This month, researchers from Boston University School of Medicine announced the findings of their product review research. Their conclusion? Product reviews can predict foods that are unsafe for human consumption.
Food sales on the internet have increased in recent years, and with that has come an increase in reviews. When purchasing food online, consumers are unable to use visual or olfactory cues to judge the quality of that product. This research study matters immensely, with over 200 food-borne illnesses that can cause death. Sick, elderly, pregnant, and young people are at the highest risk.
The program used deep learning AI to look at food product reviews on Amazon between 2012 and 2014. Of the 1.29 million reviews they analyzed, they recognized 20,000 reviews suggesting unsafe products that had not previously been recalled.
It was able to detect phraseology like “foul”, “ill” or “rotten”, combined with the star rating to correctly identify issues with 74% accuracy.
What Else Can Product Reviews Predict?
As mentioned, food purchasing online is somewhat in its infancy, but rapidly growing. This growth has been emphasized by Amazon’s purchase of Whole Foods in 2017. But plenty of other product categories are old vets in the ecommerce space. And the leaders understand the predicting power that product reviews have for their business. Some examples of the foreshadowing that product reviews can provide:
- Fraudulent Product: Like it or not, fraud is a part of marketplaces like Alibaba and Amazon. This can be predicted with product reviews by analyzing the text for terms like “fake”, “not real”, “no box” or “generic”.
- Product Recalls: Beyond food related recalls, product reviews can also indicate problems your legal department should stay abreast of. This is especially important in products that affect the safety of loved ones, like children, pets and the elderly.
- “Up and Comers”: Monitoring product reviews of new products and competitors’ products can help predict which products will experience an upswing in demand. Measuring the volume of reviews, star rating averages as well as the review text can help you gain a full understanding of what consumers are loving about these “up and comers”.
- Manufacturing Problems: When product problems are uncovered in your reviews, you will be able to pinpoint specific design elements and even trace them back to the factory in question. This type of evidence early on can help you eradicate the problem quickly.
- Industry Changes: By maintaining a pulse check on your reviews, you are able to see how the industry is changing. For example, are consumers starting to care more about the sustainability of your product’s production? Are they increasingly moving to competing products? The possibilities are endless.
If you’re a leader in your space and you like to learn from your consumers, what other ways have you found that product reviews can predict issues? Share with us!