5-step method to quickly find data analysis ideas

At work , A little friend often meets ： There is no way to do data analysis . If it's daily work, it's ok , You can copy a copy of the previous report . But I have no idea during the interview , It may end with hate . Let's share it today , How to find ideas quickly .

For example, being asked by the interviewer ：“ How do you analyze a APP”. At this time, the brain suddenly short circuited , I don't know where to start , What to do ？ take it easy , In five steps , Step by step .

First step ： Set the scene

First , Don't talk about “APP Analysis depends on XXX indicators ”, It's easy to hang up . because APP There are too many kinds of , Different APP Function differently , There must be differences in the indicators . Therefore, the best way is to specify a specific APP, Lock the problem scenario , Let's talk more about , It's not easy to roll over .

At this point you can ：

• Take out a cellular phone , Point to an e-commerce / social contact / Short video / Take-out food / Take a taxi APP
• Download the interview information in advance APP
• Point to your former company APP

And said, ： Let's take this as an example . In short, to a specific , It will be much easier .

The second step ： Clear indicators

With specific scenes , Can be clear ： Data indicators to be analyzed in this scenario . Data indicators are the basis of analysis , Set the target first , Let's talk about the classification dimension , Let's talk about how to analyze .

And in common scenarios , The analysis index is fixed ：

• Product analysis ：DAU、MAU、 Activity rate 、 Active duration 、 Main path conversion rate
• Sales analysis ： Sales volume 、 Sales profit 、 Number of customers purchased 、 Purchase conversion rate
• Commodity analysis ： Inventory 、 Sales volume 、 Moving rate 、 Inventory turnaround time

……

So as long as there are specific scenes , You can easily find the key indicators , Analyze . If you are not familiar with common indicators , You can refer to the previous post ： Great collection ！ Common data analysis indicators in the Internet industry

Be careful , In this case , The interviewer asked APP analysis ,APP It's a lot of things , Fully expanded, with a lot of content , You can give a frame first , Then pick one aspect and go deep . For example, e-commerce APP, You can look at active users first 、 Consumer users 、GMV Several core indicators .

The third step ： Look at the rules

After setting the data indicators , First pick the key indicators , Draw the development trend line . For example, product analysis , It can be assumed that the product DAU The trend chart is as follows , Then take the initiative to tell the interviewer ： Through observation DAU The trend of , You can find ：

• Whether there is a periodic change law
• Major events affect
• The trend is good / To bad

（ Here's the picture ）

This step analyzes , Lay a foundation for further analysis . Here, we can use the periodic analysis method among the nine data analysis methods , See previous sharing . Of the nine big data analysis methods ： Periodic analysis

Step four ： Find exceptions

In the third step , We give the normal law , Corresponding , It can also be monitored through data , We found something unusual . So as to lock the problem points to be deeply analyzed （ Here's the picture ）

So first draw a shape , Then discuss whether there is a problem , Can make the interview easy . Some people skip this link , Finish the index , Just say ：“ If the activity rate is low , I want to disassemble XXX Analyze ”, It's easy to be challenged by the interviewer ：“ How do you know that low is abnormal ？”

What if the data is wrong ？

What if it's a normal fluctuation ？

What if it's a slight fluctuation ？

What if it's cyclical ？

These questions , Can be easily resolved through a picture . Combining graphics , It also shows that we understand periodic analysis , Will look for the law of fluctuation , Kill two birds with one stone .

Step five ： Check the cause

After finding the exception , You can go deep into the cause . The simplest method is structural analysis + Index disassembly method . Disassemble the problem , Find the point where the problem occurred . Not all possibilities are necessarily exhausted here , The main classification dimensions can be listed clearly , Let the interviewer see the ideas . Facing the open questions of the interview , That's enough to answer （ Here's the picture ）.

Summary

After these five steps , The answer is perfect , Remember to summarize after answering ：“ The above five steps , Is that we started from 0 Start analyzing a XX class APP The process of . So even if you don't know what to do at first App, You can also find out the data rules step by step , Find the problem .” If the interviewer is not picky , You can get through it ！

matters needing attention

matters needing attention 1： Set the scene stage , Don't look for complex business scenarios . There are some APP It looks simple , The business behind it is complex , Take out, for example APP, From the user's point of view , Just order a takeout . But then , It also includes booking 、 Travel 、 Cycling all kinds of business , And the platform side 、 The business of the merchant , It's complicated . Tell me about this complex example , It's easy to roll over . Therefore, it is recommended to use a simpler business , Direct sales APP/ Applet （ For example, I personally like X Black duck Mall ） So the probability of rollover is much lower .

matters needing attention 2： Activity questions , hold “ Look at the rules ” Switch to “ Set goals ”. The above example , In order to App For example ,App Operation is daily work , Therefore, there is a long period of continuous data . There are other jobs that are not routine , It's periodic . Like promotions , Like advertising .

here , You can put the third step “ Look at the rules ” Switch to “ Set goals ”, General phased activities 、 Extension , They all have clear goals . Like getting XX A newcomer , increase XX Ten thousand sales . After setting a goal , Through the difference between the actual data and the target , Go to step 4 to find the abnormality . Some people don't pay attention to this , There was no goal in the interview , Just blabla Talking about a bunch of indicators . The interviewer asked ：“ Why do you say activities are bad ？” Just ask ya Meng . Goals matter , Be sure to book it in advance .