Data Methodology | 5 Steps to Analyze Your Own Product Operation Data

Due to the internal development of the company, Tuanzi Xiaobai received the task assigned by the company, responsible for platform data statistics and analysis. In this regard, Tuanzi will briefly describe the process and content of data analysis based on his own experience, and share some commonly used data websites with you.

For a company, data reflects the company's recent development status, whether the recent development status is upward, stable or downward, and the company's current dynamics. Secondly, you can get specific analysis results through specific data analysis, such as your target users , user whereabouts, usage of a certain function, conversion rate , retention rate , etc.

Data analysis can generally be divided into three categories:

1. Industry data analysis
2. Competitive product data analysis
3. Own product operation data analysis

The uses of the three types of data analysis are also very different. Industry data analysis and competitor data analysis are mainly used to help with the writing of BRD and MRD, understand social trends and competitors, and learn about the overall situation of the entire market and future development trends . The analysis of product operation data is mainly to help with the writing of weekly and monthly reports, the development of the current situation of the product, and understanding the overall status of the platform . At the same time, Tuanzi mainly describes how to analyze the operating data of its own products, what to analyze, and what key indicators are needed during the analysis.

When conducting data analysis, the following steps are generally followed: purpose - data collection - data statistics and description - induction and conclusion - suggestions/improvements.

1. Purpose

What is the purpose of data analysis? Tuanzi thinks that this is why we need to analyze data. You want to get something from the data to help you prove something. Once, Tuanzi's department wanted to modify the platform's incentive mechanism for merchants. To this end, Tuanzi needed to conduct a questionnaire survey to find out what merchants thought of the previous incentive mechanism, the degree of incentive, and how they felt about the future modified mechanism. How much do they need it? This is what I want to achieve through research data.

When the company assigned me to analyze the data of the platform every week, the first thing I learned was writing a weekly report. Tuanzi will check the documents output by the previous person in charge for deep learning. From this, Tuanzi summarized a few tips through the previously output documents:

1. Emphasize the importance of data (mark key data in red).
2. Important conclusions drawn from data analysis are highlighted in red.
3. If the conclusion can be supported by data, use data to support it; data is the most reliable evidence.
4. When analyzing data , indicate the source of the data. So that readers can know where the data comes from and whether it is feasible?

2. Data Collection

When collecting data, the first thing we should know is what key data to collect? What kind of data to collect depends on your purpose.

For example, the internship company where Tuanzi works is engaged in e-commerce live streaming . In the weekly report, the following indicators need to be reflected:

• APP-related data (total number of new APP downloads, status of APPs in competing industries, download data of Gedantang application market )
• User retention rate
• Average visit duration of users
• Order data
• Merchant order data
• Live broadcast related data (average number of viewers per game, number of live broadcasts, live broadcast room forwarding, data of each live broadcast type)
• Number of user registrations
• User churn rate

In start- up companies, there is not much experience in building their own system database, so they usually look for third-party platforms and outsource them. Because if the background needs to establish powerful data, a lot of technology is needed to support it. However, some special data platforms will be developed by ourselves, such as Tuanzi’s live broadcast data, order data, and user information. For startup companies, if outsourcing can reduce the company's costs, they will outsource as much as possible, after all, it is linked to current demand value.

3. Statistics and description of data

Before we conduct data statistics and description, we first need to understand what some key data indicators represent and the formulas for how to conduct statistics.

(1) Total number of new APP downloads

This indicator is mainly used to understand the current platform products and users' download situations in major platforms during the corresponding statistical period, and at the same time, to understand the situation in major application markets. Statistics on the total number of new APP downloads, divided into Android and ISO.

Tuanzi’s company mainly uses third-party data monitoring platforms ( Umeng and Tencent Cloud) for statistics. If you want to get data about your competitors, Tuanzi recommends a free platform for you: ASO100 .

(2) User retention rate

The number of times a user uses the product at a time over a period of time. Tuanzi gives an example: 1,000 new users were added on July 10, 500 users were added on July 11, 400 users on the 12th, 300 users on the 13th, 200 users on the 14th, and 100 users on the 15th. Therefore, the 10-day retention rate is 50%, and so on, as shown in the figure. How to make a funnel chart? Ask Baidu~

(3) Average user usage time

The time users spend on the platform is measured through an average algorithm to understand the overall usage time and eliminate abnormal usage. It is calculated as the sum of usage time of all users / total number of people. The higher the average usage time of users, the more important the platform is to the users. Therefore, we can use usage time as one of the indicators to divide users into different levels.

(4) Number of user registrations

The number of new users registered on the current platform within a certain period of time.

(5) User activity

Within a certain period of time, for example 15 days, if a user is active on the platform for one day, then his activity level is 1. If it is activated twice within 15 days, then its activity level is 2. By analogy, the higher the activity level, the higher the contribution to the platform.

Once in a product social group, someone asked how to segment users? Tuanzi believes that the division of users is mainly based on the degree of contribution of users to the platform. The degree of contribution can be calculated based on two dimensions: user usage and payment situation. The weight of the calculation is based on the current platform situation. For example, if the platform focuses more on payment, the proportion of payment will be relatively high.

(6) User churn rate

Active users in one observation cycle who are inactive in the next observation cycle are called churned users. User churn rate = churned users / total users. Use churn rate to understand where the platform is going.

