"What is the use of the data analysis you do?"
"What is the use?"
"What's the use?"
This is a question that is often asked during interviews, and it also makes it difficult for many students. Either I don't know where to start, or I was stunned back after I answered. Today we have a systematic answer.
First of all, anyone who directly answers this question "can help companies make money/save money" will have a high probability of death. Either I was perplexed by the interviewer, or I got stuck in a big hole after I joined the job, and I couldn't even survive the probation period.
Because data is essentially a staff member, not a soldier who really fights on the front line. Just as the sniper scope has a higher multiplier, the sniper is the last to kill, it is a bullet, it is a rifle, but it is not a sniper scope. To win the battle on the front line, what is needed is the approval power of funds, technical personnel, business teams and bosses. The role of data is at the end of the line.
Secondly, a direct answer to "can help the company make money/save money" is often confused with one sentence : Does your analysis report include an implementation plan? If it is not included, you are not doing the landing. Why do you make the money? If it includes, does your plan have resources and human resources, then how do you prove that it is the result of data analysis rather than resource investment?
Of course, most data analysis reports do not have a supporting landing plan at all. This problem has been smashed to death in the first step. I was lucky enough to survive to the second question, but I couldn't explain it. After all, data is data. Data can help business but cannot replace business.
Of course, there are also some students who have turned into passionate chuu elder brothers, boasting: "As a data analyst, my ability is really strong. No need to invest money, no need to do development, just take my analysis, the money It's just a flowery!" Such an exaggerated statement will definitely be directly dismantled by someone who knows it.
But in case it is the same thing that I don’t know how to do, I really believed these words and recruited Chuchu into the company, hoping to analyze the situation. Brother Chuuchi will most likely die during the probation period, because no one will care about him. Who will be okay to hear a newcomer beep. There is no funding, no development, no sales, and a ppt with two lines of code can make a fortune.
In the end, all the awesomeness that blows out will disappear.
Sniper scope, and sniper killing people are two concepts.
Data analysis aids decision making, and data analysts make decisions are two concepts.
Data analysis helps business, and data analysis is business and two concepts.
Remember to remember
So, what is the correct answer?
First of all, the biggest and most direct effect of data analysis must be the production of data.This is the result of the real data analysts themselves. No need for fancy packaging, the data itself is very valuable. Just as you must look at the speed and tachometer when driving, you don’t need a model at all, no thinking, no concept. Would you like to drive without looking at the speedometer? It's that simple.
1 The first answer, examples of words are as follows:
▌Filling the gap: I created 5 new promotional activity reports, so that the business department can keep abreast of the promotion activities; and for the three types of promotional activities frequently carried out by the business department, I newly designed a standard analysis template to enable the business department to Compare the effects of previous activities and choose a more suitable method.
▌The question was answered: the business department did not agree on which product scheme to use. I designed ABtest and verified the data of the two schemes. The data support scheme A performed better. Supported business department decision-making. (Note that data is not the only decision-making criterion in many cases. So even if you have done ABtest, don't say it to death, boasting that decision-making is driven by my test, and will be stunned by people who understand the business)
▌Verified the hypothesis: the business department has always had doubts about the effect of XX channel drainage. Through data analysis, I verified the business department’s hypothesis that the channel is indeed poor in ROI, and after many adjustments, the efficiency still hasn’t improved.
▌Improved efficiency: I improved the efficiency of running counts . I upgraded the data updated in the past 3 days to the next day update, which was praised by the business department and improved the efficiency of decision-making. I designed the report style on the mobile terminal, which increased the usage rate of the report and enabled the business department to get more data support.
These four items are all effects that can be directly achieved through data reports, models, tests, and analysis reports. It is the data itself that is at work, so it can be used as the first pick. Please pay attention to the order of the above four sentences, this order is the order in which the data comes into play.
From 0 to 1, the effect of new data is the biggest. From 0 to 1, the effect is easy to observe. And there is no data at all, and the situation with black eyes is like driving without looking at the speed, which is very dangerous in itself. The work from 0 to 1 is the easiest to reflect achievements and the easiest to be recognized.
When the business is indecisive, provide data standards to determine which is a 70-point pass and which is a 50-point rough, helping the business to achieve 60 points from 1 to the second largest. Because there is no extensive operation supported by data, it is easy to be killed, which is a waste of resources. Data analysis is more reliable than clapping your head.
As for optimizing efficiency, it is quite difficult to increase performance from 60 to 90 points, and the actual effect is limited. Because to make extraordinary achievements, it is often impossible to copy the right time and place. This is the time for the business department to play its genius. No matter how awesome the analysis model is, it can't simulate Jobs' brain, which is roughly what it means.
So to summarize: data itself is value. The value of data is more embodied in the end of the smear situation and provide the correct direction to the business department. Of course, if the business department sees how to do better, it will naturally trigger the second step: make decisions based on data conclusions. This brings us to the second step of the role of data analysis: indirectly contributing to performance.
2 The second-ranked answer, examples of words are as follows:
▌I found business opportunities: By analyzing the attributes of lost users and comprehensively evaluating users, I found user groups with high retention value and low retention difficulty. The business department adopted the opinions and carried out retention activities, which increased the user retention rate by 5% compared to before the hierarchical retention.
▌I found a business problem: By establishing a response rate prediction model for outbound calls and scoring users, I found the user groups with a higher response rate. The business part adopted a list of outbound call ratings, and selected top 50% of the outbound users, and the outbound call response rate was increased from 2% to 8%, thereby saving outbound manpower/cost waste and improving outbound call efficiency.
