Business review data, also known as company review data, includes reviews about companies, their points of interest, and the quality of the company’s service or employment. These reviews are found on public company review sites and may be written by customers or employees. Both review types provide important information that you can use to improve your company culture, products, or service.
Most company reviews are available free online, with a convenient 5-star rating system that can be easily quantified without signing up for the site’s analytics and feedback service.
Additionally, current and former employees, writing anonymously, pen the largest number of company reviews compared to managers or C-level officers. As these are lower-level employees, they may only feel truly comfortable providing honest feedback anonymously online.
Social media sites, surveys, and feedback requests also provide a great amount of business review data.
As noted, reviews with rating are easy to quantify, but individuals rate certain information as more important than others. They may also completely omit some factor that you are particularly focused on. For instance, some reviews only speak to a company’s work-life balance but neglect to mention career advancement opportunities. With this kind of written content, natural language processing (NLP) programs are able to draw actionable meaning from these reviews.
Business people use this data to analyze competitors, manage crises or projects, predict prices and stock changes, monitor brand engagement, plan changes or improvements to products and services, and identify and target market audiences.
Investors and lenders use the data to determine which companies would be good investments.
Finally, potential employees use this data to determine which companies to apply to or to avoid.
As always, your data set must be complete and frequently updated. Beyond that, while web scraping tools may be useful for finding mentions of your company online, any reasonably active review or e-commerce site will likely have good quality reviews as there are increasingly strict consequences for fake reviews.
The quality of your NLP program may be more of an issue. However, a good program that can grade user sentiment down to the sentence fragment and recognize neutral sentiments is not too difficult to set up or maintain.
The most important factor to consider when vetting company review data is the authenticity of the review and the reviewer—not the rating. It serves the company’s interest (and reputation) to leave negative reviews up as long as they are authentic for two reasons:
1. They may provide important information for managers and executives about the company’s product or service. (And provide an opportunity for the company to publicly show their service to readers.)
2. Even negative reviews are useful to potential customers, particularly as they seem more honest than a totally glowing endorsement across the board.
Using our Bi-LSTM model, we can accurately predict the sentiment of a review on a 1-5 star scale 46.4% of the time. This model shows initial signs that Glassdoor reviews have the potential to help managers assess internal reviews.
Quandoo Restaurant Database tracks reviews, cuisine type, and location data on more than 17,000 restaurants all over the world
Product Hunt Database consists of tech company products, projects, and site news. Members also receive a daily curation of new offerings to comment and vote on
PagesJaunes Database contains information about professional services as well as their locations, departments, and reviews