Some Yelpers Soured On Whole Foods After Amazon Acquisition
Others Warmed to the Discounts
By Jenny Yu and Yinghan Fu
In June 2017, Amazon announced a $13.7 billion deal to acquire Whole Foods Market. Today marks the one-year anniversary of the acquisition. A vocal minority of Yelp reviewers — those who mentioned Amazon in their reviews — soured on Whole Foods, even before the deal closed. The takeover of Whole Foods by a tech giant has harmed the reputation of this grocery chain among a subset of consumers.
Since the announcement of the acquisition, some consumers started to mention Amazon in reviews of Whole Foods. About 5% of all Whole Foods reviews mentioned Amazon during the two months between the announcement and the acquisition. The share of newly posted reviews mentioning Amazon surged to 10% to 15% after the acquisition, and has risen to almost 20% in the past two months.
Consumers who mentioned Amazon in their reviews of Whole Foods gave lower ratings than those who didn’t. Reviews of Whole Foods that do not mention Amazon have an average rating of 3.2 stars, a level that has remained stable over the past two years. However, the average rating of reviews mentioning Amazon is about half a star lower than ratings of reviews that don’t mention Amazon. The divergence in ratings of these two types of reviews started right after the announcement of the acquisition, and has remained nearly constant over time, indicating a negative perception of the acquisition among a small group of consumers — which began even before any real changes to Whole Foods stores from the acquisition could have taken effect.
For the most part, the people writing Whole Foods reviews on Yelp before the announcement of the acquisition are different than those writing reviews afterwards — or they’re reviewing different Whole Foods locations. After all, each Yelp user only gets to have one active review (and star rating) of each business. There is, however, one subset of Yelp reviewers who have reviewed the same Whole Foods location before and after the announcement, using a Yelp feature that allows users to update a review after it has been published. We investigated how those users changed their reviews to see what they thought of the new ownership. We found that their perception of the acquisition was markedly negative.
Reviews that were both initially published and updated before the announcement only changed the average rating by less than one-tenth of a star. The same is true for reviews that were originally posted after the announcement and then updated. However, the picture is totally different for Whole Foods reviews that were originally published before the announcement but updated afterwards. Reviewers who wrote updates after the announcement that mentioned Amazon lowered ratings by, on average, more than half a star from the ratings in the original reviews published before the announcement. In comparison, reviewers who wrote updates after the announcement that did not mention Amazon decreased the average rating by about three-tenths of a star from their original reviews published before the announcement.
Some examples of the review updates suggest the negative rating change of these updates is at least partly caused by the consumers’ perception of the Amazon acquisition.
Jade M. updated a three-star review published before the announcement to a two-star one afterwards. An excerpt from the updated version: “I wasn’t too impressed when Amazon acquired Whole Foods Market…Customer Service — is lacking in terms of friendliness and knowledge…You can even see advertisements of NOM NOM Paleo all over the hot and salad bar — so Amazon is taking part of the advertising”
Alexandra J. updated a five-star review published before the announcement to a three-star one after the announcement. From the updated version: “I get the impression the quality has gone down. The employees don’t seem to love their jobs as much anymore, so I wonder what happened behind the scenes with Amazon.”
To further understand how consumers perceive Amazon within the context of Whole Foods reviews, we processed the text of reviews using a two-layer neural network model called Word2Vec. The model is able to learn word similarity in the context of these reviews, which can reveal what words are perceived similarly by these consumers. By examining the words that are most similar to “Amazon” and “good” and most different from “bad,” we are able to understand what Whole Foods shoppers tend to associate with a positive image of Amazon. The top words related to “Amazon” and “good” are mostly related to discounts and reduced price. (Amazon Prime members get discounts at Whole Foods.)
The same method can be used to identify the top words Whole Foods reviewers associate with negative views of Amazon. The top words perceived as similar to the combination of “Amazon” and “bad” are mainly about the acquisition itself. The model learns from the review text and correctly associates the name of “Jeff” “Bezos” with “Amazon,” and the Amazon founder and CEO’s first and last names are on top of the list.
Our analysis demonstrates that Yelp reviews reveal how consumer sentiment changes after a major acquisition. Time will tell whether the potential advantages of the acquisition for shoppers will turn consumer sentiment around.
Peter Weir, Carl Bialik, Liina Potter and Travis Brooks contributed to this blog post. Graphics by The DataFace. All stats are based on Whole Foods locations in the U.S. and Canada.