I look for fake reviews by measuring the relationship between 1, 2, 3, 4 and 5 star reviews. There is a normal distribution and when I find outliers, it usually

I look for fake reviews by measuring the relationship between 1, 2, 3, 4 and 5 star reviews. There is a normal distribution and when I find outliers, it usually indicates that something suspicious is happening. If I understand what you mean by inventory co-mingling, there is actually a drill-down that's built into Amazon itself. For example, the reviews of Samsung's HW-T450 look awesome at first blush: https://smile.amazon.com/dp/B085WTFCQ7/ https://smile.amazon.com/dp/B085WTFCQ7/ - but, if you filter the ratings to only display that particular model, you'll see that 22% of the reviews are 1-star: https://smile.amazon.com/product-reviews/B085WTFCQ7/ref=cm_cr_arp_d_viewopt_sr?ie=UTF8&filterByStar=all_stars&reviewerType=avp_only_reviews&formatType=current_format&pageNumber=1#reviews-filter-bar https://smile.amazon.com/product-reviews/B085WTFCQ7/ref=cm_c... This an automated check that I perform, but I also investigate the reasons for these disparities. (In this case, it's because the soundbar lacks a digital decoder to support the surround sound speakers that are sold as an additional accessory). On rarer occasions, the negative reviews can be discounted, even if they are legit. One example of this is when lots of customers misread the dimensions of a product (e.g. I assumed it would be larger...).