Let’s talk about the daily work of product data analysis!

There is still a lot of work to be done in data product operations , from being able to discover problems in the data, forming effective suggestions for product optimization, to finally providing guidance on commercial products.

Data analysis is a boring job. It requires you to have good logical analysis skills, proficient business capabilities, and keen industry insights, so that you can ultimately use data to drive commercial decisions.

Data analysis positions are often referred to as data operations . I think this is another position that comes into contact with users besides PM and customer service. A good data operator can even become an excellent product manager or company manager: not only can he or she have a good understanding of user behavior, but also be able to monitor activities and effects in a comprehensive manner; he or she can even predict industry development trends and give decisive suggestions for major company decisions.

Regarding the daily data analysis work of product data operation, I think we can start from the following angles:

User portrait analysis:

User portrait analysis is one of the main methods to obtain target users for products and is also a part of daily data analysis work. By labeling users with information such as gender, age, income, and region, and if other user behaviors (such as access behavior and payment behavior) can be connected through the account system, a complete user portrait database will be formed, which will play a decisive role in precision marketing (the e-commerce websites such as JD.com and Taobao should do the best job in user portraits. It can even predict that you may need a certain product in a certain period of time in the future, and then push targeted product promotion information through email, text message, WeChat, etc.).

There are many ways to obtain user portraits. Here we focus on some common ones. You can choose the most suitable one according to your own product positioning and internal resources:

Method 1

Relevant functions for user basic information can be embedded in the product, and users can be encouraged to complete their personal information through task guidance and appropriate reward systems. I have also seen some products where some advanced features are enabled by completing personal information, which is also a good approach. It is important to note that users should not spend too much time to complete their information, and their privacy should not be compromised to avoid user disgust and loss.

Method 2

Utilize some third-party monitoring platforms: such as Umeng , Google Analytics (GA), or Baidu Index, etc.; these platforms have statistics and analysis of basic user portraits (of course, they are subject to the influence of cookies. When users clear or refuse to read cookies, there will be some deviation in the data, which requires data collection and cleaning). The advantage of using these platforms is that they can be connected with advertising data, and can also obtain data development trends of industry competitors.

Method 3

Listen to users' voices regularly, such as through questionnaires and return visits, and use sampling methods to predict the overall user portrait level. At the same time, due to the flexibility of question setting, you can also obtain a lot of information about competing products in the same industry.

Traffic monitoring:

Traffic monitoring is something that needs to be done from the day the product is born, because it not only involves the direction of product iteration, but also can use data to tell us which functions are useful, which functions are not useful and need to be optimized, and even which functions are useless and need to be abandoned. For event operations, traffic monitoring is also one of the most important references for summarizing event results.

Whether the internal technical team collects data by themselves or uses third-party tools for data monitoring (I would like to add one more sentence here. The choice of self-development or using third-party tools for monitoring and statistics must be considered based on the actual needs of the product and team resources. Large companies have more resources and often choose to develop by themselves because it involves data security and accuracy; small products can consider choosing better third-party tools on the market for data collection), it must be done as early as possible and as carefully as possible.

For a website, a complete site map is one of the essential functions. Each page needs to be placed with the correct monitoring code to monitor user access (PV/UV), bounce rate, page dwell time, page access depth (i.e. how many pages are visited), access channel source (from which website and in what way), retention rate (next day traffic, 3-day retention, 7-day retention, 14-day retention, 28-day retention), etc. Key processes must be deployed correctly, such as the registration process (involving new users), the purchase process (involving conversions), etc. At this time, the conversion funnel is an important tool to help us do page analysis. Through the funnel, you can see the traffic entering and conversion of each key page, as well as the user leaving ratio. If there is an abnormality in the process data of a funnel, you need to focus on whether there is a problem with the product function. If you use monitoring tools such as GA, you can achieve the interconnection of advertising delivery and user access behavior data, and use the attribution model to analyze the shooter channel and assist channel. This can not only optimize advertising and increase conversion rates , but also discover new cooperation channels and even new user concentration groups.

For App, DAU, MAU, Interactions, access depth, etc. are the data we need to focus on. Compared with website monitoring, app data monitoring is more suitable to start from the account system. Each user is an independent individual and has independent access behavior. At the same time, by connecting with national energy and portrait data, we can obtain important basis for different types of users' product access behavior and product function requirements.

Revenue (conversion) monitoring:

Revenue monitoring is an important basis for measuring the commercialization level of products. The target form of products is to achieve commercialization, so different types of products must have continuous monetization capabilities, otherwise they will gradually be eliminated by market competition.

The data monitored daily include revenue flow, profit, profit margin (year-on-year, month-on-month), subsidy, subsidy rate, number of users’ first payment, number of repeat payments, retention rate, etc. Generally speaking, this type of data is written directly to the backend database, which means that only internal employees of the product can view it, and they may be assigned different viewing permissions. Some companies also require product data operations personnel to have certain SQL capabilities, be able to read database codes, and be able to write or describe requirements clearly so that technology can help you write them.

The above is a brief introduction to the daily monitoring work of product data operations and some of my own thoughts at work. Data analysis work itself requires employees to be familiar with the industry, and at the same time constantly accumulate experience in their work and make good use of some resources to integrate and analyze data. Simply looking at one item or one aspect to obtain information is definitely one-sided.

There is still a lot of work to be done in data product operations, from being able to discover problems in the data, forming effective suggestions for product optimization, to finally providing guidance on commercial products.

There may be minor errors in the opinions expressed in this article. We welcome your criticism and correction, as well as interactive communication.