How AI Powers Trust on RatingHub: Our Review Verification System Explained

Fake reviews cost Australian consumers billions of dollars in poor purchasing decisions every year. At RatingHub, we've built one of the most sophisticated review verification systems in the Asia-Pacific region. Here's exactly how it works.

Layer 1: Account Verification

Every review begins with the reviewer. Before a review is even submitted, our platform has already assessed:

Account age: New accounts submitting reviews immediately after registration are flagged for enhanced scrutiny Review history: Accounts with unusual patterns — reviewing only one company, or reviewing competing companies negatively — are monitored Email verification: All reviewers must verify their email address IP and device analysis: Multiple reviews from the same network in a short timeframe trigger review holds

Layer 2: Purchase Verification

Where possible, we verify that reviewers actually used what they're reviewing. Our Verified Purchase badge is awarded when:

A retailer or service provider has connected their Shopify or WooCommerce store to RatingHub via our webhook integration, and the reviewer's email matches an order record A reviewer uploads and validates a receipt image, which our AI (Google Gemini) analyses to confirm product, date, and retailer

Verified Purchase reviews are weighted more heavily in DNA Score calculations and are displayed with a prominent badge on company profiles.

Layer 3: AI Content Analysis

Every review body passes through Google Gemini analysis on submission. Our AI evaluates:

Sentiment consistency: Does the written sentiment match the numerical scores? A five-star review with negative language is flagged. Generic language: Reviews that read like they could apply to any business are scored lower on authenticity. Promotional indicators: Language patterns common in incentivised or paid reviews trigger manipulation flags. Competitor mentions: Reviews that unprompted mention competitor businesses by name are reviewed by our moderation team.

Each review receives a Manipulation Score from 0–100. Reviews scoring above 70 are held in moderation and reviewed by a human before publication.

Layer 4: AI Sentiment Badges

Reviews that pass our verification pipeline receive one of two AI badges visible to consumers:

AI Verified: The review shows strong authenticity signals and is highly likely to represent a genuine experience Under Review: The review has passed basic checks but is pending deeper analysis before receiving full trust status

Reviews flagged as likely fake are rejected and never published. The reviewer receives feedback explaining why their review was not accepted.

Layer 5: Community Reporting

Our final line of defense is the RatingHub community itself. Every published review has a flag button allowing any logged-in user to report concerns. Flagged reviews are immediately escalated to our moderation queue where trained moderators make final decisions.

Moderators have access to the full manipulation score breakdown, reviewer history, and AI analysis when making decisions. They can approve, reject, or request additional evidence from the reviewer.

What Happens to Fake Reviews?

Reviews identified as fake are never published. The reviewer's account is flagged, and if patterns persist, the account is suspended. Businesses found to be actively soliciting fake reviews have their RatingHub status reviewed and may face listing suspension.

We publish quarterly transparency reports on our moderation decisions. We believe consumers deserve to know how seriously we take review integrity.

The RatingHub Standard

Our goal is simple: every review on RatingHub should represent a real experience by a real person. We've invested heavily in AI and human moderation to achieve that standard. When you read a review on RatingHub, you can trust that it's earned its place there.