[Video] An excellent operator should have the ability to analyze data in these four aspects

Open class video recording (about 50 minutes)

In the past, the concept of “traffic is king” made the responsibilities of operators focus on attracting new customers. As the market environment changes, the channels and methods of operations continue to increase, and there are more detailed classifications.

In the context of refined operations, how to use data analysis to solve growth problems in traffic operations, user operations , product operations , and content operations. Today, I will share with you GrowingIO’s experience in data operations.

1. Traffic operation: multi-dimensional analysis and channel optimization

Traffic operation mainly solves the problem of where users come from. In the past, extensive traffic operation only focused on vanity indicators such as PV and UV, which is far from enough in today's refined operation.

1. Traffic indicator system

We need to use multiple dimensions and indicators to determine the basic traffic situation, including magnitude indicators, basic quality indicators, and visiting user type ratio indicators. Magnitude indicators involve different platforms. The Web side mainly looks at visits, PV, and UV, and the APP mainly looks at the number of launches, DAU, and NDAU. Basic quality indicators include the average visit duration of users, the average number of pages viewed in one session (i.e., visit depth), and bounce rate. These indicators can be used to determine the user's activity. The product life cycle model is widely used in Internet operations. In different product life cycles, the types of visitors are different.

GrowingIO - Overview

An excellent operator should be familiar with the traffic profile of his or her product. By checking the traffic of the website every day, the operator can clearly grasp the traffic indicators and their changing trends, which is convenient for evaluating the past and predicting future trends.

2. Multi-dimensional traffic analysis

In website traffic analysis, the main focus is on access sources, traffic entrances, advertising, etc.

First of all, the sources of visits include direct visits, external links, search engines, and social media. Under this analysis framework, we need to break it down layer by layer and conduct traffic analysis on each channel.

Take the blog of GrowingIO as an example. This is a sub-site of content operation, which contains many articles on data analysis and growth hacking. Through visit source analysis, we found that compared with other channels, the number and quality of users coming from Weibo are relatively low, which reminds us that we need to optimize content channels.

The second is the landing page, which is the entrance for users to reach your website. If users are directed to invalid or irrelevant pages, there will generally be a high bounce rate.

Finally, advertising is also an important part of current traffic operations. The advertising analysis generally involved includes advertising sources, advertising content, advertising formats (clicks, pop-ups, effect guidance) and sales sharing, etc. We need to optimize advertising through multi-dimensional analysis.

The above three angles are mainly for analysis on the web side. For APP analysis, factors such as distribution channels and APP versions also need to be considered.

3. Conversion funnel analysis

In the growth model, after traffic enters, it needs to be further activated and converted, and activation requires certain processes and steps. A funnel like this can clearly show the conversion status of each step.

Taking the above picture as an example, we analyzed each step of the conversion and found that the loss rate from the first step to the second step was the highest, which requires targeted optimization. By analyzing the conversion rate of different channels, we found that the conversion rate from Baidu Brand Zone (bzclk.baidu.com) was as high as 44%, while the conversion rate of other channels was less than 3%.

4. Channel optimization configuration

After conducting a series of traffic analysis and conversion analysis, we can formulate corresponding strategies, including search terms, landing pages, advertising optimization, etc.

For channels with low cost and high quality, it is necessary to increase investment. For channels with high cost and high quality, it is necessary to evaluate the cost. For channels with low quality, it is also necessary to conduct an evaluation. Overall, the channel configuration should be managed and optimized based on comprehensive factors such as cost and traffic conversion.

2. User Operation: Refined Operation to Improve Retention

If traffic operation solves the problem of where users come from, then user operation is about establishing and maintaining relationships with users.

1. Refined operations

Users have many interactive behaviors on the product. We can classify users based on their behaviors, and then conduct refined operations based on the characteristics of different groups to promote user return visits.

Taking forums as an example, users' behaviors on forums include: visiting, browsing posts, replying, commenting, posting, forwarding, sharing, etc. We establish a behavior index for each type of user behavior. For example, we establish a "communication behavior index" based on the user's forwarding, sharing and other behaviors, and classify users according to these indices. In this way, users on the forum are divided into four dimensions: A browsing, B commenting, C dissemination and D content production. A user may have only one label index or may span multiple index dimensions.

