Eight processes and seven common ideas of data analysis

8 processes and 7 common ideas of data analysis

In the process of product operation, data analysis has extremely important strategic significance and is the core brain of product optimization and product decision-making. Therefore, data analysis is one of the most important links in product operation.

so how to do a good job of data analysis of payment? The following summarizes the eight-step process of data analysis and seven common analysis ideas. Before starting data analysis, novices should confirm each step of the analysis process with the supervisor or children's shoes with rich data experience.

I. Eight processes of data analysis:

Why analysis?

first of all, you need to know why? Find out the purpose of this data analysis. For example, why do you want to do this analysis in the data analysis of SMS? All your analysis revolves around this why to answer. Avoid repeated rework that does not meet the target. This process will be very painful.

who is the analysis target?

who is the analysis target? Keep in mind the clear analysis factor, whether the statistical dimension is order, user, amount or user behavior. Avoid counting orders as users, and count users as orders (last week's real case of operating classmates), and the calculated results are very different.

what effect do you want to achieve?

Find the real problem by analyzing users and orders in all dimensions. For example, the analysis of the XX channel this time, completely offline, or maintaining the status quo, does not conform to the principle of maximizing interests. Through analysis, we can find out the real root of the problem and find that it is very necessary for users to operate finely.

what data is needed?

the amount of data paid is vast, and there are a lot of data, so it is no exaggeration to describe it as "sea". What source data are needed? Total amount paid, number of people paid? New and old user dimensions? How many times do you pay? Transfer number? Retention rate? User characteristics? Portrait Organize your thoughts first and make a list. Avoid data department students running a data today and another data tomorrow, and data department students will be more annoyed.

how to collect?

direct database retrieval? Or give it to Cheng Xuyuan for export? Write SQL yourself? Operation students might as well learn SQL and be self-reliant.

how to organize?

sorting out data is a technical job. I have to admit that EXCEL is a powerful tool, and the skillful use and skills of pivot table are essential for payment data analysis, and various functions and formulas need to be understood a little to avoid inefficient data sorting. Spss is also an excellent data processing tool, especially when the data volume is relatively large, and when the field consists of special characters, it is easier to use.

how to analyze?

after sorting, how to comprehensively analyze and correlate the data? This is a test of logical thinking and reasoning ability. At the same time, in the process of analysis and reasoning, you need to know the products like the back of your hand, the users well and the channels well. What seems to be a simple data analysis is actually the embodiment of various capabilities. The first is the technical level, the understanding and understanding of the principle of data source extraction-conversion-loading; In fact, it is a global view, and has a clear understanding of the business at the seasonal and company levels; Finally, it is professionalism, and I know the business process and design like the back of my hand. Practicing the power of data analysis is not an overnight achievement, but a continuous growth and sublimation in practice. A good data analysis should be value-oriented, look at the overall situation, base on business, and use data to drive growth. It is easier for operation students to get together at a certain point and walk in circles.

how to present and output?

data visualization is also a science. How to show it with a suitable chart? What is the moral of each chart? Here are eight commonly used charts:

(1) Line chart: suitable for continuous data that changes with time, such as income change with time and growth rate change.

(2) Column chart: mainly used to show the differences between various groups of data. There are mainly two-dimensional column charts, three-dimensional column charts, cylinder charts, cone charts and pyramid charts. Such as the difference in coverage between Alipay and WeChat.

(3) Stacked column chart: Stacked column chart can display not only the size of each data in the same category, but also the total amount. For example, when we need to express the number of people and the total number of people in each payment method.

(4) Line-column chart: This type of chart can not only show the comparison of the same category, but also show the trend.

(5) Bar chart: It is similar to a horizontal bar chart and has the same display effect as a bar chart. It is mainly used for comparison of various categories.

(6) Pie chart: it mainly shows the proportion of each item. Pie charts are generally used with caution, unless the proportion difference is very obvious. Because the naked eye is not intuitive to distinguish the proportion of pie charts. Moreover, the number of items in a pie chart should generally not exceed 6. After item 6, it is suggested that the column chart is more intuitive.

(7), composite pie chart: generally, it is the next analysis of a certain proportion.

(8) Pie chart of mother and child: it can directly analyze the composition and proportion of the project. For example, among the users who had the ability to pay by SMS last time, there were X% users who didn't have the ability to pay by a third party, and X% users who didn't have a bank card and a WeChat payment account.

charts don't need to be too flashy, just say one question in a table. Using friendly visual charts saves readers' time and is also a respect for readers.

There are some data, which have been painstakingly sorted out and analyzed. Finally, it is found that it has nothing to do with the conclusion output. Although a lot of work has been done, data cannot be piled up to reflect the workload.

during the presentation, please indicate the data source, time, description of indicators and algorithm of formulas, which not only reflects the professionalism of data analysis, but also respects the report readers.

Second, seven ideas for data analysis:

Simple trends

Learn about product usage through real-time access to trends. Such as total flow, total users, total success rate and total conversion rate.

multidimensional decomposition

according to the analysis needs, the indicators are decomposed from multiple dimensions. For example, new and old users, payment methods, game dimensions, product version dimensions, promotion channels, sources, regions, equipment brands and so on.

transformation funnel

according to the known transformation path, the transformation situation of the whole and each step is analyzed with the help of funnel model. Common conversion scenarios include order rate, successful conversion rate and so on.

user grouping

In the refined analysis, it is often necessary to analyze and compare the user groups with a certain behavior; Data analysis needs to take multi-dimension and multi-index as grouping conditions, optimize products and improve user experience. For example, this time, for users such as SMS, if there are third parties and no third party payment ability in SMS, we need to operate in groups again.

Close inspection path

Data analysis can observe the user's behavior track and explore the interaction process between users and products; Then find problems, inspire or test hypotheses. For example, our operation of new users this time is also very interesting.

retention analysis

retention analysis is to explore the relationship between user behavior and return visit. Generally speaking, the retention rate we talk about refers to the proportion of "new users" who "return visits" within a period of time. Find the growth point of the product by analyzing the retention differences of different user groups and users who have used different functions.

A/B test

A/B test is to test several schemes in parallel at the same time, but only one variable of each scheme is different; Then choose the best scheme by some rules (such as user experience, data indicators, etc.). Data analysis needs to select reasonable grouping samples, monitoring data indicators, post-event data analysis and evaluation of different schemes in this process.

not only the data analysis of payment, but also other product operation data analysis processes and ideas are also applicable, but the payment data has many dimensions and combination dimensions compared with other products, so it needs clearer ideas and overall situation to avoid falling into the data ocean.