This is too much, this [operation analysis] idea is really on point

In this data-driven era, operational analysis has become an indispensable part of corporate decision-making. This article explores the application of data analysis in operations and provides a series of practical strategies and ideas to help operations personnel better understand data and make more informed decisions.

In the article “What is data operation and how to perform analysis?”, we listed the difficulties that operations actually encounter. Today, we will continue to share how data analysis can help operations solve difficulties.

As mentioned in the previous article, data analysis has provided a lot of support for operations, but unfortunately it is limited to the stage of understanding the current situation. So what else is needed to support the iterative upgrade of operations? This requires us to start with what the iterative upgrade of operations is actually doing.

1. How to perform iterative upgrades in operations

The reason why operations like to talk about iteration is firstly because operations have a large number of basic routines, templates, and cases to refer to, and there is no need to start from scratch. Therefore, there is no need to use words like "innovation, design, and creation".

Boys can recall the games you played, whether you get XX for the first deposit, XX for logging in for seven days, medals for competitive games, the routines are very similar. Girls can review the various discounts, discounts, and lucky draws on shopping websites, whether they look very similar. That's the feeling!

For example, AARRR is what operations like to talk about the most. In fact, there are a lot of routines in each aspect (as shown below):

Secondly, as the environment changes and the scale of enterprises expands, routines cannot be stuck to the same path and will always change over time. This change can be divided into five levels (as shown below):

These five levels of change will be carried out according to a process:

After reading the above, do you feel that data can do a lot of things? But wait, how much it can do depends on the specific type of operation. If you don’t choose the right service object, it may be superfluous.

2. Data requirements for different operations

Although they are all called operations, operations actually involve a lot of work. Different operations have different specific pain points. For these pain points, the degree to which data analysis can treat them also varies.

In essence, data analysis methods represent rationality, logic, and calculation. But this is not all of the work. There is a lot of sensibility, emotion, and creativity in the work. Therefore, some jobs do not naturally require data help, and just look at the current situation and results, while others require careful calculation and analysis. Combining the above factors, it can be summarized as follows:

This is why all the detailed analyses you see are related to users. Because user operation itself is a very strategic job.

User operation itself is also very important. In order to go public and raise money, many Internet companies need to achieve a certain level of user volume, user growth rate, and payment conversion rate. They are willing to spend money on channel operation (attracting new users) and user operation (cultivating old users).

3. What questions can the data support?

Data analysis is suitable for solving rational problems, so after reading the above classification, you can probably know which problems data analysis is suitable for. But don’t forget that the biggest problem of operation is lack of money. Therefore, we have to add the degree of expense demand for each type of work, and solve the expense problem of those departments that lack money first (as shown in the figure below).

Therefore, in theory, the first step of data-supported operations should start with "sharing money". First answer questions related to money, which are very rational and strategic, such as: The company found that the goal is (industry No. 1? Revenue exceeds 10 billion?) ● Based on this goal, XX million new users are needed, and the old active users are maintained at XX level ● Based on the number of new users, according to the current market price, the channel cost is XXX billion ● Based on current measures, the cost of maintaining old users is more than XXX billion. The goal is to achieve x% through phased promotions, and daily channel/user investment is X%

With these analyses (actually business analysis), money, time, and responsibilities are clearly separated, and subsequent operations are very refreshing! I am never afraid of high goals, but I am afraid of not paying enough. With financial support and appropriate time arrangements, it is also easier to choose specific implementation methods in the future (as shown below):

As for the specific implementation level, there are too many subdivisions to explain it in one article. We will update it slowly when there is a chance.

4. How to implement the data

However, just having these analytical ideas and methods is useless! The more important part is implementation. Everyone understands the principle, but it is painful when it comes to implementation:

1. Analysis and decision-making are out of touch: This is the biggest, biggest, biggest problem. Decisions are often made on the spur of the moment, based on experience, copying opponents, or following instructions, without real analysis. The analysis is actually just updating data, without opinions, interpretations, or insights.

2. Decision-making and execution are out of touch: This is the second biggest problem. Often, the direction, cost, and strategy are decided by the upper-level leaders, and the grassroots students are busy every day: making plans-asking for instructions-changing plans-asking for instructions-changing plans-asking for instructions. They are completely confused about why they are doing this and where they should stop. No analysis is useful.

▌3. Theory is separated from practice: This is the third biggest problem. We are familiar with AARRR, but when it comes to a specific industry, a business, an activity, or a copywriting, we have no idea what the data format is and what the appropriate amount is.

▌4. Lack of historical experience accumulation: There is no collection and accumulation of past data. Even many students who work with data don’t even know what the business is currently doing, let alone what they have done in the past. It’s impossible for them to perform analysis.

▌5. Lack of activity, planning, and copywriting labeling system: Just as it is difficult to understand users without user labels, without these business labels, it is impossible to specifically classify and compare businesses, let alone summarize routines.

To sum up, a good data support system is always the result of integrated business data operations and collective efforts. There has never been a god-level data analyst who can solve all problems with a simple “ah la la la”. This also applies to case sharing. Many students like to say: Here are two awesome cases. In the end, it was discovered that awesome cases were always catalyzed by awesome companies. If you want to replicate, you still have to practice basic skills, such as how to label business.