How do reviews on the Internet affect the restaurant’s revenue?
American economists have found that raising the restaurant’s rating even by a half a point on a five-point scale significantly increases the number of visitors.
This is what every restaurateur (and even more widely: any business owner in the service sector) knew even before the Internet era: bad reviews reduce the number of customers and reduce the amount of revenue, and good reviews – vice versa.
However, not all owners of such businesses today realize how important the ranking of a place among Internet users is for the success of an enterprise.
Nowadays, when, when thanks to technology, everyone can become a critic, ratings have become more important than ever.
The work of two Californian economists, Michael Anderson and Jeremy Magruder, published in the September issue of Economic Journal, is the first attempt to trace the relationship between the Internet rating and the decision to buy or visit a place.
Scientists investigated the attendance of 300 restaurants in San Francisco and compared it with the ratings of these places.
It turned out that the improvement in the rating even by half a point (on a five-point scale) increases the attendance of the restaurant during peak hours, that is, starting at 7 in the evening. So, if before the increase in the rating of places in the restaurant was not in 30% of cases, after – already in 49%.
This increase was in no way connected with changes in the menu, improvement of the quality of food or service, the scientists note:
The results of this study show that, although social networks can directly return investments to restaurateurs, they nevertheless play an increasingly important role as consumers judge them about the quality of goods and services.
It is worth noting that the study has a certain amount of error due to the fact that the points on Yelp.com are rounded off. Thus, a restaurant with a rating of 3.74 will be rated at 3.5 points, and those who score 3.76 will be at 4, although there will be practically no difference between them.
However, scientists note, changes in consumer preferences occur even when the quality of food and service does not change. The effect of the rating is even stronger when it is problematic to find information about the place in places.
The researchers considered the question of the reliability of ratings:
Of course, in such a situation, restaurateurs have an incentive to independently influence the rating and leave fake reviews on the site. However, Yelp.com has a wide range of tools to verify such fraud.
Nevertheless, if you look at other appraisal services, including in Russia, we will encounter the fact that restaurateurs and owners of other businesses in the service sector strive to influence the Internet ratings.
There are known precedents when bots left many positive reviews of a particular place or company in order to raise it to a higher position in the top. Often we can encounter examples of such reviews on the site of the Afisha magazine, in Yandex.Market.
As a rule, business owners who use these tricks do not think too much about making their reviews realistic. The reviews are usually of the same type and appear in large numbers in a short period of time. Users who leave them, as a rule, are registered on the service recently, they have no reviews of other places. Often they don’t even have a userpic.
Some services have already begun to deal with fake reviews. So, TripAdvisor marks with a red flag the names of those hotels for which a suspiciously large number of rave reviews appear.
According to the company Gartner, by 2014 about 10-15% of all reviews posted on social networks will be fake.
Social networks themselves begin to slowly but surely fight this. So, Facebook in September 2012 strengthened its automated system aimed at the removal of dubious likes.
Note that, until gaining control over fake tags, about 1% of all likes were removed as dubious.
Scientists are trying to create their own systems that would distinguish fake reviews from real ones. So, in 2011, researchers at Cornell University in the United States created a similar algorithm. True, it was tested only on reviews of 20 hotels in Chicago, but calculated the fake in almost 90% of cases. After studying a lot of reviews, scientists have found out which words are often found in fake reviews and very rarely appear in real ones. Based on this, the system was developed.
This method, with all its high efficiency, in my opinion, is rather flawed. In my review, I have the legal right to use any words without fear that the review will be absorbed by the spam filter. The widespread introduction of such a system could be useful with a certain refinement of it. For example, it would be worthwhile to teach the system to recognize avatars, track the date a user is registered on the site and his other, in addition to this review, activity on the service.