In subsequent user operations, classification can be carried out according to these tags. For example, UGC forums need to maintain the activity and growth rate of Category D (content production) users; at the same time, in the promotion and dissemination of the forum, it is necessary to stimulate Category C (dissemination) users to expand the dissemination and influence of the content.

2. Improve user retention

Internet products generally focus on user retention. Only when users stay can they further promote monetization and dissemination. Retention analysis generally uses cohort analysis, which is to analyze people with the same characteristics within a certain time frame.

GrowingIO - Retention Chart

In the retention chart above, the horizontal comparison shows the retention rate of new users added each week in the subsequent weeks, and the vertical comparison shows the retention performance of new users in different weeks over the next period of time.

Retention time and cycle are related to the complete cycle of product experience. Different businesses and products generally have different time group division methods. For example, the daily retention rate of high-quality products better reflects the relationship between users and products, while the weekly retention rate of tools is more meaningful than the daily retention rate.

It is very important for user operation to discover the changing trend of user retention through time dimension analysis, discover the differences between users in different groups through behavior dimension analysis, and find the growth points of products or operations.

3. Product Operations: Using Data to Analyze and Monitor Functions

Product operation is a very big topic, which I will mainly share today.

1. Monitor abnormal indicators and find out what makes users angry about your product

There are many small functional points in the large product process. The user experience is built on these small functional points, and the usage of these small functional points becomes the key to our conversion at each step.

Taking the registration process as an example, mobile phone verification is generally required. Sending a verification code is one of the key conversion nodes; when the number of times users click to resend surges, it may mean that there is a problem with this function. This is what makes users angry: they cannot receive the mobile verification code in time. By monitoring key indicators, we can identify problems and fix them in a timely manner.

2. Test the effect of new features through retention curves

For products that have been online for a while, new features are sometimes added. After going online, it is necessary to evaluate the effectiveness of the new function, whether it meets the core needs of users, and whether it can bring value to users.

From this retention curve, we can easily find that the proportion of people who used the new function on the first day and continued to use it was very low, which shows that this function did not solve the user problem well; this reminds us that we need to rethink the newly launched functions.

4. Content Operation: Accurately Analyze the Effect of Each Article

Before doing content operations, you need to understand whether your content is a product (such as Zhihu Daily) or an auxiliary function of the product. Only by understanding your own position can you set clear goals. In order to expand the effectiveness of content operations, we need to analyze user needs, such as the content that users are interested in, the proportion of content reading and dissemination, etc.

1. Category of Content

Take GrowingIO’s technical blog as an example, this blog belongs to the PGC model. The content in the blog is divided into different categories. In order to reduce the cost of users obtaining information, we have designed entrances to different sections on the blog homepage, including category navigation on the left, article recommendations in the middle, and hot recommendations on the right.

We found that users mainly read articles through the navigation bar on the left and the recommendations in the middle, and rarely click on the hot recommendations on the right. Therefore, on the mobile side, we cancelled the hot recommendations on the right and only retained the category navigation and the recommendations in the middle. It not only saves space, but also maximizes the user's content needs.

At the same time, we also analyzed the content of the category navigation bar and found that users were most interested in case analysis content, which is a very good inspiration for our future content selection.

2. User-based recommendations

Recommendations in content operations are sometimes closely related to refined user operations. Every user has his or her favorite content and categories. When we push content based on the user's interests, the efficiency will definitely be higher.

Taking the blog of GrowingIO as an example, we counted the clicks on articles by visiting users and obtained the results in the above table. Obviously, user 8 has his own preference for "growth secrets", and users 6, 7, and 9 prefer "case sharing" articles. Then in the actual content push, we can push growth secrets articles to user 8, push case analysis articles to users 6, 7, and 9, and push to other users indiscriminately.

The core of Growth Hacker, which has become popular in recent years, is essentially to achieve refined operations and achieve the goal of growth through technological innovation and data analysis. An excellent data operation personnel should have data-driven thinking and master certain data analysis tools. In actual business work, we constantly raise questions from the data, keep trying, and use data to optimize operational strategies, thereby achieving customer and business growth.