Many successful companies, such as Amazon, Walmart, and NETFLIX, actively use data, and everyone recognizes the importance of data in improving business results. Nevertheless, there are probably many people who feel uncomfortable with data and feel that it is something that is far removed from them. 

In this article, we take another look at what can be achieved through data science and the great potential it has, and also speak to XICA CEO Yoshiaki Hirao about the meaning and significance of companies today engaging in data science. 

CEO of XICA Corporation
Yoshiaki Hirao 

After his father's bankruptcy, he strongly wanted to "eliminate the hopeless sadness in the world." He encountered statistical analysis while studying at the Faculty of Policy Management at Keio University, and founded XICA Co., Ltd. in February 2012, just before graduating. Before founding the company, he also had a unique career as a band member. 

The promise of data science to achieve "disruptive growth" 

── In recent years, attention to "data science" has been rapidly increasing. What do you think is the reason for this? 

The situation surrounding data science has changed dramatically since I founded XICA nine years ago in 9. Although data science was considered important back then, it has only been in the last five years that it has become more widely known. 

In the early days of data science, the importance of collecting and accumulating various phenomena as data was discussed under the keyword "big data." After that period, interest in processing and organizing collected data grew. Around this time, the market grew significantly, with companies providing BI (business intelligence: organizing and visualizing big data to support management decision-making) tools going public one after another. 

In recent years, as there has been an increasing need to analyze organized data and use it for "next actions," data science has naturally begun to attract attention. 

There is no point in simply "collecting," "storing," and "organizing" data; data needs to be "analyzed" in order to achieve goals and solve problems.--Looking back now, it seems like an obvious conclusion. However, as the amount of data circulating in the world has become huge and diverse, and tools for managing and utilizing data have become widespread, I think that a change that has occurred over the past 10 years or so is that more and more people and companies are thinking, "I want to make effective use of the data I have at hand." 

In addition to this, several events have accelerated the growing momentum for data utilization. 

One book that changed the tide the most was the publication of "Statistics is the Strongest Science" (by Nishiuchi Kei) in 2013. Statistics and data science became widely known and recognized by the general public, including people who were not normally familiar with data. I think it was around this time that data scientists and other people who handled data started to be recognized as "cool" (and lucrative). 

The movie "Moneyball" (*1) was released around the same time as GAFA began to grow at an accelerated pace. Stories of people and companies who had succeeded by making full use of data were packaged and shared as "cool," and data science quickly became something people were paying attention to. 

One key point is that each of these cases is an example of "disruptive growth" achieved through the use of data. 

The Oakland Athletics in Moneyball made it to the playoffs four consecutive years from 2000, and twice achieved the feat of winning 4 or more games in a season. GAFA, which started out as small startups, have risen to become major companies in a dramatic way by making full use of data. 

Data reveals things that were previously invisible or unnoticed."If we make full use of data, we may be able to win even without abundant resources."I think this expectation is what has attracted people even more to data science. 

In an era when it has become difficult to win or survive by continuing on the same path as before, I believe that data science has been embraced as a long-awaited "weapon" by many people and companies. 

(*1) Moneyball 
This film depicts the life story of Billy Beane, a real-life baseball team general manager who turned a struggling Major League team into a perennial winner with his unique Moneyball theory. The Moneyball theory uses sabermetrics, a method of objectively evaluating players based on their statistics.

The two effects of data science: "Increasing certainty" and "Bringing about unexpected discoveries" 

── What exactly can data science achieve? 

In a nutshell"Increasing certainty"Whether it's politics, business, or sports, the power of data science is to enable you to "win when you are meant to win" and "increase your chances of winning." 

Therefore, data science can also be described as "the process of deriving rules for winning." 

There are two main ways to use data to "increase certainty": "verifying a hypothesis" and "getting hints for formulating a hypothesis." 

Taking a company's advertising activities as an example, data can be used to verify the hypothesis, "In order to increase sales, should we strengthen advertisement B rather than advertisement A?" If the hypothesis is contradictory,Exploring previously unimagined advertising media and creative possibilitiescan also do. 

Of course, advertising is not the only field where data science can be used. The range of things that can be achieved by "deriving winning principles" is wide, and the following are some examples of what has been achieved with data science: 

As long as you can digitize phenomena, you can perform data science. 

With advances in technology, almost all phenomena in the world can be digitized in some form, and any data can be analyzed in some way. It is no exaggeration to say that there is no goal that cannot be achieved or problem that cannot be solved by data science. 

Because it can be used to achieve any goal or solve any problem, there are countless examples of companies that have succeeded by using data, regardless of their size or industry. 

NETFLIX started its DVD rental and sales business by mail order in 1998 with only 30 employees. The company hired data scientists with programming skills and promoted recommendation functions, personalization, and the planning and production of original content using data analysis. This led to exponential growth for NETFLIX, which achieved sales of $2020 billion (approximately 250 trillion yen) in 2 (*5915). 

Walmart has risen to become the world's largest retail chain by leveraging its data analysis capabilities. The company is adept at analyzing not only quantitative sales and inventory data, but also qualitative data such as regional needs and changes in needs due to weather. It is also known for supporting the growth of the entire industry by providing the data and analysis results it has obtained to its suppliers (*3).  

There's no need to go into too many examples here; using data to improve business outcomes is becoming as commonplace as breathing. 