It should be noted that the above-mentioned language skills are likely to arouse the other person's curiosity and raise more questions, such as "how did you make it", "talk about experience in detail" and so on. So if you really have experience, it is far from enough to simply explain how the data is done. At least the following three points are needed:
▌First , we must clearly explain the basis of analysis
Most projects do not start from zero. Unless you are involved in a brand new project, there is no business design yet. Otherwise, business departments have some conventional practices. For example, in the above retention retention, it is possible that the business department will wake up when the user does not log in for 7 days/30 days/60 days, without distinction. The role of data is not drawn out of thin air. It is precisely because of the previous data basis that it is possible to judge good or bad through analysis and form experience. These foundations are indispensable. Many interviews made a lot of talk, and the messy data analysts who entered the job did not pay attention to the basics and ended up in the mud of the new environment.
▌Second , it is necessary to explain the combination of analysis methods clearly
Most of the projects are not a single model to the end. Let's take the example of retention retention above. At least five steps are required: divide retention groups, analyze user preferences, compare retention programs, predict retention success rate, and monitor retention effects. At least five analysis topics can be separated here. In actual work, participating in the entire business process as much as possible and building an analysis system is more effective than expecting a solitary model.
▌Third , we must make clear the business cooperation conditions
Most projects are not covered by data. For example, the above outbound call response model, the level of the outbound call itself, the selection of outbound products, whether to cooperate with the promotion policy, the outbound time period, the outbound time node and other factors will all have an impact. All of these must rely on the support and cooperation of business departments to achieve the desired state, and data alone cannot be calculated. Good data analysis results must be obtained from the cooperation between data and business.
With this three-point blessing, how data can help businesses and ultimately generate benefits is a perfect explanation. This method of explanation would seem very intimidating, and some students may not understand, and say: "Is it necessary to distinguish so clearly? I will say that my analysis produces performance, so what!". In addition to avoiding the problem mentioned at the beginning, there is a deeper reason for such a careful distinction:
When people don’t understand something, they will show two extreme attitudes. They will dismiss it when there is no problem, and expect it to continue their lives when something goes wrong. This attitude will only degenerate science into metaphysics and degenerate scientific work into counterfeit drug dealers selling health products. Putting aside the system of data analysis and operation, to trumpet that relying on a magical algorithm model, a powerful underlying thinking can bring back to life, cure diseases and save people. The essence is to harm oneself, and it is the continuous and healthy development of the circle of data.
Data analysis is not the elixir of being too arrogant, it is not the magic medicine for renewing life. Data analysis wants to help business, it has always been built on the basis of systematic operation.
1. Collect data from 0 to 1, and establish a data monitoring system;
2. From 1 to 60 precipitation experience, screening methods, and accumulating characteristics;
3. Establish a fixed analysis model from 60 to 90 points to continuously improve business efficiency.
This is the path for data analysis to help business. This process seems scary and hard, but it is indeed a sustainable development path.
This article is very long again, and the students who persisted here will have benefits again. The reason why most students can't answer "What's the use of the data analysis you do?" is because they are at the beginning-they don't know what the business uses for data. There are even a lot of people who have worked for 2 years, and they don't even know how many business departments there are or what they do. So mastering the work of the business department is a good breakthrough.
In order to train newcomers to explain "what is the use of data analysis", I have briefly sorted out what the business departments of each enterprise do, so that newcomers understand the responsibilities of each department first, and then when meeting customers, they must understand clearly that the customer is Which departments have demand. Then you can follow the 0,1,60,90 method to introduce.
For example, if a customer you want to meet today has a demand for sales BI, you can first ask: Have you established a performance monitoring system? If not, let's talk about the significance of 0 to 1, and talk about the various chaos of companies that do not do data management.
If there is already monitoring, then talk about the various meanings from 1 to 60, how to find channel problems through data results, and how to find opportunities for new channels.
If you even have this, you will have a variety of cool operations from 60 to 90, such as how to develop mobile reports for channels, how to make automatic follow-up reminders, how to create a small data secretary, and so on. With this sorting, newcomers are also able to communicate clearly with customers, which is much easier to use than what they said before.
Students can do this by themselves. Look at the picture above, figure out which departments they serve, and then ask one by one:
1. Do they have data reports for their work? Do they need to create new ones?
2. Is their work effective? Do you need to develop a scientific standard? Do you need a verification standard?
3. There are good and bad evaluations of their work, so why is good? Why is it bad? Have you sorted out the reasons?
By analogy, not only can we sort out the usefulness of the data from the seemingly ordinary work, but it will also be clear what we can discuss with them in the next step. Of course, it is taken as an example of an entity company with an e-commerce department. You can draw inferences and add more. welcome.
There is also a bigger problem, that is, often you do a lot of hard work and get dumped:
° "This is very common!"
° "The business department has known for a long time!"
° "I know this without analyzing it!"
° "Can you make an analysis that the business department doesn't know at all, yet has significant meaning!"
Let’s share this question in the next lecture. Welcome to the bottom right corner to watch and support Mr. Chen, continue to follow the drama!
Original selection:
The most useful report in the world, novice data analysts have not noticed it!
The ultimate question of data analysis: how to calculate the natural growth rate to be reasonable!
Looking at the data, interpreting the data, and discovering the role of the data is an indispensable ability for people in the workplace. Mr. Chen explained in detail in the "Business Knowledge One-stop" video course. In the course of study, you can also join a group of students to discuss issues with Teacher Chen one-on-one to improve your ability.
Click " Read Original " in the lower left corner to listen to Teacher Chen's lecture!
I’m Chen Xiaoxiao, scan the QR code to add the assistant WeChat~