a launch result. The show is Helium 10's own podcast, so expect its tools to come up as the data source. The takeaway that holds without any tool: ChatGPT can't see Amazon sales data, so give it a file of real numbers before you ask what sells.

How we picked these

As of Nov 18, 2026, this site has 374 public, sourced cases. 34 of them include reviews as a topic, and 20 of those analyze review text in their workflow or prompt: the cases whose main topic is reviews, plus side-topic cases we checked one by one. The other 14 only mention reviews in passing (store agents, sales and inventory tools, general research) and stay on Reviews cases. 7 of those cases include a prompt published by the original source; 11 include a starter prompt written by us and appear on this page, and 2 are not yet on this page.

Section 1 includes all 7 source prompts, unchanged. Section 2 includes all 11 starter prompts we wrote for the cases on this page, grouped by review job. This page covers 18 of the 20 review cases. We did not pick prompts to promote a tool. Vendor, vendor-affiliated, agency, Amazon, and template cases are labeled as such.

Every case links to its original public source. Every result and note on this page is quoted from the case. We do not independently verify results. Counts reflect what we found, not market share or how many sellers use each tool. See how we source cases. Related pages: Reviews cases, Which AI agent for Amazon review analysis?, and State of AI Agents for Amazon Sellers.

License and how to cite this page

The counts, groupings, and original text written by AI Agents for Sellers, including starter prompts, are CC BY 4.0: reuse them with credit and a link to this page. Quotes, prompts marked “source”, and linked material belong to their original authors.

How to cite this page: AI Agents for Sellers, “Best AI prompts for Amazon review analysis”, as of Nov 18, 2026, https://aiagentsforsellers.com/prompts/amazon-reviews/