The strongest strategy in history! Hand in hand to teach you how to build a “data desk”!


The article was transferred from Chief Digital Officer.


Author: Li Guohuan

On the evening of April 24, the first lecture of the online micro-lesson series “Hard Nucleus Exchange of Data Station”, which was co-created by the experts of Shulan Science and Technology and Jinkuang, was launched. 1000 CXOs are listening at the same time. The group members ask 30 + questions. Hi, everyone! This article is for this micro-class sharing content collation, missed 24 evening micro-class students do not panic, dry goods articles are ready, come and taste it!

“The DT era is really coming! Today, the whole industry mode has changed, from Internet + intelligence to intelligence, to the role of big data and artificial intelligence in various application scenarios, all of which have injected new vitality into the development of enterprises. “- Jiang Min, Co-founder and CTO of Shulan Science and Technology

First, why do enterprises need to build “data station”?

The concept of Taiwan was first put forward by Alibaba Group. In 2015, Alibaba Group launched the strategy of Taiwan. The goal is to build an innovative and flexible mechanism of “big platform, small front desk”, which is in line with the era of big data on the Internet. That is to say, the front-line business as the front desk will be more agile and faster to adapt to the ever-changing market. Zhongtai will gather the operation data capability and product technology capability of the whole group to form a strong support for the front-end business. The so-called “Zhongtai”, that is, the enterprise Internet architecture, will build public resources together and share services.

Over the past two years, people’s perception of China and Taiwan has become more and more obvious, which is closely related to the pattern changes of the whole industry. At the earliest time, our traditional to B mode was only background + front desk, and the front desk was mostly in the background of a single business scenario. This ability was mostly for internal management. The left traditional to B mode in the following figure includes CRM, ERP and other corresponding systems, mainly for internal management.

With the development of Internet in recent two years, including the development of mobile Internet, the business model of enterprises has changed greatly. In the past, a manufacturing enterprise may not have the final business model, but in recent years, through the mode of Wechat, mobile Internet, small program, etc., it has also begun to carry out to C business model.

There are many situations that can’t match the management of the earlier period quickly, such as CRM which was recorded after sales visits; today, when we reach customers, it may be through an online system, such as marketing system, that we can reach the end customers directly, and there will be a problem among them. That is, data on both sides cannot be connected in series.

At this time, how do we solve this problem, in order to get through these data, so when building the data desk, enterprises need to better aggregate the original data of various formats, or data under various business models, and then fuse them based on these data, and then data will be processed. The corresponding value mining, service application, and ultimately the service feedback to the existing business system.

What is the relationship? In the digital transformation of enterprises, there are two concepts, one is business platform, the other is data platform. Firstly, the business platform solves the problem of business data. Enterprises are formed by business data, which is a mineral resource that can be excavated in the future. These minerals are the source of raw materials for the entire data platform. Although its value may not be perceived at present, in the future, with the application of new technologies, including scene mining, these data can bring tremendous value-added for enterprises.

Because the business platform often solves the OLTP scenario, the response to business, including transaction, processing relevance and so on, is more a process of production data; and the positioning of data station is to mine the value of these data, data station is not a process of production data, but a process of production data. A process of using data, so it is defined as “data business”.

From the point of view of data desk, we refine the value of data minerals deposited in the business system through new technologies, such as big data modeling, machine learning, and so on, so as to form the final enterprise data assets. Then, through the service ability, we can empower the existing business system, such as promoting commodities and personality. Recommendation, etc. The results are recorded by the business system and then fed back to the data platform, which then models and analyses the data based on these results, and gradually optimizes the business strategy.

Data stations are more oriented to use scenarios, maximizing the value of precipitated data, serving existing businesses, and even expanding new business forms, bringing new value-added to enterprises.Data desk and business desk are mutually complementary and mutually assistant.

2. Data application process mechanism to build data application pipeline for enterprises

Enterprises build data desks, in which the key is whether the enterprise can quickly build up the process mechanism of using data. It is impossible to connect in series if we only buy tools to fill a capacity gap without considering the process mechanism, which is the process of data value mining.

Once we construct a set of data application process mechanism that matches the enterprise, it is like buying a data application pipeline for the enterprise, which is the closed-loop of data value mining. Data will be input according to the required process, which can be verified, such as: which data is valuable in which scenarios, which data is not valuable in which scenarios, this process is a very core part of the data desk.

Rejecting “three pats” and shaping enterprise data culture, business departments often fail to recognize the value ability of data while promoting business, and thus ignore the existence of data value. This value does not necessarily directly generate income, but reflects in making the whole thing more efficient, more accurate decision-making, more personalized and accurate service, these are the value of data to enterprises.

To build enterprise data culture, we sometimes joke about making decisions. Without data support, it is likely to be a three-pat pattern. One pat on the head, this is what I want to do; two pats on the chest, this is something I must do well; three pats on the butt, it is not good to pat on the buttock and walk away. However, data capabilities will be improved to avoid this situation, so that enterprises refine their operations. Support the whole management process and service process through data, and provide corresponding technical support more pertinently.

There are two main cores of data station: first, technology product system. Data desk is not a product, each enterprise’s data desk should be different from other enterprises, because the business of the enterprise has its own characteristics. The second core is the construction of data assets system. This part is equivalent to a person’s cognition and soul. If there is no such thing, it is a vegetable. Therefore, it is not feasible for an enterprise to build a data platform only with a tool platform, in which data assets are a very important part.

