Amidst a wave of excitement around the surprising capabilities of generative and conversational AI, the tech and media words continue to reach around and speculate how it applies to their products. Much of that involves local media and marketing as we recently discussed in light of SMB websites.
Now the latest area of applicability has emerged: reviews. We’re talking about the reviews that populate local listings and guide consumer purchase decisions for everything from landscapers to where they get their next pizza. It’s central to reviews pure-plays like Yelp, as well as search giants like Google.
Speaking of tech giants, that’s where AI’s applicability to reviews advances. Specifically, Amazon has announced a new initiative to generate AI-summarized product reviews. The goal is to create a sort of Readers Digest summary box for online shoppers to get a quick sense of meaning from many reviews.
To be clear, Amazon’s efforts here are purely in the context of eCommerce, not local business reviews. However, we’re making the leap that the latter is where it could advance. Indeed, Yelp has already gotten a jump on several additive and fitting AI integrations, including LLM-powered review snippets.
Good Problem to Have
Going deeper on Amazon’s AI reviews play, it’s specifically utilizing GPT-like conversational AI and large language models to derive succinct meaning from reviews. This will take place in a new section on the product detail page that lists product qualities as well as narrative customer sentiments.
Both of these elements are plucked from reviews, representing structured and unstructured data. The former involves keyword tags that display product attributes (think “ease of use”). Clicking any tag will bring users to the full reviews where those terms were used. Amazon already did a version of this.
As for the unstructured elements, these will be summarized narratives that are composed from the larger bank of reviews that sit further below the fold on product pages. The idea is that the reviews volume on several products on Amazon has reached a level that it’s untenable to read through everything.
Though this is a good problem to have in the chicken & egg-challenged reviews game, Amazon has decided that saving users time with summarized insights could serve its users. More realistically, Amazon likely believes that this move will boost key metrics like conversion rates and basket sizes.
Up in Arms
Further delineating the structured and unstructured parts of this effort, they will work well together but also have dynamics of their own. With structured data for example, keyword tags could be a good utility for users to skim features and attributes with the option to jump to related reviews for a deeper dive.
Additionally, where our minds go is a sort of taxonomy for Amazon to structure its site experience around product tags. Beyond the primary purpose as a reviews summary tool on individual product pages, these tags could be useful in enhancing other key Amazon functions like product search and browsing.
As for the unstructured data, the value of summarized narratives is up for debate. Though it could be a user-centric move to save time, we foresee misgivings from sellers about the veracity of AI-summarized reviews. Many are already up in arms about reviews in general, withouth any AI involvement.
We’ll get many of the above questions answered as Amazon rolls this out. It will do so gradually over the coming months to a “broad selection” of products and mobile users. More importantly, the rollout could offer more perspective on whether this type of AI integration will fly or die with local business reviews.
