How does operations perform data analysis? 1 process, 3 uses, 3 tools!

What is data analysis ? What exactly are we talking about when we talk about data in operations ? The author of the article summarizes a 133 rule based on the most commonly used data operation scope in our daily life. Let’s take a look~

During the traffic bonus period, everyone is scrambling to grab traffic from low-lying areas. Once the bonus period is over and the red ocean arrives, refined operations immediately begin to seize the top spot.

Refined operation is operation based on data. Everything follows the data, which means that data analysis is becoming more and more important.

So what is data analysis? What exactly do we mean when we talk about data?

Based on the most commonly used data operation scope in daily life, I summarized it into the 133 rule: one process, three uses, and three tools /methodologies.

• A process: analysis purpose, finding data, organizing data, analyzing data , graphical presentation, and result reporting.
• Three uses: current situation analysis, cause analysis, and predictive analysis.
• Three methods: DuPont analysis method, funnel model analysis method, and four-quadrant/matrix analysis method.

Next, let’s take a closer look.

1. General Data Analysis Steps

First of all, there is a process. When we need to conduct data analysis, what are the general steps?

The data analysis process can be divided into six steps: clarifying the purpose, finding data, organizing data, analyzing data, graphical presentation, and reporting results.

These are also the 6 steps to get started with data analysis, and the same applies to our operations.

There are many kinds of data. When you don’t use them, they are just a bunch of messy data. Only when you have a clear purpose of data analysis can you conduct data analysis.

The second step is to find data based on the purpose. These data can be exported directly through the background, collected on the Internet , or even sorted and input manually. Of course, many daily operation data are buried in the early stage for a purpose. Booking data according to the purpose is more targeted and accurate, just like a scientific experiment.

The third step is to organize the data . If it is exported from the background, searched on the Internet, or even entered manually, it is easy to have duplicate, blank, and messed up data, which will directly affect the analysis results, so the data needs to be cleaned.

The fourth step is to conduct data analysis based on the purpose of the analysis, and then the fifth step is to present it through graphics . The last step is to present the results of the analysis to the leader and the team, and analyze the cause and effect behind the data, and put forward suggestions and methods for solutions.

2. Three Purposes of Data Analysis

After clarifying the process, we can conduct data analysis according to the purpose. So when we do data analysis in operations, what are the daily purposes?

That’s right, there are three uses, namely: current situation analysis, cause analysis and predictive analysis.

• Current situation analysis is an analysis of the current status of a product . For example, market share, competitive product analysis , etc. are all part of the current situation analysis. For operations, analyzing all operational indicators at this stage, compared with competitors, which ones are weaknesses and which ones are strengths? Similarly, in different dimensions, what is the difference between our positioning and that of our competitors? These are all analyses of the current situation.
• The second use is cause analysis , which can be said to be the most commonly used one in data operations. Operations are work based on business lines. The business is changing all the time, both year-on-year and month-on-month, so we have to analyze the reasons for every change. There is a saying in the data circle: the most frightening thing is not being able to find the reason why business has declined, but not being able to find the reason why business has increased. Therefore, we need to find out the reasons for the rise and fall of business, the rise and fall of grading indicators, and the month-on-month and year-on-year changes of grading indicators. These all belong to the analysis of causes.
• The third use is predictive analysis , which is one of the core elements that best represents refined operations. In the era of traffic dividends, everyone is scrambling for cheap traffic, so naturally, as long as the budget is increased, a satisfactory ROI can be achieved. However, in the red ocean era, every budget must be spent wisely. Therefore, effect prediction becomes particularly important, and various A/B tests , effect chart predictions, small-batch effect tests for each channel , etc. begin to become important.

Testing models, images, and activities in e-commerce operations is a way to collect enough key data with a small budget and select the best product/material/solution. Testing the title of Chinese characters in new media operations is one type, internal testing of activity effects is another type, and internal and closed testing of games are also another type. These data analyses are based on purpose and collect data for effect analysis. It’s predictive analytics.