In addition to data assets, there is also a part of our service system, which can reduce the trial and error cost of enterprise data value. Therefore, when building the asset service system, it is necessary to consider the combination of enterprise scenario precipitation and common performance. Business departments can quickly assemble the desired capabilities through data plus services, and then they can self-help complete the whole data value mining process. This is an external output capability that data stations need to provide.

So how to operate data as a business can be summed up in four words:Research, determination, operation and excellence

Firstly, we need to study and analyze the problem by means of data; secondly, we need corresponding data support when making strategy or making corresponding decision. Thirdly, data are used to support the action. Fourthly, to optimize the final results, enterprises need to record the results in the form of data, and then provide corresponding optimization capabilities through the way of data.

Three Core Values of Data Platform

The core value of data stations has three key points. The first is innovation. How can the data desk help enterprises dig out the value of data, so that business personnel’s business knowledge can be precipitated? Why does Ali operate better? Our own understanding is more about their operations, that is, to do the whole business in a data-driven way.

The second is scene-driven mode. Demand-driven may cause problems, first of all, the response must be lagging behind; on the other hand, the response process depends on the entire infrastructure capacity-building, many times after a demand is put forward, it will take two or three months to support, and the time will be longer.

So we need to consider scenario-driven. That is, to digitize the whole process of things, when new problems arise, it may be based on the accumulated knowledge and ability experience, which can quickly provide solutions to support the problem solution.

The third is experience precipitation. In building the whole data platform, we need to transform experience into a part of the data assets system to guide enterprises to use data in the future, stand on the shoulders of giants and constantly optimize and iterate, so that data assets become more and more valuable.

However, it is difficult to find the business value of data for technicians, technicians may pay more attention to the technical level in many cases, so the understanding of whether data can serve a business scenario is not profound, because the understanding of business in enterprises must be first-line business personnel, so our data and business. There is a gap between them. How to build the bridge on the ditch is actually a problem that we need to consider when building the data platform.

Notably, the bridge has a feature to capture: how to turn data into a readable and understandable asset.

From a business perspective, we need to translate it into a form that business people can read and understand, and then give it to them. So once we convert data into business perspective, it will be easier to understand, but business will have many ideas and innovations, such as demand-driven, scenario-driven, so once these data settle down, it can provide good support for future business decisions.

Experience precipitation, first of all, asset system will be divided into three dimensions: people, things and scenarios. Then, it will provide some basic data service capabilities, such as query, analysis, recommendation, wind control, etc. Of course, service capabilities may have different choices according to the situation of the enterprise, but once these capabilities have settled down, the industry will be able to provide some basic data service capabilities, such as query, analysis, recommendation, wind control, etc. When doing precise marketing, staff can know what kind of recommendation methods they need to use today, which dimension indicators they need to consider when recommending, and which features they need to train the promotion algorithm. Then they can try. Once successful, they will see whether the results are good or bad, and then continue to optimize.

IV. Evolution Path of Data Station

The evolution path of data stations can be divided into four stages.

Statistical analysis. The first stage is mainly based on statistical analysis, business needs oriented, and then add a small number of statistical analysis methods to do. Common forms such as business system plus statistical report module, directly do statistical work on business system data.

Decision support. The second stage begins to make decision support. At this time, some data concepts begin to build the basic data warehouse to support the whole data decision-making process through the method of data warehouse.

Data services. The third stage is the data service stage, in which the business needs start to be driven by data. At this time, data and business begin to integrate gradually. But in the process of integration, there may be more business needs, and few common problems will be considered. So in this process, in order to support business well and bring value to business, there will be a lot of repetitive construction.

Data-driven. The fourth stage is to look at the problem from a global perspective. Phase 3.0 starts from the requirement scenario; Phase 4.0 is typically driven by business scenarios, that is, building infrastructure capabilities, building business requirements, and which scenarios applications can be quickly assembled through existing service capabilities under data-driven, and applied to specific businesses to form a whole. Closed-loop data application.

For the evaluation of data maturity, in fact, every enterprise can do some corresponding evaluation according to the current stage. As for the whole data platform to promote landing, we summarized 16 words, that is, to promote, to pass-band storage, to save training, to calculate and use. How to understand these 16 words? In many cases, the investment of enterprises in data platform is relatively large, and the risk is also high.

Use to promote. First of all, we have a clear understanding of the data station, what the data station looks like; how the whole link closed loop is; so when the enterprise first landed, it is necessary to find an application point which can best reflect the value of the enterprise. With the application point as the entry point, its data capacity building will eventually be generated from the entry point. And see business value, data value.

Passband deposit. In the application scenario, the requirement for storage is very high after data pull-through. In this process, storage considerations are needed, such as how to support the development of future business, whether storage capacity can be combined with in-depth learning knowledge and learning framework, etc.

With data base, including computing power and data base of algorithm, the feasibility of algorithm and intelligent application scenario can be established. So we need to train AI capability with stored content, or algorithmic model capability, extract the asset content of these data, and then return to the application scenario.

So the key is how to avoid being affected by the business on one hand, and then to build a data station which only supports a single business. Similarly, we need to consider more when we build the whole platform, in order to find a specific scenario to cut in, and then really highlight the value before expanding.

(Small partners are welcome to join the Shulan community to learn big data ~)