(*2) Source: "Analytics: The Power of Business (Thomas H. Davenport and Jane G. Harris, Nikkei BP, 2008)" and "Harvard Business Review (https://hbr.org/2018/01/data-can-enhance-creative-projects-just-look-at-netflix)”, “Netflix, Inc. Financial Results (https://s22.q4cdn.com/959853165/files/doc_financials/2020/q2/FINAL-Q2-20-Shareholder-Letter-V3-with-Tables.pdf)'  

(*Four) 
Source: "Analytics: The Power of Business" by Thomas H. Davenport and Jane G. Harris, Nikkei BP, 2008 "

The most important thing in data science is "experience and intuition" 

── How exactly do you go about data science? 

Data science consists of eight processes: ❶ to ❽. 

1. Purpose: What do you want to achieve? 
❷Challenge: What should we solve first? 
❸ Hypothesis: What is the cause? 
4. Data: What data do you need? 
❺Analysis: What analysis should be performed?
6. Interpretation: What decision should be made?
❼ Involvement: How to move forward within the organization
❽ Execution

Looking at steps ❶ to ❽, the word "data" appears exactly in the middle of the entire process. You don't start working with data right away. Also, only steps ❹ and ❺ require expertise in statistics and data analysis. 

Data science is actuallyThe proportion of elements other than "data" and "analysis" is higher.Therefore, I think there are many people who think of data science as something that "science people do" or "something that requires statistical expertise," and feel that it is something that is far removed from them, but this is a big misunderstanding. 

In fact, the most important thing in data science isSet the direction of your data analysis with "purpose," "challenge," and "hypothesis."To do. And"Interpretation" involves finding relationships in the data presented to you and turning it into a storyIs to do. 

If this is neglected, no matter how much data you obtain or how highly accurate your analytical methods are, you will not be able to derive useful insights from the data or translate the analysis results into results. 

On the other hand, as long as you have the "purpose," "issue," "hypothesis," and "interpretation," you can leave the "data" and "analysis" to experts. 

── It seems that even people with so-called “liberal arts” backgrounds can practice data science. 

In fact, it may be that in many cases, people from the humanities who have a vague dislike for data science are more suited to it. 

To imagine causal relationships between various phenomena in the world and form hypotheses."Hypothesis ability" is an essential skill for practicing data scienceThat's why. 

Long ago, a certain convenience store chain hired a large number of data scientists to conduct data analysis to find "factors that affect sales," and the answer they arrived at was the obvious one: "Sales decrease when it rains." Of course, this result itself is not wrong, but it is an anecdote that clearly shows that you cannot get useful suggestions from data if you analyze it without a hypothesis. 

Without a hypothesis, some phenomena would not be quantified or analyzed.Without the ability to hypothesize, there are things you cannot notice or learn.As analytical technology becomes commoditized in the future,"Hypothesis power" not only gives you an advantage in data analysis, but also in the business itselfIt can be said. 

And the basis of this "hypothesis power" is,Each person has their own "experience and intuition"I think 

For example, "rain" is generally said to have a negative impact on store sales. When converting this phenomenon into data and analyzing it, digging deeper into a hypothesis such as "sales decrease when it rains" as shown below will increase the chances of obtaining useful suggestions. 

■ Examining the amount of precipitation based on the hypothesis that the more rain it rains, the lower the sales will be.
■ Under the hypothesis that sales will decrease when the amount of rain exceeds a certain level, we will investigate the relationship between precipitation and sales.
■ Based on the hypothesis that sales differ depending on the amount of precipitation, rather than simply being high or low, we categorize the amount of precipitation as ○mm to ○mm, △mm to △mm, and □mm to □mm, and examine the impact of each on sales.
■ Based on the hypothesis that the combination of precipitation and temperature affects sales, we will investigate the relationship between precipitation as well as temperature and sales.

Having experience in retail is a great advantage when formulating these hypotheses and analysis plans. Having experience that "a certain number of customers visit the store even on rainy days" and intuition that "sales seem to fall on days when it is raining and the temperature is high" allows you to formulate sound hypotheses and think of appropriate ways to turn them into data. 

Data science is something that every person, including managers and marketers, should possess.A tool to make the most of your experience and intuition in achieving goals and solving problemsIt can be said. 

Data science brings us to an era where we can compete based on our "human skills" 

── It seems surprising that experience and intuition can be useful in data science. 

There was a time when people, especially in the marketing industry, said that "data science is a methodology for discovering patterns and increasing reproducibility based on data, without relying on human experience or intuition," but this is a big mistake. 

I believe that this misconception has created resistance to data science among ordinary business people other than data scientists, and has been one of the factors hindering the spread of data science. 

Data science is a means to make the most of people's experience and intuition. By spreading this correct understanding, I hope that more people and companies will put data science into practice. 

── How will society change as more people and companies begin to practice data science? 

"I'm fully prepared, now I just need to compete with my own talent."In fact, the desire to create such a society was what led to the founding of XICA. 

Currently, data science is a "haves and have-nots" situation.Polarization between companies that can put this into practice and those that cannotAs a result, a situation has been created in which the former continues to win. 

If we can spread data science to more people and companies, we will be able to win in the end with our "human power." In business, we can use our human power to develop a customer perspective cultivated by continually interacting with customers, the observational skills to identify trends in society and the market, our own unique sensibilities and intuitions, and the courage and passion to face things head-on.The time is coming when we can make better use of uniquely human abilitiesI think. 

To create a world where people can compete with their own talents without feeling frustrated and regretful thinking, "If only we had more resources (people, things, money, information) we could have won." That is my goal, and it is also the potential of data science. 

[Interview and text] Chiaki Saito 
[photograph]Daisuke Koike 
[Planning and editing] Yuko Kawabata (XICA)