Ratings and Reviews at the most powerful moment - the point of sale

AI and ML will power the next generation of recommendation innovations

 

Bobby Figueroa, CEO Gradient

Online ratings and reviews have helped fuel the rapid growth of eCommerce by recreating an important experience from the physical shopping world – getting advice from people we trust. In the digital age, the pool of people that can give recommendations is exponentially larger, and those recommendations become more powerful the closer they are to the purchase decision – the point of sale.

Getting the right recommendation can make a big difference. A 2018 study by TurnTo, a provider of online shopping technology, as reported by eMarketer, found that most consumers will actually pay more money for a product with higher ratings. Customers, the survey found, are most interested in reviews covering product performance, buyer satisfaction and product quality -- both short term and over time.

Knowing all this, major eCommerce sites like Amazon and eBay have invested in robust review platforms, as have Google and social networks like Yelp and TripAdvisor. But shoppers know review systems to be manipulated at times, and many consumers look at online reviews with skepticism. That’s a missed opportunity.

For online merchants, that opportunity goes beyond the immediate sale. Review content can also be analyzed to deliver a better overall shopping experience to all their customers.  

Bogus reviews by bots and humans

Bad actors weaken the integrity of the marketplace. They come in two main categories: bots that generate synthetic reviews, and human reviews sponsored by companies hoping to boost their own products or disparage the competition. This is not an expression of what customers believe – it’s bad advertising practice.

Some companies have taken extreme measures to punish review fraud, not only taking down the reviews, but also removing the products themselves and even ending relationships with some sellers. That can be controversial – product owners often claim they have nothing to do with the bogus reviews.

A more sustainable approach to curation begins by weeding out people who didn’t actually purchase the product. For example, an eCommerce site may follow up on confirmed orders with an email asking the buyer to rate the product. For a longer-term view, we could ask purchasers for input six months after the transaction, this ensures the feedback includes a broader opinion of the product.

Not all reviews are created equal

To add more value, we can ask repeat buyers to share their feedback, creating a tier of top reviewers, whose insights are even more trustworthy. Review systems also gain credibility through “ratings of the ratings” – asking shoppers to tag reviews that were most useful. We can also use artificial intelligence to review and evaluate those ratings, while mining for customer insights that can improve products.

As for bots, many can be filtered out using algorithms to spot repeated or generic review language. For example, restaurant reviews that hail “fresh” or “delicious” food should rank lower, while comments on the juiciness of a steak or the use of real wasabi in a sushi restaurant are more valid and should rank higher. “I like this product,” is no competition for “I fell asleep in the business class seat and woke up in London feeling refreshed.”

The goal is not to eliminate negative comments or positive comments. It's understanding, through technologies like machine learning, what language patterns determine a real voice versus a manufactured voice. Those findings help a company share relevant perspectives on products with the audience that most values them. It’s another way of getting the right product on the right shelf at the right time.

Identifying the words that trigger certain consumer actions helps design future products that will appeal to their target audience. This turns the relationship between technology and customers on its head – machines aren’t used to change what consumers' feedback looks like, but to learn from it. We can then surface the most effective and relevant reviews to similar consumers.

Spam, robocalls, and review fraud

Bogus online reviews are the evolution of spam email and unwanted telemarketing phone calls. But just as we used technology to authenticate email and screen phone calls, we have tools today to detect and minimize review fraud and improve the integrity of the review system

Our eCommerce platforms and ratings services will need to continue investing in improvements that capture the best review content and apply it more programmatically to drive better online shopping experiences.  

Those companies are doing solid work today. They understand the importance of verifiable, genuine reviews and the benefits of getting the system right. Like spam and robocalls, we may never get to zero fraudulent reviews. But with the available technology, consumer participation and a motivated eCommerce industry, we are poised to drive steady improvements that provide the most value for everyone involved.