- What use is it for your business!
- What you did is not what we want!
- Can you think about it a little deeper?
Three consecutive vomiting blood...
Can't the business make it clear by itself, what exactly it wants? ——It's actually really difficult. Because abstracting a specific problem into a problem that can be calculated, tested, and predicted with data is definitely a professional ability. People who don't have this professional ability will just pack everything they don't know well into: "You come to analyze and analyze" and throw it out.
Unfortunately, many newcomers who do data do not have this ability themselves, so they can run just a few times, and they will get confused when they encounter the word "analysis". Today we will explain the system.
1. Four questions about the soul of sorting out needs
The first question of soul torture: What is the difference between the following three sentences?
A. What was the sales amount yesterday?
B. What was the sales situation yesterday?
C. Is yesterday's sales performance okay?
Think for a minute
answer:
A asks about the value of a specific indicator.
B asks about “situation”, which may correspond to many indicators.
C asks for a conclusion “OK or not”, which requires index + standard
This is the first key to sorting out data analysis requirements: clear problem types.
As long as what you hear is not a specific indicator, you have to play up your spirit. Because people without professional training don't know what they want to ask. One sentence of "analysis" is over. It is common to analyze problems in confusion. There are five basic question types for data analysis. When sorting out requirements, you must combine these five types to clearly distinguish what the business wants (as shown in the figure below).
The second question of soul torture: What is the difference between the following three sentences.
A. What was the sales amount yesterday?
B. What is the sales amount today?
C. What is the sales amount tomorrow?
Think for a minute
answer:
A is something that has already happened, just take the number according to the conditions.
B is what is happening. There must be supporting monitoring and real-time updated data.
C is something that will happen in the future, and there must be forecasting methods and forecasting assumptions.
This is the second key to sorting out data analysis requirements: clear time status.
The time status of the data directly determines the work content, work difficulty, and work style. For real-time monitoring, the monitoring requirements, data update frequency, and monitoring indicators need to be mentioned in advance. For forecasting, forecasting purpose, and forecasting accuracy, you have to talk clearly in advance (as shown below)
Soul torture third question: What is the difference between the following four sentences.
A. Commodity operation asks, for the two previous activities, which one is better, half price or buy one get one free?
B. Commodity operation asks, we are planning a promotion, which is better, half discount or buy one get one free?
C. Commodity operation asks, we have a half price and buy one get one free plan, which is better?
D. Commodity operation asks, we just did buy one get one free, would it be better if it was changed to half price?
answer:
A asks about what has happened, and specific things can be directly evaluated according to the conditions at the time.
B asks only about intentions, which require multiple rounds of analysis to in-depth.
C asks about the comparison of two specific schemes, and can directly test
D. Possibility, can only be deduced according to certain logic
This is the third key to data analysis requirements: clear business actions.
Note that this question is more likely to appear in the form of "half price or buy one get one free?" They probably won't elaborate on what the state of the matter is. The advantage of data analysis is to evaluate the problems that have occurred. Problems that have not occurred can be tested if they are specific; if they are not specific, they can only be deduced logically. In the end, they have to be implemented in specific plans for further analysis. Therefore, it is very important to clarify the state, and the corresponding methods for different states are also different (as shown below):
The fourth question of soul torture: What is the difference between the following four sentences.
A. What was the sales amount yesterday?
B. Yesterday’s sales amount was 30 million?
C. Why was the sales amount yesterday only 30 million!
D. What happened to the sales situation yesterday?
Think for a minute
answer:
A is asking about the value of a specific index.
B is questioning the accuracy of the value.
C is asking about the reason behind the value.
D is asking knowingly, there is a pit warning ahead
This is the fourth key to sorting out data analysis requirements: probing the business purpose.
Except for option A, business parties who will ask BCD questions are obviously coming with a purpose. At this time, you can't just give the value stupidly, but have to prescribe the right medicine. Especially when there are already fixed reports and fixed assessment standard output, the business side still comes to ask about the situation. It is very likely that there is a deeper purpose hidden. At this time, you can sort out the ideas below.
This is the basic method of sorting out needs, get it?
Here is a comprehensive test
2. Failure to carefully sort out the tragic situation of demand
Comprehensive test: The operation stated that we need to resume user portraits, segment the population, and achieve push according to the needs of the population.
Question 1: Is this requirement clear?
Question 2: What do you want to do when you get this requirement?
In fact, quite a few newcomers shouted when they heard similar needs: "It's a big job!"
Then began four tragedies:
In this demand
First, there are no specific indicators. It is completely unclear whether the existing data can meet the needs and what needs to meet.
Second, there is no time state. Does grouping based on past behavior have any guiding significance? Still have to group based on predictions, to what extent? Not at all clear.
Third, if there is no business plan, has there been grouping and what is the effect of grouping? Is there a push currently, and what is the effect of the push? Not at all clear.
Fourth, there is no business purpose. "Push on demand" is just an action. What results do you want to achieve? What index to improve? What is the current indicator? Not at all clear.
Therefore, when encountering similar problems, what needs to be done is in-depth communication and complete a long combing list. In addition to regular understanding of project goals, time requirements, output products, these, the most critical point is the third point: business plan!
1. What are the current dissatisfaction and dissatisfaction of group leaders?
2. Is the current push 70% success rate or 5% success rate?
3. Existing data can only meet 30% of the demand. Is there a plan to improve the collection?
4. Does the prediction accuracy of 90%, 80%, 50%, and 30% affect business development?
These are the foundations of the new project. The quality of the foundation directly determines the construction method. If the foundation is too bad, you have to start from the simplest place; if the foundation is too good, you have to think carefully: can it really increase from 90% to 99.99% with such a great effort? ? ? Such reasonable planning can guarantee everyone's satisfaction when landing. If you are interested, follow the WeChat public account [Ground Qi School], we will share in the next article, how to design a data-driven solution based on different scenarios, respectfully.
Author: Chen grounded gas, micro-channel public number: down to earth school. A data analyst with ten years of experience has launched a series of data analysis courses and has more than 20,000 students.