Analysis of online reviews using AI better than market research

Machine learning and AI could make classic market research increasingly obsolete. This is because it will make it possible to efficiently analyze user-generated content (UCG) such as online reviews, blogs and social media postings. One in the Journal Marketing Science According to a published study, it is actually faster and cheaper for brands to find out what customers want - and possibly better than, for example, through traditional surveys.

As more and more people shop online and also share opinions about and experiences with products, a UCG data mountain is emerging that is massive in many areas. "On Amazon alone, there are over 300,000 reviews of health and personal care products," says Artem Timoshenko, a marketing doctoral student at the Massachusetts Institute of Technology (MIT) and lead author of the study. If this freely available data could be used to efficiently identify customer preferences, this would be very attractive for consumer brands. However, many marketers still have doubts as to whether this can be done efficiently enough.

Timoshenko and marketing professor John R. Hauser have therefore looked at whether this can be done with the help of machine learning. The researchers are using a hybrid approach: a learning system discards garbage and duplicates in Amazon reviews to deliver only relevant data. An expert then analyzes this data to identify customer needs.

The team tested this approach for oral hygiene products and compared the results with those from traditional face-to-face surveys.

"Ultimately, we found that UCG does at least as well as traditional methods that rely on a representative group of consumers," Hauser says. That's because the machine-learning approach has actually meaningfully reduced the review data mountain to a selection that a human expert can work well with. However, technology is unlikely to replace humans entirely in the foreseeable future. "Face-to-face review remains an important final step in the process, as professional analysts are best able to assess the contextual nature of consumer needs," says the marketing professor. (pte) 

Graphic: Manfred Steger via Pixabay