There is only one scenario data analysis that may be useful for income, that is, a business unit +B42 is really badly done, and something is wrong. At this time, if some benefits can be improved through analysis, they will not be happy. This is why many mature data mining projects are aimed at customer service outbound, SMS sending and EDM. Because in these places, the natural conversion rate is horribly low, and the copywriting, products and advertisements of business departments do not play much role. At the same time, these channels are point-to-point push, and the data accumulation and modeling environment is relatively closed. The data model can improve the natural conversion rate from 1% to 2%, and the business department has been thankful.
In fact, data analysis is helpful to enterprises, which is more reflected in posts, such as performance evaluation, result assessment and result optimization. Interestingly, many practitioners themselves don't want to understand this. For example, Fan Ruan also has the answer to this question. You can have a look. The examples given in it are all about how to cut costs instead of increasing income.
However, Fan Ruan's answer itself is very professional. Because cutting costs is easier to reflect the credit of data analysis than increasing income. Let's review the process of increasing revenue by the above new products. If the data analysis says that I did this performance, at least six departments will take credit with you. But if the data analysis says that there is a product here that is rubbish and can be cut down, then at most one department (the department that designed this product) will be offended, and the remaining five departments will still support you (because there is no need to waste time). Therefore, intelligent data analysis always proves value from the perspective of internal control, not from the perspective of external income increase.
However, this leads to the second embarrassing place, that is, should I do this for the last data product of wool? Even wool, I have to hire a data analyst to do this? Because the data of invoicing is also in ERP, theoretically I want to know which product has poor efficiency as long as a programmer who knows SQL runs from ERP! Therefore, if the value of data analysis is only linked to internal control, then the importance and professionalism of data analysis will be very low. The bosses of all departments will analyze it themselves. Do you know anything about sql? What do you care if you don't understand business?
At this time, further packaging is needed to reflect the value of data analysis. The core is the final product! Just like the concubines in the harem, they will please the emperor for a while when they are young and beautiful, but in the long run, they still have to have a child. Having a child will ensure your position. For example, sales can use paper bills, why use pos system? That is, the pos system is online and the business process is running, so he has no reason to stop. The child has been born and still has to be raised.
There are several kinds of children who are familiar with data analysis: management-oriented dashboards and boss-oriented data products suitable for scientific management theory. It may be a recommendation system, an accurate marketing model, a business assistant or a data mart. In short, it is a link that must be used in the daily work of business departments. Packaging, packaging with data, packaging into product-oriented marketing reminder tools and operational data guides. Let the sales staff have a look every day, it will be uncomfortable not to look. Let the operators have to look at the heat ranking before writing the copy, not at the bottom of their hearts. I won't go into details. How to attract the attention of the boss, how to win over the business department, and how to let the front line use it, it is enough to write a book. After consulting for so many years and contacting a large number of Party A and Party B, all intelligent data people finally embarked on the road of doing internal control → attracting management's attention → launching products → cooperating with business departments → expanding organizational structure. And those who claim that the big data system can make a profit of XXX in the end are basically bad.
In the past two years, the concepts of big data and artificial intelligence have been on fire. The position of data analysis has been favored by the bosses of major enterprises like young and beautiful concubines, and countless students have poured into this field. Therefore, I sincerely remind you that we can have many methods and complicated concepts ourselves, but whether the enterprise finally makes money from us is the capital for our long-term settlement. If you only play an auxiliary role, export a product around a specific business scene as soon as possible and combine it closely with the business, so that your position will be stable. Finally, for example, we should pay attention to the distinction between algorithms, because algorithms can be applied to both production systems (such as photo recognition, material distribution, route planning and process control) and analysis systems (such as recommendation, prediction and BI). If applied to the production system, their position is relatively stable, because the production line will not be completely replaced, but will be continuously optimized. However, if it is applied to the analysis system, there will be too much water. We should carefully look at what this algorithm is for before making a decision. As early as 20 13 "big data era" became popular, there was a wave of "big data analysis". As a result, I shouted to my boss at that time: "We can use big data XXXX to analyze and improve our performance." Now it is estimated that the grave grass is as tall as my baby ... As a senior, I have an obligation to tell you the truth of this industry. The value of data can be varied, which does not necessarily directly increase income. The data is really useful, but it doesn't mean that the bosses recognize this use, nor does it mean that we can get a promotion and a raise from here. In addition to technology, how to create value may require the assistance of something other than code and algorithm. * * * with everyone.