Underhood measures the online reputation of companies.
In this article, we explain the basic principles behind our technology and algorithms. We welcome everyone to comment, share their research and ideas, and help us. In this way, we hope that we can give something back to the research community.
At Underhood, we treat all companies equally. Everyone can see everyone else’s metrics; we take objectivity and transparency very seriously.
Our philosophy can best be summed up by the following quotation from Daniel Kahneman, the Nobel Prize–winning expert on behavioral economics:
Whenever we can replace human judgment by a formula, we should at least consider it
Kahneman (2011)
Reputation is earned. It’s the way the world sees a company.
Reputation is not just a public image of a company, nor is it a brand. It is created within the company and is deeply linked to the company’s mission, respect, trust, and other emotions—and ultimately to the reason the company exists.
Corporate reputation rests on assessments made by individuals outside the organization
Burke (2012)
Underhood measures the reputation of companies by analyzing all the main stakeholder groups:
Currently Underhood analyzes social media discussions about companies, but we will add metrics from other stakeholder groups in the near future.
Yes, it is. Reputation is directly linked to the financial success of a company. According to Deloitte, most executives take reputation very seriously:
Reputation problems have the biggest impact on revenue and brand value
Deloitte (2014)
A company with a good reputation is loved by its customers, gets better business deals, saves on employee costs, and attracts investors.
Everyone wants to play with a good guy, so the good guy makes money!
Social media has become an essential communications arena for all kinds of companies. Even the most boring and traditional B2B companies are active on social media nowadays, especially if they don’t want to be left behind.
This change has been swift, and not everyone has understood the magnitude of the revolution yet.
Never before have companies been so out of touch with their communications than now
Aula & Heinonen (2015)
If the company isn’t interacting with its customers and other stakeholders on the internet, it’s taking huge risks. It’s like driving a car blindfolded.
Luckily, social media provides a lot of reputation data that can be measured. This is where Underhood can help.
Underhood analysis is based on the data we collect daily from the internet. The main data sources are Facebook and Twitter
Social media data is normalized by dividing it by the number of Facebook likes on the company’s page or by followers on Twitter, depending on the data we analyse. This is how we make different kinds of companies comparable.
After normalizing, every data point is given a value between 0 to 10.
The scales are calculated from the existing company data and calibrated daily.
The Underhood score is an average of sub-scores (Similarity, Dialogue, and Visibility). It shows the reputation of a company on a scale from 0 to 10, based on online data.
The Similarity score measures the word similarity of a company and its audience. It is commonly believed that companies should communicate with the same language their audience uses—that’s successful communication. So the more common words, the better the score.
We also measure the audience tone of voice by using sentiment analysis on the comments a company receives on Facebook. Soon we will add Twitter and other social media services to analysis too.
Sentiment analysis is typically used to determine the attitude of a writer towards a certain topic – in this case, towards a company.
All the texts are translated to English before analysis.
Dialogue score measures the quality of the dialog a company has with its audience.
It’s important to be active on social media, but just posting something daily isn’t enough.
Research has shown that genuine dialog with the community can create a better reputation and lead to financial success. This makes the response rate to audience comments a revealing meter: If a company posts to social media and gets replies from the audience but doesn’t bother to continue the discussion, the Dialogue score goes down.
We also measure the average shares, likes, and comments a company gets. Note: The rating depends on the number of likes/followers. With a big audience, a company needs to have more reactions than firms with a smaller fan base.
The impact of the richness and responsiveness of a firm’s social media efforts on the firm’s market performance is positive and significant.
Chung et al. (2015)
Visibility score measures how well known a company is.
On average, famous companies perform better than unknown ones.
Prominent visibility is a good thing, because the more prominent a company is, the more likely it is that potential customers, employees, and investors will find it.
A firm’s public prominence is the degree to which it receives large-scale collective recognition, public attention, and salience in the minds of stakeholders.
Rindova et al. (2005)
Underhood measures prominence with its Visibility score, by analyzing how many followers a company has on social media. These values are not normalized. But they are compared to other companies to judge if the result is bad, moderate, good, or excellent.
Facebook buzz is calculated by dividing the Facebook “talking about this” value with the number of page likes.
What you see in this current version of Underhood is just a start.
So far, Underhood has been focusing on social media, because that’s where we can get the most data related to reputation. But we’re already collecting data from many different APIs and other internet sources, including traditional media articles. So in the next Underhood versions, we’ll use that data in our analysis.
According to studies, a company’s performance is closely linked to its reputation, so Underhood will find correlations between company reputation and financial success. By doing that, we can show the current situation of a company.
The method we’re using is called nowcasting. According to Wikipedia, “Nowcasting is defined as the prediction of the present, the very near future and the very recent past in economics. The term is a contraction for now and forecasting and has been used for a long-time in meteorology.”
Most financial analysis services focus heavily on forecasting the financial success of a company. But the future is harder to predict than the present. Anything can happen tomorrow, but until then, most facts can be analyzed.
To put it briefly, Underhood aims to show weak signals about companies, by focusing on reputation analysis.
The other term closely connected with Underhood is “alternative data,” which is becoming more and more popular among investors.
Business Insider UK described alternative data as a method by which “obscure data sets can be turned into tradable information” and as “anything that is raw or unstructured, and is distinct from things like company filings, historic market prices, or investor presentations.” Alternative data can be anything from satellite images of grocery chain parking lots to social media activities.
A famous example of alternative data’s possibilities was when Foursquare was able to predict the massive decline in US-based restaurant chain Chipotle’s sales in the first quarter of 2016, when the check-ins to Chipotle’s restaurants dropped.
Underhood’s goal is to become a valuable tool for investors who don’t want to rely solely on the official information provided by companies.
There’s so many possibilities and so much work to be done! If you’re interested in unstructured data, machine learning, natural language processing, econometrics, and social media analyses—or any other field we need to look at—we invite you to participate in the development of our methods, algorithms, indexes, and especially the science behind our operations.
We’ve already partnered with some research institutions, and we’re willing to give our data and algorithms to researchers in this exciting field of alternative data, nowcasting, and reputation analysis.
If you feel we can help you with your research or in any other way, please email us at science [at] underhood.co
Kahneman(2011). Thinking, Fast and Slow. Farrar, Straus, and Giroux.
Aula, Heinonen (2015). The Reputable Firm. Springer.
Burke. (2012). “Corporate Reputations: Development, Maintenance, Change and Repair.” In: Corporate Reputation. Eds: Martin, Burke, Cooper. Gower, Abingdon.
Deloitte (2014). Reputation@Risk.
Wikipedia. Sentiment analysis.
Chung, Animesh, Han, Pinsonneault (2015). “Firms’ Social Media Efforts, Consumer Behavior, and Firm Performance: Evidence from Facebook.” In: CORS/INFORMS International Conference, Montreal, Canada.
Rindova, VP; Williamson, OA; Petkova, AP; Sever, JM. (2005). “Being Good or Being Known: An Empirical Examination of the Dimensions, Antecedents, and Consequences of Organizational Reputation.” The Academy of Management Journal Vol. 48, No. 6, pp. 1033-1049.
Wikipedia. Nowcasting (economics).
Turner (2015). “This is the future of investing, and you probably can't afford it.” Business Insider UK.
Glueck (2016). “Foursquare Predicts Chipotle’s Q1 Sales Down Nearly 30%; Foot Traffic Reveals the Start of a Mixed Recovery.” Medium.
Your subscription has now started!
Manage your subscription on the billing section on your profile page.
