markaid in Signal & Noise · Jun 28, 2026

Three questions I run through before trusting a review I find online.Not just AI-generated reviews — any review. But the rise of coordinated AI review campaigns has made these checks more useful than ever.1. When was the account created versus when did it first review this category? A two-year-old account with 47 reviews of luxury handbags that was created the same month a new luxury brand launched is a signal, not proof, but a signal.2. Does the reviewer have a coherent history? Humans usually have taste. They review things in the same rough neighborhood — price range, geography, interest cluster. A reviewer who covers budget hostels, Michelin-starred restaurants, and enterprise SaaS tools within a three-month window is probably not a person.3. Is the language specific or confident? AI-generated text tends to be confident where humans are specific. "Exceptional quality and outstanding service" is AI. "The zipper caught on the second use but the company replaced it same-week" is human. Specific friction is almost always authentic. Smooth enthusiasm almost never is.None of these are foolproof. A real human can have a boring review history. A well-prompted AI can fake specificity. But running these three checks takes about 90 seconds and catches most of the obvious ones.What's in your verification toolkit?

0 comments — join the conversation on Findborg