Zhang Ximeng Silicon Valley big data analysis and data science expert Former Senior Director of Business Analytics at LinkedIn Founder and CEO of GrowingIO After the news that LinkedIn was acquired by Microsoft for a whopping price of approximately $26.2 billion on June 13th (U.S. time) spread, many friends asked me why Microsoft wanted to acquire LinkedIn and whether such an acquisition price was reasonable. So I would like to explain from a data perspective why LinkedIn has such a high valuation.The first point is the value that the data itself creates for users. LinkedIn is the world's largest social networking company, with nearly 430 million users. These users are the backbone of society, with a lot of decision-making power and huge purchasing power.
LinkedIn is a data company. User experience generates a large amount of data. The data itself is used to extract higher value and newer products to further serve users. A seamless loop is formed between the three points, creating a lot of value for many ordinary users.
For example, users can find many new job opportunities, new business solutions, new marketing methods, business expansion methods and functions that match their own needs, etc. These provide value to many ordinary users.
Secondly, at the enterprise level, LinkedIn connects ordinary white-collar workers, high-end users and enterprises. Companies need all kinds of talents to meet their development needs. LinkedIn brings together top talent pools (especially in Europe and the United States). The professional social relationships on the platform help companies find the best entry path and improve conversion efficiency.
For example, for a good job position, the conversion rate of talent introduced through recommendation and introduction is several times higher than the conversion rate of people looking for jobs internally, which invisibly improves the efficiency of the company's human resource acquisition costs. In addition, the elite pool on LinkedIn itself is also the best pool of sales leads for many companies to think about their future products. LinkedIn provides a great platform for people looking for business opportunities, from marketing operations to sales.Finally, in terms of LinkedIn’s internal operational efficiency, first of all, by using LinkedIn’s data, we can acquire users in the fastest, simplest, and most scalable way. For example, by analyzing many user profiles, nearly a million companies can be abstractly tracked. By analyzing the social networks among people within the company, we can find the internal decision-making level of the company. By analyzing the social network relationships of sales personnel within a company, you can find people who know decision makers. For a company, previous decision-making cycles required people to make a lot of judgments and collect information, which made it impossible to achieve standardization and unification. But after being driven by data, everyone is very unified, the value brought to the company is very focused, and the final implementation method is also very practical. This will greatly shorten the entire sales cycle, improve the efficiency of acquiring users per unit time, and reduce the cost of acquiring users.
According to Wall Street's analysis for several years, LinkedIn is the second largest enterprise service provider in the world, and the cost of its customer acquisition services is nearly 50% of that of ordinary enterprise service providers.
For the same resources, LinkedIn has an advantage over other competitors in terms of the speed, quality and breadth of acquiring customers, which is one of the reasons why its valuation on Wall Street and the stock market has remained high.
Therefore, we can see that the ordinary user side, enterprise side and ultimately the investment side of LinkedIn services can be seamlessly unified through data. This is one of the core reasons why LinkedIn was valued very highly when it was acquired.Value, parallax, and pain points in the field of big data Data creates various values, but there are also great disparities and pain points in the field of big data. Leaving LinkedIn to start a business is based on the above two problems.(1) First, in my work over the past five years, I have seen the creation of various values of data. I think the first core of the value generated by data is based on belief, that is, whether we really realize that we are sitting on a gold mine of data . Such belief is ultimately interpreted as imagination and creativity before the belief is realized. Data itself is the source of our creativity and imagination. Imagination is abstracted into its product form and value embodiment. This is one of the reasons why I returned to China to start a business. Over the past decade, the Internet industry has undergone two revolutions. The first PC website revolution led to the birth of a series of successful start-ups, such as a series of Internet website innovations from portal-type to platform-type transaction-type. Starting from 2008, a new wave of mobile Internet emerged. But China's overall economy has transformed from an incremental or high-growth customer acquisition economy to a stock or value-based economy. Taking smartphones as an example, the annual growth rate of their ownership has been between 50% and 30% in the past few years, but in the past two years, especially in the past year, the annual growth rate has hovered around 5%. The number of new customers is slowly declining and the market is approaching saturation. In this case, companies need to transform their previous traffic-driven thinking into a lean operations thinking. This lean thinking has been implemented in many countries in Europe and the United States for several years. My first job in the industry was to do data-driven lean operations. I believe that this kind of thinking is a stage that society must go through in its development. If companies hope to gain a foothold in the future competitive landscape, they must make achievements in this area.
