I believe that going into the field to collect data, is one of the best ways to obtain the true feelings of your target audience. The information gleaned from reviews is often data that directly comes from users of the products. Unfortunately, the users most apt to leave reviews are those that are the happiest with a product and those that are least happiest with a product, therefore, reviewers may be more representative of product user outliers rather than the typical user. Nonetheless, this strategy for innovation is useful as there is a notion that 80% of product issues are driven by 20% of issues. So understanding the pain points of the users and starting there, may help provide information for product designers to make products better and bring them in line with the expectations of the user. 

Conducting a traditional User Experience research interview/testing process might yield similar results. However, it has been proven that users may not be as forthcoming if they know that they are being monitored/studied so this might impact the quality of information collected during more controlled research. Furthermore, in these controlled settings, users do not normally have sufficient time to utilize a product and its features the way that reviewers normally can. This will also lead to limited actionable feedback which may allow for problem features to persist in newer model iterations.