How do traditional companies find their anchor points in the big data age?

In Hong Kong, there is a Japanese store. The store swept through the territory in a very short time and opened several chain stores. Many people know that the seafood is very fresh and affordable, only 70 percent of the price. I have also asked the chef friend, what can be done so good business? The chef mysteriously asked me if you saw the camera on each table. That's our secret weapon. "It turns out that the seafood store goes through the camera every day to check the order of the diners, the ordering of the meal, and the weight of the leftovers." Through such an inventory, the restaurant's owner can accurately grasp the consumer's preferences, so that Hokkaido's seafood purchase is relatively accurate. Also because of this, the restaurant's source of circulation quickly, the cost is also reduced. This is an interesting case. A traditional restaurant without ERP system, through the camera to achieve the procurement of information management: the collection of user information, analysis and then for the second day of procurement decisions, cycle repeatedly, in order to reduce business costs. For many people, big data is just a buzzword. At the same time that the data is far away from their business, the traditional enterprise is also afraid of the future: what kind of business can use data? What specific business problems can data solve? Who needs big data? The store, a century-old traditional retail store, began  data on competitors ' prices six years ago. Recently, they have been doing a dynamic pricing engine, and are also working on automated systems that match the product to the crowd. In the field of electric business, we can divide the user's cognition into three kinds: the visitor, the purchaser and the consumer. Traditional grocery stores don't know which stores people walk into the store (browsing the data), what they buy in each brand store (the data they buy), what bank cards they pay for, not to mention the experience data they use when they are finished shopping. The most painful point of manufacturing enterprises is that I know who is helping me sell, but I don't know who is buying it. The problem with retailing is that I know who is buying, but I don't know how the guests make the decision, or how they use it or what problems they have. This is because the old model, the data can not be traced to the store, resulting in production and use is disjointed. But in the big data age, production companies can use social data and even sensors to track how users are used. What's wrong with the product, the production enterprise can even understand the problem before the user perceives, and provide the solution. What if traditional department stores could have this data? They can know what brands their members prefer, what kind of payment they prefer, and can send orders to manufacturers to purchase items that are in line with their members ' interests. The data can help the retail industry to match the needs of the population with the rapid availability of goods, the greatest value is here. When it becomes easier to get data, businesses will find that they lose a lot of opportunities without data. Every enterprise in the future will become a data enterprise, and each product will become a data product. Because the optimization points inside are all dependent on the data innovation, the data will become the enterpriseThe driving force of development. How to enlarge data with limited resources? Small and medium-sized enterprises in the data is the biggest problem is limited resources, not too many resources for trial and error, trial and error space is very small. Therefore, SMEs should collect key information rather than collect all data. You can choose a smaller scene for data  and analysis. This scenario satisfies the following conditions: 1 Do you have the required data? 2 Accurate data? 3 What is the real time of the data? 4 data and algorithm matching 5 how to learn from the error, the last one of the data reflux can continue to optimize, is that these reflux data can improve our previous understanding. In the case of restaurants in Japan, the choice of consumers is their most critical decision-making basis, so it is a priority to collect such data. The large data is based on the enterprise data integration, algorithm innovation and product. For example, the reason Google Maps can tell you the road ahead of traffic jams, in fact, depends on each use of Google Maps location to share real-time integration. Therefore, I think that the government's promotion will enable small enterprises to reduce the threshold of data and increase the data function of the industry, so that small enterprises can enjoy large data technology. From the industrial chain, the alliance of small companies, the data unified, using data to solve some of the industry can not solve the problem. It is not easy for small and medium-sized enterprises to have large data teams like big companies. Therefore, SMEs in the use of data, there must be a more secure way to pay attention to the use of data efficiency, can try to start from small projects, and then gradually expand. Another noteworthy is that the nature of the operation depends on the founder's direction and management, we can not cart the horse before the horse, just look forward to the data can solve all the challenges of the enterprise. Why is the data fragmented? I recently met a manufacturer who makes computer hardware. Internal production could be computerized, he said, but it was found to be heavily fragmented with sales demand, which seemed to be out of reach. "Why is this happening?" One of my common metaphors is that restaurant owners tend to order menus, but the lowest-level buyers buy them every day. So very few restaurants can often become famous, because the chef can't set the menu, also can not use suitable good raw materials. The innovation of the data is all the time, the innovation period of the algorithm is slightly long, and the innovation of product is often a sword of ten years. Therefore, people