There are many apps nowadays, and the time a user spends on a certain app or website is very limited. How to improve conversion rate per unit time is the first fact that companies need to pay attention to today. The second fact is how to keep users after using the APP. The third fact is how can a good product be spread in a very low-cost way?
The era of using large amounts of money to buy traffic is gone. We must acquire customers at a lower cost and faster speed per unit time, especially the ability of customers to continuously generate value on the platform. From this perspective, data-driven is inevitable.
In the early stages of a company’s general development, especially when the Internet demographic dividend era has not yet ended, many decisions may not need to be data-driven. But in today's fiercely competitive environment, if companies want to survive, they must increase their output per unit time . This is also a core reason why data-driven is becoming increasingly important. In addition, previous data decisions remained at the management level, which means that it did not require many people to make decisions, but only the upper level made the decision and the lower level executed it. But in today's era of full product and universal competition, an enterprise needs to let front-end people make decisions using data to promote the overall improvement of various efficiencies of the unit, so that the products produced are needed by many people and user retention is increased.
Finally, the good reputation of the product can be spread and marketed in the lowest cost and fastest way, bringing a lot of value to the company, realizing the company's long-term mission, and contributing value to society. In this process, data-driven becomes increasingly important.
The field of data analysis has undergone two major changes in the past decade. The first change was the ERP era. In the ERP era, some analysis is generally done on transactional data, such as the quality, quantity, and region of transactions. These are generally achieved by exchanging time for space.
Because the computing and storage costs are very high, people have to spend a lot of time converting large amounts of unstructured data into very structured, small data for storage. This requires heavy manpower. This era lasted until around the year 2000, and many companies are still experimenting with this model today.
But in the past five to ten years, a new analytical framework has emerged in Silicon Valley, a big data framework based on large amounts of unstructured data and with parallel computing as its core, which is the concept of big data that is widely known to the world today. The framework of this system and technology is based on parallel processing, unstructured, and large-scale storage computing capabilities, and requires companies with high computing power and information framework capabilities to implement it.
If traditional analysis solutions can be implemented with four different tools , today it may require dozens of different tools and languages to be integrated together to realize a truly large-scale analysis system. In the past decade or so, many Chinese companies have been lacking in technology in this area. We are not like the United States, which has been operating its economy for decades and has been developing unstructured data solutions for the past decade or so. If companies want to gain a leading edge and generate efficiency in the future competitive landscape, they can seize the opportunity to overtake others, get rid of the burden of traditional analysis, and use new methods to achieve large-scale deployment in the future. For example, a methodology for analyzing intelligence in various departments of an enterprise is a methodology that is implemented using new technologies and new tools. Cloud technology is already very mature, and various analytical theories have become products. There is no need for companies to spend a lot of time, cost and manpower to build a large team. From this point I would like to talk about the issue of connection again. In fact, an analyst helps a company provide analysis methods such as methodology and conclusions, which is the same process as a doctor making a diagnosis. When making a diagnosis, a doctor must first collect some basic information, such as blood tests and X-rays. Due to the imperfections of various systems in the past, he needs the funds to build a blood test system and an X-ray system, just like within a company.
But today, we no longer need to build it ourselves, which saves a lot of time. In addition, doctors diagnose patients based on data judgment. The same is true for data analysis. By understanding the history of a certain event, we can determine its cause and predict its future.
Today's big data analysis system hopes to automatically replicate doctors in large quantities and regularly, so that everyone has a doctor to accompany him. This is a core analysis theory of big data, namely the popularization, executability and simplicity of data.
