Editor: Peng Wenhua
Source: Big Data Architect
Hello Peng friends, I am Lao Peng. Today, a friend of Peng asked me: There are many functions of data governance in the data center, can it directly replace the matter of data governance?
Alas... Some data governance concepts in the market are very small, how small is it? The data governance they understand starts after the data is put into the warehouse.
The question asked by Peng You earlier is actually an obvious representative of this phenomenon.
You can see what I'm trying to say: data governance is not just the job of the data department! ! ! Data Governance is not just for data departments! ! ! Data Governance is not just for data departments! ! !
end to end
I first heard that "end-to-end" is a technical term seen in the supply chain, and it is now being used for reference in the fields of informatization and digitization. It is said that supply chain management should be controlled from one end of the entire supply chain to the other.
Taking the vegetable supply chain as an example, one end is the farmer, which is the vegetable production end, and the other end is the consumer, which is the vegetable consumption end.
Well, let's use the simplest thinking to understand, where should we start with the quality control of vegetables? From entering the wholesale market?
Obviously, we have to start from the farmer and control the whole process, okay? After entering the market, we will start to control it. When problems are discovered, it is likely that a lot of vegetables have already been transferred to consumers' tables.
data production end
In fact, many data governance projects fail to go deep into the "data production end", not because the data governance service providers do not understand, but because of many factors.
such as time and money costs.
Once it goes deep into the data production end, it means that the cost of data governance will increase exponentially, and the results will not be seen in a short time.
To give a simple example, no matter where we are, if we find a piece of data with quality problems, how can we solve it?
Generally speaking, data quality can generally be divided into several situations:
1. If the data is not standardized, it can be dealt with directly after standardization. For example, if the gender code is not uniform, just make a mapping table to unify the standard;
2. Relatively regular, for example, if there are spaces in the name or special characters other than ·, just sort out a few rules and deal with it;
3. The key information is missing or wrong, such as the name is empty, and the ID number does not comply with the rules at all.
Among them, cases 1 and 2 can be resolved by data engineers after confirming the rules with the business side. But data engineers can't do anything about situation 3.
No matter how powerful the big data technology is, you can't guess what the other party's name is or what the ID number is, right? That's the business of fortune tellers.
At this time, it is necessary to connect the data governance platform with the business system, classify the problematic data from various channels, and return it to the business system.
After receiving the information in the business system, let the personnel of the business system start another process to complete it by communicating with customers and consulting other materials. One thing that must be done here is to confirm the ownership of each piece of data.
If you have read "Huawei Data Way", you should be impressed by their data owner. Going deeper, they believe that business is behavior, behavior is records, and records are data.
Whoever produces this data is fully responsible for this data. Therefore, every piece of Huawei data has a corresponding business department to undertake management responsibilities.
Friends of Peng who have been following Lao Peng for a long time should have seen this post: " This is a "data governance" call ", which said that the Agricultural Bank of China called me and asked me to complete the information.
This is a typical case of data governance at the data production end. Originally, the garbage that came down from the upstream, why should it be fished downstream?
data application
"Huawei's Data Way" borrows the term of the supply chain and calls the other end of the end-to-end "data consumption". Let's call it the data application side here, which is the same as the "data application layer" we wrote PPT to customers before. If we don't always come up with new terms, customers will be very annoyed, and it will be exhausting for us to explain.
In the past few years, the most typical application is the large screen, and now it has begun to go deeper into the business side. This is in line with the general law of management: first serve the high-level, and then penetrate downward layer by layer.
Therefore, the easiest way for Lao Peng to judge the degree of digitalization of a company is to see who their digitalization serves.
If it only serves the decision-making level, then no matter how advanced the technology used and how huge the investment is, the digital penetration rate is still not enough, and it will penetrate the management level.
If you can deeply integrate the actual business of each department and help each department optimize its process, then the penetration rate is already relatively high, and it can be called a data-driven enterprise.
If digitalization has been embedded in the daily process of the executive level, even to the point where it is impossible to work without digital means, it is like "sitting, standing, and walking are all kung fu" in martial arts novels, and it has reached the "walking, sitting, sitting, and lying down" in Buddhism. The point where everything is Zen” can be called a complete digital body.
Data governance needs to aim at data application, and reversely requires that all data must be "accurate" and "timely". This is the core goal of data governance.
Therefore, understanding how data is used is the first problem to be solved in data governance. So how did Huawei solve this problem?
It's very simple, their data application is not thought out by the data department, but is completely thought up by the business department according to its own business.
Therefore, the data owner of the business department has a lot of authority, and the burden is also very heavy. The positioning of the data department is very clear. There is no need to coordinate business departments everywhere to repair data, and there is no need to work hard to help business departments want to apply, as long as the intermediate data processing work is done.
the case
Friends of Lao Peng probably still have an impression. One year ago, Lao Peng shared a case " How to control the quality of CRM data?" Come, share the experience of the world's top 500 with you! ".
Here is how the data quality is managed and controlled from end to end, all dismantled and understood. Peng friends who are interested can move over to have a look. I won't go into details here. The situation is the same for other companies.
Friends of Peng who are free can estimate how much time, manpower and material resources it will take to achieve end-to-end data governance like the example given by Lao Peng.
Anyway, Lao Peng hangs out with IT people every day, raises requirements and changes the system every day, hangs out with the training department and customer service department every day, and never stops. Do the math yourself.
More exciting:
How to control the quality of CRM data? Share the experience of the world's top 500 with you!
How to do a good job in big data security access control?
[66-page PPT] Ministerial and group-level data governance project experience sharing
Kuaishou data quality assurance system and its practice in live broadcast scenarios
How to create a data governance closed loop? Take the financial industry as an example
Digital transformation requires redefining the role of data governance
Typesetting | Lao Peng
Reviewer | Lao Peng Editor-in-Chief | Lao Peng