(7) User return

Set up three observation cycles. Returning users refer to users who were active in the first cycle, lost in the second cycle, and became active again (returned) in the third cycle.

Understand the above-mentioned popular indicators, obtain the corresponding data through the platform, and then use Excel to organize the data and make charts.

When describing data charts, we always need to remind ourselves why this data appears, what causes this result, what improvements we need to make and what suggestions we need to put forward for the current data situation.

Case analysis: Tuanzi takes the recent “ Minions Collection in the City” event held by Ofo as an example, obtains their recent download situation through ASO 100, and describes the data. For example, this is the download situation of the Android system of Ofo. The event time is from July 7th to 14th. In order to let everyone better understand the impact of this event on the new downloads of Ofo, the time period is set from the 4th to the 18th, and combined with the situation of other competing products ( Mobike , Bluegogo) for analysis.

From the following charts, we can get some information:

• Judging from the time period of the data , Ofo is launching a series of Minions-related themed activities, which will continue from the end of June to mid-July. This results in a large user download base, so the total number of downloads is about 3 times that of other competing products.
• Looking at the following chart from the perspective of the overall market , we can observe an interesting phenomenon: the download fluctuations of the three companies are very similar and identical. Although Ofo has stimulated the increase in their total downloads through activities, it has also caused an increase in the download range of its competitors. If you look at them over a longer period of time, they will show a fixed pattern. This is market volatility.
• Let’s look at the data of Ofo from the user’s perspective. In the early and middle stages of the campaign, user downloads showed a downward trend, but in the later stages the downloads skyrocketed. Tuanzi believes that since the event lasts for one week, users are busy stockpiling Minions cards in the early and middle stages of the event. When the activity continued to the later stage (the product design of Ofobi always reminded users that the activity was about to end and updated the number of winners at all times), users felt a sense of urgency and desire, and many users, who were missing a certain card, hoped to obtain that card through exchanges with friends and receive a cash reward of 7 yuan. Therefore, users are encouraged to share and interact with each other, which leads to a significant increase in user downloads in the later stages of the activity.

Summary: The arrangement of related activities in the early stage was of great help to the promotion of this event.

The setting of the friend exchange function greatly increases the interaction between users and their friends, thereby greatly increasing the number of downloads by subsequent users.

The design of the event, including reminders of the number of winners and event time, stimulated users' desire and sense of urgency, greatly enhancing the atmosphere of the event.

Sometimes, the incompleteness of data may lead to deviations in the results of data analysis.

When Tuanzi was analyzing the Ofo campaign, he saw the first result and mistakenly thought that the decline in downloads was due to the rules of the campaign, which led to a decline in downloads in the early and middle stages of the campaign. When Tuanzi picked up the data chart of the overall industry, he realized that the reason for the decline in downloads was more due to the overall market fluctuation effect.

Don't be a frog in a well. Occasionally jump out and look at the overall picture, then you will know that everything is like this.

4. Induction and conclusion

As for summarizing and generalizing, Tuanzi personally does not have much experience, so Tuanzi will share some skills in summarizing.

When making a weekly report, you need to spend a lot of time on data organization and description. However, many colleagues, when browsing your weekly report, generally only read the summary part carefully, and many of them just glance over the other parts. After all, a weekly report is very long, and many colleagues do not have much time to read it.

Therefore, when writing a weekly report, we should first put the summary at the top ( the most conspicuous place ); at the same time, for important data and summary points, be sure to mark them in red so that your colleagues can quickly obtain the corresponding information points.

Regarding how to summarize, Tuanzi recommends a few books to make up for this deficiency. " The Pyramid Principle ", "Learn to Ask Questions", and "McKinsey Working Method", I hope these can help you.

5. Suggestions/Improvements

I don’t know if you have the same mentality as Tuanzi. Every time I write suggestions/improvements in my daily summary, I have a lot of ideas and suggestions at the beginning. As time goes by, I feel like I am just racking my brains and basically have no ideas at all, which is very distressing. I would like to ask you guys how you solve this problem?

Tuanzi is very happy to share his experience and thoughts on data with everyone. Although the article is short on content , this is largely due to Tuanzi himself. Although Tuanzi does not have as much practical experience as other authors, he still works very hard to share his experiences with everyone so that we can grow and communicate together.

The above is my data analysis sharing. I hope everyone will pay more attention to Tuanzi. Tuanzi will share this summer product experience with you.

Next, I will share some data websites with you. I hope it will be helpful to you in the later data work.

• ASO100: https://aso100.com
• Baidu Index: https://index.baidu.com
• 199IT Internet data: http://www.199it.com
• IT Orange: https://www.itjuzi.com
• Baidu Mobile Statistics: https://mtj.baidu.com/web/welcome/login
• China Internet Data Platform: http://www.cnidp.cn
• Ebrun e-commerce data: http://www.ebrun.com/data/
• Browser market share: http://tongji.baidu.com/data/browser
• Chinese online video data: http://index.youku.com
• Alexa website monitoring data: http://www.alexa.cn
• New Rank - WeChat public account monitoring data: http://www.newrank.cn
• Analysys/iResearch Database
• Baidu

I hope the above data website can help you~