Drivers and Strategic Takeaways
The Ultimate Guide to AI at Amazon: Ratings & Reviews
Amazon is just the latest entrant into the fray of using AI in summarizing customer sentiment. CarMax and Microsoft published a case study of how they used a similar approach to save 11 years of work optimizing CarMax’s used car product pages.
As of the time of writing, these customer review summaries appear to only a “subset of mobile shoppers in the U.S.”, but in our opinion will likely will be rolled out across all PDPs in short order if proven to improve conversion rates and customer satisfaction.
How does it work?
Below is an example of the Amazon Firestick. You can see there are two levels of AI-generated review summaries. The first level is an overall review summary, which takes all reviews into account to give shoppers a quick overview of what other shoppers thought about the product.
The second level generates a summary of reviews tied to a specific common trait (e.g. ease of use, performance, a specific product feature… etc.).
This is generated for key Positive, Neutral, and Negative traits as reported by customers.
What can I do today about this?
First step should be to see if this has affected your PDPs.
Closely monitor key ASINs to see if this has been rolled out to your U.S. listings (note that you can only see this on mobile right now).
In particular pay, attention to the conversion rate and see if this is having any material effect as compared to the last 7 days of data before August 14, 2023.
Going forward likely will need to think through how you start to solve for the “negative” categories flagged by Amazon as they now stand out a lot more prominently irrespective of the overall star ratings.
There are already examples of two similarly rated products (e.g. both 4.7 stars) with one having all “positive / green” sub-categories while the other has a mixture of “positive”, “neutral” and “negative” review sub-types.
This is a development that will continued to be monitored.