5-Star Phonies: How Amazon Fake Reviews Impact CX

Product reviews drive today’s e-commerce world. Users trust other users to an astounding degree, with 85 percent of consumers solely relying on reviews to inform their purchases. It's in this context where fake reviews become a lucrative and destructive business.

Disingenuous reviews threaten the function of reviews entirely. However, business is still booming for fake product and business reviews, with some commentators estimating almost 40 percent of Amazon.com reviews are fake. For example, this website even sells Amazon reviews at $15 each.

This is why systems which understand how to ignore reviews flagged by artificial intelligence (AI) are gaining prominence. Let’s explore how technology works to combat the five-story phonies of e-commerce.

The Why of Fake Reviews

Online business is booming. U.S. e-commerce sales grew to almost 15 percent of total retail sales, reaching $517 billion in 2018. This marks an upward trend of online purchases, and the fact of the matter is that consumers regularly use reviews to guide their purchase decisions.

For example, four out of five American adults check product reviews before making a purchase. Furthermore, research shows that consumers are more swayed by a simple star rating than what reviewers actually write. While four out of five buyers use reviews to determine whether products are worthy, the way e-commerce vendors evaluate reviews is alarmingly simplistic.

This is a widespread problem when one considers that one of the biggest players, Amazon, hosts 1.8 million vendors and sellers with nearly 600 million items that generate about 9.6 million new product reviews every month.

Finding the Fakes

So, how does one sort the fake from the genuine? Well, it's something that self-learning AI is able to do very well in bulk. Self-learning systems grow smarter with more fake reviews. Unlike human-trained AI, which relies on pre-defined keywords which can be fooled by the fake reviewers of the world, self-learning AI compares the reviews of each product to the industry’s standards and competitors. If it detects anomalies, it ignores the suspicious reviews when calculating sentiment analysis.

This is why machines which understand double sentiment based on specific product and industry benchmarks are more accurate than human reviewers. The “human” factor is responsible for about 90 percent of sentiment analysis errors. Eliminating this drastically improves false positive and false negative errors.

It's extremely hard to apply machine learning trained by humans to understand double sentiment as things like sarcasm are complex to code. However, self-learning systems sift through all industry products and categories to identify general sentiment and tone — and better understand cases of double sentiment and even sarcasm in reviews.

A Market Shift

Fake reviews sell more products, but confuse the big data customer analytics of major companies. This is because most products are sold by third parties and do not enable companies direct interaction with the end user. Therefore, consumer-centric enterprises face a challenge in understanding what customers feel towards their brand — and fake reviews only cloud this further.

In this way, self-learning systems work to better understand industry or product sentiment. This tech can easily filter millions of reviews and analyze the entire market on an ongoing basis. This removes fake reviews and presents an unbiased depiction of market sentiment to any given brand.

At the end of the day, it's all about the numbers. Self-learning AI can be applied to huge volumes of texts and reviews without any additional human training. This technology, finally unhampered by the fraudulence of review fakery, could present major customer insights going forward. Watch this space.

Niv Elad is the vice president of engineering of Revuze, the tech startup changing how organizations consume customer experience insights.