Data analysis by "demonization" (Abstract)
| Font size subscription
Why is Data Analysis "demonized" because new users always think that data analysis is extremely complicated and highly technical. They are always concerned about some professional Vocabulary (ARPU, PCU, and many letters and combinations I Don't Know), how to create charts, how to use Excel tools, and how to write conclusions...
Let me talk about how I view data analysis:
Data analysis is a method, but not the only method.
The advantage of data analysis is relatively objective, but its disadvantages are also obvious, resulting in high labor and time costs.
The data analysis in the game is nothing more than two goals:
1. Discover the essence of existing problems and solve them (99%)
2. discover some trends to make the future better (1%)
The first account accounts for 99% of the total accounts! The second goal is that I have never seen anyone do it, and I have never done anything similar. At most, the data is too much to read. When a blind cat hits a dead mouse, it finds a trend.
Therefore, data analysis is mainly performed to discover and solve problems. There are many ways to discover and solve problems. Sometimes data analysis is not the best solution.
For example, the new version is about to be updated soon, and a pile of preparation work is not completed. At this time, you find that the new equipment is sold badly, which is far less than expected. If you still spend a lot of time analyzing why that equipment is not sold well, you will be delayed for more important new versions! When time is not enough, separate the primary and secondary nodes, and avoid wasting time on data analysis. When you talk to different types of users, you can find the main problems.
Data analysis is not omnipotent
Data analysis can discover the root cause of code problems, but it is difficult to explain user behavior.
Click to open the big chart
For example, when we discover the loss of Big R in the game through data. What data can do is over. Data cannot tell us the specific cause of data loss. The data cannot tell us whether users stop playing games because the company went bankrupt or because they ran to play other games, or because they were tired and didn't want to play...
The data is often unclear, but through continuous data segmentation, we can narrow down the scope of the problem, rather than finding a needle in the sea.
Therefore, data analysts of niub are useless if they do not know products or users!
Data analysis is not just to pile up charts and text.
I have seen a lot of data analysis: the layout is neat and the chart is very beautiful. Every page has a company logo and copyright description. At first glance, I feel good!
But when I look at the content, it's a pile of garbage!
- Data Sampling is completely unscientific
- There is no subdivided data at all, and there is only a bunch of macro data that cannot explain the problem.
- No comparison data
- Data cannot support the conclusions drawn from "analysis ".
Summary: The conclusions are subjective and independent from the stacked data and charts.
Data analysis is a very rigorous task. Every conclusion should be drawn from the data. The data cannot be explained only by speculation. So when we write down every conclusion, we must know whether it is "I think so" or "I found it from the data "! For more exciting content visit, http://www.gamedatas.com
