Amazon Fake Review Flagger

WHAT IT DOES

This script analyzes customer reviews on Amazon product pages and flags those that exhibit patterns commonly associated with fake or incentivized reviews. It examines each review for multiple indicators: overly generic language (e.g., "great product," "works great," "highly recommend"), extremely short reviews with 5-star ratings, repetitive phrasing, unnatural capitalization or excessive exclamation marks, generic reviewer names (e.g., "Amazon Customer"), and mismatches between review titles and bodies. Each review receives a suspicion score based on these indicators, and is then classified into one of three levels: "Genuine" (green badge), "Questionable" (yellow badge), or "Likely Fake" (red badge). A summary banner at the top of the reviews section shows the total number of reviews analyzed and the breakdown by classification. Clicking on any badge opens a popup showing the specific reasons that review was flagged, giving you full transparency into the analysis. The script helps you make more informed purchasing decisions by surfacing reviews that may be less trustworthy.

WHERE IT RUNS

The script runs on all major Amazon domains, including .com, .co.uk, .ca, .de, .fr, .it, .es, .com.au, and .co.jp. It works on product detail pages that have a reviews section, including the main product page, the "See all reviews" page, and pages that load reviews via infinite scroll. It handles both desktop and mobile views, and dynamically processes new reviews as they load.

HOW TO USE

After installation, the settings panel appears on the left side of the screen with the gear emoji (⚙️). Open the panel to configure the script:

- The "Enable fake review detection" toggle turns the flagger on or off. When enabled, reviews are analyzed and flagged automatically.

- Select a sensitivity level from the dropdown: "Low (fewer flags)" for less aggressive detection, "Medium (balanced)" for default operation, or "High (more flags)" for more aggressive detection that may flag more reviews as questionable.

- Click "Scan reviews now" to manually re-scan all visible reviews.

- Click "Reset all flags" to clear all badges and summaries.

The script works automatically—just browse Amazon products and the flags will appear next to review titles. Click any badge to see the detailed analysis for that review.

TECHNICAL SECTION — REAL MECHANISM

The script first detects whether the current page is a product page and extracts the ASIN for caching. It then finds all review elements using selectors such as .review, .a-section.review, and [data-hook="review"], filtering for elements that contain review body text. For each review, the script extracts the review text, title, rating (from .a-icon-alt or [data-hook="review-star-rating"]), reviewer name, and helpfulness count (from [data-hook="helpful-count"]). The analysis engine then applies a series of heuristics: it counts occurrences of common generic phrases (e.g., "great product," "highly recommend") from a predefined list; checks if the review is very short (under 50 characters) but has a high rating (4.5+); detects excessive repetition of words; looks for unnatural capitalization or excessive exclamation marks; identifies generic reviewer names; and compares the review title to the body to detect mismatches. Each indicator adds to a suspicion score. The sensitivity setting adjusts the threshold at which a review is classified as questionable or fake. Reviews are cached by ASIN to avoid re-processing on page reloads. Visual badges are inserted into the DOM using textContent and style properties (never innerHTML), and a MutationObserver watches for new reviews being added via infinite scroll.

THE PANEL

The settings panel is built with BotGentz's BGPanel component. It is draggable via its header, allowing you to reposition it anywhere on the screen. The panel snaps to the left edge by default but can be moved freely. Click the gear emoji to collapse the panel into a small bubble, and click the bubble to expand it again. The panel remembers its position per page session. Press the Escape key to quickly close or minimize the panel. The panel includes a status flash bar that provides visual feedback when settings are changed or actions are performed. All settings save automatically via GM_setValue.

PLEASE NOTE

This script requires the free BotGentz browser extension to run. It does not function as a standalone Chrome extension and does not use chrome.* APIs. The script uses heuristic analysis and may produce false positives or false negatives—no automated system can perfectly detect fake reviews with 100% accuracy. Use the flags as a helpful guide, not as definitive proof. Amazon frequently updates its page structure, which may break the selectors used to find reviews and extract data. If the script stops working, please check for updates. The script stores only the enabled and sensitivity preferences via GM_setValue. No data is sent externally—all analysis is performed locally in your browser. The script never reads or writes password fields, hidden inputs, or file inputs. All DOM operations use safe methods (style properties, textContent, appendChild) and never use innerHTML.