(2) There is a huge disparity and pain point in the field of big data, which is the problem we want to solve. For example, in the past ten years of our work, we found that many companies even have a lot of trouble with data collection. Several Internet companies where I worked before spent a lot of manpower, material resources and time on data collection, organization, cleaning and even basic standardization.But in fact, the real process of creating value lies in the application of data. Only when it is applied on the business side, service side, and client side can it create real value for people. The problem we want to solve is how to get people, especially engineers, analysts, product managers , and operations personnel, to spend their time on business rather than on data cleaning and collection. Therefore, we developed a new technology called a data collection solution without embedding points . Basically, full information collection is achieved on websites and apps without the need for engineers to do a lot of management, helping companies save a lot of resources in the process. In addition, we recommend that these data be immediately accumulated as the base for enterprise data analysis to ensure real-time analysis.The value of the data itself will gradually decay over time. The closer the data is to us, the more likely it is to increase customer conversion rates in the future. The farther the data is from us, the less impact it will have on potential users. The system we are developing today needs to be real-time under super-large-scale processing conditions, which requires the use of technology. At present, the system we have developed hopes to enable many people to use data to make decisions, provide business analysis intelligence, and help front-line employees within the company make decisions and judgments. This is different from previous design ideas. But achieving this goal is actually very difficult.When deploying and implementing this solution, enterprises basically have three concerns, two of which come from people's cognitive problems. The first point is whether companies can realize the huge value that data can bring. The second point is whether the company has certain experience and methodology to implement the entire process of data-driven decision-making. Why is this a big challenge? Because China is developing very rapidly, there has not been much data-driven experience accumulated in the past decade or so. Therefore, we hope to export the methodology directly to enterprises to help them quickly obtain theoretical and practical operational foundations and solutions. The third point is practical operation ability. A lot of enterprise software itself requires a lot of people to spend time learning in order to overcome this huge and slow learning barrier. We must try our best to make our products simple and easy to use so that many internal users can use them at the same time. This is also the mission of our company. We cover the three areas of value recognition, methodology implementation and final operation, hoping to help more companies improve production efficiency, create more customer value, and truly use the company's internal resources to the best effect. Why Einstein was smart: The importance of connection The data itself is very similar to the understanding of the human brain in brain surgery, including LinkedIn's social network and everything connected by the data and the previous structure of the human brain. The real value of these things lies in the connection. Perhaps you have read an article about Einstein before. Einstein's intelligence is not due to his large brain capacity, but because of the many connections between his brain cells. From a data perspective, the value of a piece of data itself may be 1, but after several data points are connected, its value is no longer a simple addition of numbers, but a geometric increase. When different data sources, devices, and businesses slowly aggregate this data together, the value generated is enormous. This kind of correlation and geometric iteration is a manifestation of huge value.We found that a person's social influence is also related to the density of his social network. A person's status in society is directly proportional to the number of his or her connections, that is, the number of a person's connections reflects his or her potential value in society. If there is various collaborations between employees of two organizations, many internal connections are potentially formed. When two social networks are pushed in the same direction, it is a process of organizational merger and acquisition. Sometimes the challenges of mergers and acquisitions come from the fact that people in the two organizations do not know each other, which creates a lot of friction when collaborating. One of the analytical methods that the human resources department does during the merger and acquisition process is called talent analytics, which analyzes the social networks between people. Many things are learned by analogy. A mathematical method is abstracted from a physical fact, which in turn explains the connections between many businesses in physical reality and amplifies these connections. We can use the social relationships between people to deduce the structure of an organization, and then deduce who is qualified for what position and predict the future development direction of individuals. These potential things are told to us by the data.
What the data tells us is a basic physical fact and future trend of the entire universe, that is, we are constantly evolving from low to high, from coarse particles to fine particles, and constantly evolving upward in this process of energy.
Information is one of the most advanced reactions in the process of energy transmission. It is full of value, energy, creativity and possibilities.
