Some students asked: "At present, the data team's analysis ability is weak, and they want to improve their analysis ability, which can be combined with the company's business to achieve quick results"...... Today, the system introduces that the data analysis capability is divided into three layers and needs to be improved layer by layer.

Three levels of capability requirements

• primary capability: use SQL,Python and other tools to extract data as required to meet the viewing requirements
• intermediate capabilities: based on business processes, actively sort out indicator systems, develop data Kanban, and replace scattered and temporary data retrieval requirements with fixed data monitoring.
• Advanced Capabilities: based on data monitoring, proactively detect business problems, provide landing suggestions, and design experimental test Business Solutions. Only by doing this can the data be truly "effective"

at present, there are many introductions about primary abilities on the Internet, including:

1. A large number of questions are brushed to improve the proficiency of taking numbers.
2. When recruiting, take the complicated query in the work as the examination questions, and select the ones that can be done correctly.
3. Provides sample tables and query logic, and uses AI tools to assist SQL writing.
4. Establish a "wrong question book for taking numbers" and practice repeatedly on the indicators for counting faults.

In a word, practice makes perfect. If you practice more, you can create a qualified counting tool.

However, starting with intermediate capabilities, data is required to be combined with business and have their own insights. This piece is rarely shared online, and most of them introduce popular practices such as AARRR. Today, I 'd like to introduce it to you in depth.

Point 1: replace scattered numbers with standard templates

there are four common data analysis requirements:

1. Monitor business conditions

2. Analyze the cause of the problem

3. Forecast Business trends

4. Test business ideas

the monitoring class has the largest demand and takes the most time. If you want to improve your analysis ability, you should divide it into business lines (such as sales, operation, products, marketing, logistics, etc.) Sort out the indicator system and generate a fixed monitoring template (as shown in the following figure), which can deepen novice students' understanding of the business, greatly reduce the workload of temporary data retrieval, and release manpower, to study more advanced things

point 2: summarize the trend of business Regular

"Those who do data cannot understand the data" is the pain point of many basic students. After establishing monitoring indicators, we should first understand the regular trend of key indicators of business assessment (such as sales revenue, profits, number of new users, number of active users, etc.)

there are three types of rules to pay special attention to (as shown in the following figure):

1. Natural cycle: whether the indicators are related to seasonal changes and holidays

2. Lifecycle: the trend of the main indicators of the business from online to offline

3. Group changes in the same period: Trend of N time cycles after user registration

only after understanding the trend of the regular can the newcomers understand what "normal trend" means, and then find out the real indicator abnormality (instead of disassembling for half a day with a 1% fluctuation). This is the starting point for in-depth analysis.

Point 3: go deep into the business process

to gain insight based on data, you must first go deep into your business.

For example, sales, you can understand:

1. How many steps are there in the sales process and what number records are there

2. There are several types of sales channels, and how does each type perform

3. There are several kinds of products sold, and what is the proportion of each kind

for example, supply, you can understand

1. How many steps are there from raw materials to production

2. What resources are consumed in each step

3. What results are produced in each step

in the process of understanding the business, note:

1. How much data is collected in each link

2. Which links do the business itself pay most attention

3. What are the common business practices

after understanding, you can establish an indicator system based on business processes and common practices (as shown in the following figure). In this way, data can be monitored without dead corners and problems can be considered from a business perspective (rather than "what indicators are in the database, I can pour out all my brains")

point 4: quantify the effect of business actions

A quantitative monitoring mechanism must be established for key business actions.

For example, the business wants to improve sales performance

1. If the marketing department is doing a promotion activity, you can use data to record which orders are promotion orders, see the growth of promotion orders, and calculate the activity revenue.

2. If the sales department holds a sales ability training meeting, there is probably no data to record how much each person has improved. At this time, we can only take the second place, record who participated in the training/which companies participated in the training, and then see if the indicators have changed.

In doing so, on the one hand, it can deepen the newcomers' understanding of the business; On the other hand, it can first understand the effects of various business actions from the results. The material can only be obtained after in-depth analysis. Some companies don't know the data about what the business has done, and they have to ask afterwards...... This is not a "in-depth analysis"

point 5: split business issues to form analysis assumptions

learning to make assumptions first and then finding evidence is a key step from intermediate to advanced (as shown in the following figure)

the analysis assumes that there are three sources

1. Make assumptions based on past rules, experiences and trends

2. Make assumptions based on the concerns of the business side

3. According to the structure/hierarchical analysis, the problem found is big point hypothesis

these three methods require the accumulation of the previous steps. When there are many possible assumptions, such as a simple "impact of rain on performance", senior analysts can split several situations (as shown in the following figure). This kind of thinking from coarse to fine requires long-term training. Only by breaking through this step can I be promoted to the ranks of Masters.

To sum up, the improvement of analysis ability requires

1) go deep into the business process and understand the business details

2) grasp business rules and measure business practices

3) reasonably put forward assumptions and find out the key elements

moreover, a lot of training is essential.