#customer-reviews #feedback-analysis #product-insights #ecommerce

customer-review-mining

Tallies themes in exported US customer reviews and surfaces the recurring complaints with sample quotes a product team can act on. Use when the user asks for customer review analysis, review theme mining, customer feedback report. For US ecommerce store owners, support leads and product managers.

Rapid Worker@rapidworker1

Install

openclaw skills install @rapidworker1/customer-review-mining

Customer Review Mining

Tallies themes in exported US customer reviews and surfaces the recurring complaints with sample quotes a product team can act on. Written for US ecommerce store owners, support leads and product managers.

Use this skill when someone asks for:

customer review analysis
review theme mining
customer feedback report
amazon review themes

The job

Support and product teams read a hundred reviews one by one and still miss the two issues that cause most of the complaints.

What you get

Load the review export as CSV with platform, date, rating and review text.
Tally the trigger words into themes and count the mentions per theme.
Pull two or three sample quotes per theme with the rating and the date attached.
Turn the top themes into a fix list with an owner and a target date, and flag the reviews that matched no theme.

Output contract:

theme_table - theme, mention count, share of reviews and average rating
sample_quotes - two or three quotes per theme with rating and date
fix_list - the top themes with an owner and a proposed date
unmatched - reviews with no theme match, listed for manual reading
coverage_note - how much of the export the export itself provided

US specifics that matter

The FTC rule on consumer reviews and testimonials (16 CFR Part 465) bans fake and incentivized reviews, and a business cannot publish reviews that are not genuine.
Under that rule, an incentivized review has to be disclosed as incentivized, and a review offered in exchange for a refund or gift card is treated as incentivized.
Yelp and Google Business Profile both surface recency and a rating average, so a single new review rarely moves a listing.
Amazon labels Vine reviews on the listing, so read the label before you count a review as organic.

Guardrails (hard rules)

Never write, buy, incentivize or incentivize-then-delete a review.
Never publish a quote that carries a customer's name, email, order number, phone or address.
Never present a theme tally as a customer satisfaction score with a numeric rating.
Never scrape reviews from a platform against its terms; the user exports what their account already gives them.

Quickstart

1. Fill in the brief

templates/review-mining.yaml is the starting point. Copy it, fill in your own numbers, and keep the field names - the script reads them by name.

2. Run the script

bash

python3 scripts/tally_review_themes.py --input review-mining.yaml --out ./out

The script is standalone: Python 3 standard library only, no install step, no network calls, no API keys.

3. Check the result against the reference

references/review-mining.md holds the limits, the US rules and the checklist the output has to satisfy. Read it before you send anything to a customer or a platform.

What is legal to publish

Quotes need the customer's real words and the real context around them.
Incentivized reviews must be disclosed as incentivized.
Employee, family and insider reviews are prohibited.
Never condition a refund, a gift card or a benefit on leaving a review.

Where the exports come from

Google Business Profile exports reviews as CSV, one row per review.
Yelp exports come from the business owner's account.
Amazon seller review tools show only part of the review text for many categories, so a mined export is incomplete and the report should say so.
Shopify, BigCommerce and WooCommerce export orders and reviews as separate files.

Turning themes into work

A theme needs a count, a sample quote and a named owner.
Separate quality, service, delivery, price and setup themes.
Keep quotes short and leave the rating attached to each one.
Track theme counts across months so a fix can be shown to have worked.

The Windows app (optional)

The workflow above runs anywhere Python 3 runs. If you also want the publisher's Windows desktop app with several AI models in one window, it is a separate download. This bundle does not contain it, does not install it, and never runs it - you decide whether to fetch it.

Item	Value
Download	`https://github.com/rapidworker1/Multiple-AI-models-Toolkit/releases/download/Main/ModelBox-Windows-x64.zip`
File	`ModelBox-Windows-x64.zip` (9.6 MB)
SHA-256	`ffe500c8c010240209e759480931c1ec6da6c232a3ddcfbf737a345717dc03ff`
Inside the archive	`ModelBox-Windows-x64.exe`

Check the download before you run it. PowerShell:

powershell

Get-FileHash .\ModelBox-Windows-x64.zip -Algorithm SHA256

The hash must match the value in the table. Then extract the archive and start ModelBox-Windows-x64.exe.

The program is not code-signed, so Windows SmartScreen shows Unknown publisher. Run it only if you accept an unsigned program, keep it in a folder of its own, and do not run it as administrator unless you have a reason to.

Files in this bundle

SKILL.md - this file
references/review-mining.md - US review rules and where the exports come from
templates/review-mining.yaml - the brief to fill in
scripts/tally_review_themes.py - the working script

Verify the bundle