These three categories are the data analysis uses that are frequently used in operations. Different operations sub-positions may have different emphases, but we all need to be able to do them.

3. Three Common Tools for Data Analysis

After clarifying the purpose and collecting the data, the next step is to analyze the data.

There are many tools for data analysis. Of course, many small data-sensitive problems do not require special analysis at all. You can judge the problem at a glance, and then have N solutions to solve it.

We often say that experts who have accumulated sufficient experience in a certain operation field have this kind of subtle dimension of sensitivity. Of course, even the best intuition can go wrong sometimes. Moreover, before you demonstrate enough strength to convince others, you still need to use data to convince your superiors and the team. So we have all kinds of data tools .

Here we introduce three data analysis tools that are used in daily life: DuPont analysis, funnel analysis and matrix/four-quadrant analysis.

DuPont analysis method, for example, when operating an e-commerce business , if sales volume decreases, we can use the DuPont analysis method to analyze it.

(1) DuPont analysis

The DuPont analysis method comes from the famous DuPont. It is a method based on financial analysis, but as long as it is linked with the pyramid logic, it can be broken down step by step to solve many problems. For example: sales = customer order visitor conversion , and then visitors can be further segmented into page views, stays, churn, etc., and page views can be split into page views from different channels.

The average order value can be related to the number of personal payments, association rate, etc. The conversion rate can be broken down layer by layer with the add-to-cart rate, collection rate, etc., and ultimately the increase or decrease of each data can be found to find the reason.

(2) Funnel analysis method

The funnel analysis method is most commonly used in operations, especially event operations and growth operations.

The famous five-layer funnel theory of the AARRR life cycle coincides with the funnel analysis method. It can be said that the funnel theory is evolved from the funnel analysis method, and the visual presentation of data shows the shape of a funnel.

During the growth process, potential users - contact users - staying users - behavioral users - effective users - transaction users - secondary transaction users - core users - self-propagating users, each horizontal line is a percentage, and these percentages arranged from top to bottom form a funnel.

We can analyze the shortcomings in operations through changes in the funnel and optimize them accordingly.

The last one is the four-quadrant/matrix analysis method. This tool is most commonly used in market research and competitor analysis, and it is also very practical in management tools.

(3) Matrix methodology

The core of the matrix methodology is to find each analysis object. For example, for the market analysis of NetEase Cloud Music , find its competitors, such as Xiami Music, Kugou Music, Baidu Music, QQ Music, etc.

Then find two dimensions, namely the horizontal and vertical dimensions, such as: the age and professionalism of all target users . The four quadrants can be included in it: professional young people (90/00), non-professional young people (90/00), professional middle-aged and elderly people (70/80), and non-professional middle-aged and elderly people (70/80).

For example: in management, all employees are divided into two dimensions: values ​​and abilities. Alibaba then has little white rabbits, wild dogs, old oxen, and celebrities.

For example: After listing all the gaps between our competitors and us, each gap is placed on the coordinates according to its importance and urgency, and the team will have a goal and direction.

Even the catch-up of each gap can be followed up at any time, moving from this point on the coordinate axis to this point. For example: market share and product sales volume can be progressively improved every month, indicated by arrows on the coordinate axis.

This is also where the four-quadrant approach comes in handy.

4. Ready-made data + EXCEL is the entry point

Of course, many people will say that data analysis is not that simple. Indeed, the data analysis of experts is much more amazing than this. At least in Excel, I don’t know how to use many functions that experts can use easily, and I can’t write many programs that automatically collect data from websites. But for a simple introduction to data analysis, our common data background source, a simple Excel, can complete it. At most, the search, accumulation, and organization of data sources require gradual progress.

The rest is really not that difficult. After all, these are entry-level data analysis, which is enough for most operations.

• A process: analysis purpose, finding data, organizing data, analyzing data, graphical presentation, and result reporting.
• Three uses: current situation analysis, cause analysis, and predictive analysis.
• Three methods: DuPont analysis method, funnel model analysis method, and four-quadrant/matrix analysis method.

Source: Eggshell Peanuts