Online reviews influence what we buy, but should they have that much power over our choices?

Katie Mehr For The Conversation

Alberta: Imagine you’re looking to buy a new grill. You want to make sure you purchase a well-built, easy-to-use grill for you and your family. How can you determine which one is best to purchase? On the one hand, you can rely on information the manufacturer provides to understand things like what material the grill is made from, how big it is and whether it has additional features like a grease management system.

But this information doesn’t really tell you what it’s like to own the grill, or whether the grill will work well for your summer barbecue aspirations.

For that, you probably want to hear from people who have bought and used the grill and can speak to its quality.

This example highlights the appeal of product ratings and reviews: by providing insight from people who actually bought and used the grill, aspiring grill owners learn more about what owning it will be like.

Predicting experience

People rely on reviews because they want to predict what their experience will be like with a product. They see reviews as a good source of information for making this prediction.

Reviews are also plentiful and almost costless to produce and access, bolstering the likelihood that people use them. And, people can sort through reviews to find information about specific attributes and benefits of the product (for example, whether a grill evenly cooks steak), which can help address specific queries or concerns.

Taken together, these benefits lead people to rely on reviews to determine whether they should buy a given product.

In fact, reviews are so heavily relied upon that they influence product sales and even stock prices. Given up to 98 per cent of consumers read reviews before making a purchase, the out-sized role reviews have makes sense.

But should people rely so heavily on reviews?

The answer to this question is much more nuanced. On the one hand, product reviews are easy to access, provided by a third party (not the same entity trying to sell the product) and are often written with good intentions.

On the other hand, academic research, including my own, has shown there are many reasons to suspect reviews are not quite as valuable as they may seem.

Bias in reviews

Many of these reasons stem from a common argument, which is that reviews may not provide an objective, unbiased measure of product quality. Indeed, a number of seemingly irrelevant factors affect the star ratings and reviews that are given.

For example, asking raters to fill out both an overall rating and several attribute ratings leads them to give a higher overall rating when their experience with the product was subpar.

Additionally, filling out a review on a smartphone leads reviewers to provide more emotionally driven, less specific reviews.

The context of product use can also affect ratings given; a winter jacket is rated more favourably when the outdoor temperature is warmer because raters attribute their comfort not to the warm temperatures, but to the coat.

And, receiving a special designation, like being a “Superhost” on Airbnb, can actually decrease average ratings, as raters now compare their experience to higher expectations when determining what rating to give.

Previous research has also documented how the way reviews are displayed affects review readers’ product perceptions. For instance, people often make categorical distinctions between favourable and unfavourable ratings, while being insufficiently sensitive to differences between ratings of the same valence (for example, between 1 and 2 stars or 4 and 5 stars).

Additionally, people often heavily weigh a product’s average rating, at the expense of considering important quality signals, like the number of ratings and price.

AI and fake reviews

More recently, additional concerns have been raised about review quality. Fake reviews can make up a sizeable proportion of available reviews, and businesses that are more affected by these reviews, like smaller, independently owned restaurants, are more likely to engage in review fraud.

Additionally, the proliferation of AI has led to an increase in chatbot-authored product reviews, which can be difficult for both companies and consumers to filter out.

Taken together, reviews can be a useful source of information, but have a number of important flaws and limitations. In theory, providing information about what owning a product is actually like from a neutral, third-party source is extremely useful.

In practice, however, the execution of this vision leaves room for improvement and future research.

The author is Assistant Professor, Marketing, Business Economics, and Law, University of Alberta