who have the right to decide in a business are often the ones who have decision-making power over products. It's hard to find a combination of data and business if you just stand on a single point of view. According to my observations, there is a dearth of data management talent: He wants to have a good understanding of the business, understand what the data can do for the business, understand the relationship between the technology update and value, learn from data collection to processing, the integration of new data and history, the convenience of using data, and so on. The understanding of business and business is definitely the basic requirement for data managers, but if you want to achieve excellence, you must know how to attract and retain talent in a large data industry with a shortage of people. For business people,Ask yourself: Does the data you have now help me solve the problem? Assuming all the data is available, what data do I need to solve the problem? How do you make it easier to get the data you need? For example, when I used to see traffic on the road, I thought about the possibility that taxi services would improve in big cities. I was thinking that if a taxi had a light to show what the customer had done to him in the past, the driver would provide a better standard of service in order to maintain a good rating. This is a simple example of how data can be resolved. The next step is how to design an easy way for customers to evaluate. And now call the car software is a good implementation case. It's a good way to train data sensitivity, and it's the way I've been doing it over the last decade--training data sensitivity through things around, making numbers talk. The most difficult part: your understanding of your own big data applications are about crossing boundaries and innovation, and, more accurately, the value of large data comes from being able to look at the same thing from a multiple perspective, and panoramic observation can reduce errors and create new opportunities. But it's not about asking people to recognize all the outside world, but to make other people's data available to you. The hardest part of big data practice is your understanding of yourself, plus, insiders, the externally integrated data may be valuable but there is also a lot of noise, and you don't fully understand the source and definition of the data. How to see yourself? Based on past experience, I think the first thing is to start small. Traditional enterprises in the initial stage do not start a very large data project. The data is suitable to start from the small and specific, easy to evaluate the effect as a starting point of the project, to exercise their own collection, processing, use of data to make decisions, as well as the ability to measure the value of the data, that is, small knowledge. Starting from a small scene, with data in the business scene optimization. Axciom's chief data officer, Geoscience, once proposed a three-level state of data. : Data 1.0 What data is generated by its own business, what data do we use for analysis and optimization; Data 2.0 intersects existing data with its own historical or upstream data, which optimizes the data, and data 3.0 is the purchase of external data or the sharing of its own data, which is mutually miscible, creating a new product experience in blending. These three levels, all need enterprises have different technology and structure to achieve data extraction, processing and product integration. This is actually a process of constantly using data to describe and restore business. Recently, Ali data team successfully promoted a quick taxi success rate. We superimposed the use of the data once and two times. We integrate real-time data with historical data. The original app in sending a taxi demand, is the location of the taxi people as the origin, every few minutes spread to the vicinity of 300 meters, 600 meters of taxis. The push of this message is based on the location of the push logic. But if nearby drivers actually do not want to go to the destination, the success rate of the orders will be reduced. Therefore, we put the driver priority destination "this data into the push system,The data has been redesigned to make the drivers more willing to take orders "higher". It also improves the overall success rate of the orders. Of course, this is just one of the ideas of optimization. In my opinion, all data products are related to decision making. As a result, data optimization should be traced to every aspect of the decision making in people or machines, constantly updating your anchor points. To break a decision, you first need to know how people make decisions and how to change decisions with new data. What is the difference between the two? What value will it bring? Big decisions are often made up of a series of small decisions. Like a quick taxi. The key to efficiency is how to correlate the driver's data with the user's data, and how to continually cross over the historical data to find the most efficient match. The key is how to measure the effectiveness of data reflux, in the dynamic, to find new anchor points. Now the traditional enterprise has come to the need to integrate into the Internet time, this time real-time data is your new data. Among the most critical is the ability to restore real-time data, refining, and for enterprises to use. This is a dynamic process in which the data "continuously optimizes the decision-making process-see yourself". "If GE was an industrial company when you went to bed last night, you would wake up today as a software and data analysis company," said General Electric CEO Immelt. As a representative of the traditional industry, GE has figured out, and people say, I already have tens data points, traditional enterprises have what to hesitate? You can also read the following related articles The Big Data play of home electric builddirect; Large satellite data shines on agriculture; Industrial data is the core competitiveness of China to win the new manufacturing revolution; Why do 90% executives not look at the data, intuitively?