What kind of data analysis should be done from the start-up to the maturity of APP?

DAU, MAU, retention rate , frequency, duration...What data do product managers need to analyze? Combining the data analysis practices of overseas mobile products, the author will take you through the different stages of product development from inception to maturity, and see how data analysis is applied to product design and product operations .

According to the popular classification, the product life cycle (PLC) is divided into the start-up stage, growth stage , maturity stage, and decline stage . At each stage of the product, the work weight and analysis focus of data analysis are different. Let's talk about it according to stages and combined with cases.

1. Start-up period

The focus of the start-up phase is to verify the core value of the product, or to verify the product's hypothesis: a certain product or service can solve a problem for a specific group of people. This stage should follow the MVP (Minimum Variable Product) idea, verify the entrepreneurial idea at the lowest cost, and quickly iterate to adjust the solution based on user feedback, which will ultimately be verified by data.

Examples:

Take a foreign mobile forum social application we developed before as an example. During the idea stage of the product (around 2012 and 2013), we found that forum users often complained that accessing the forum from the mobile Wap page was slow, there were many advertisements, and there was no mobile adaptation at all. So we proposed a hypothesis: to make an App to connect the forum system with users, so that forum users can also enjoy a smooth forum access experience on mobile devices, and users are willing to pay for this experience.

Therefore, in the early stage, the entire product was completely explored around the two core scenarios of reading posts and posting posts. It was promoted in the forum and sold for $18. We found that many users paid for it, and the retention rate of these users reached 60%+ (of course, this is related to the fact that users paid), and half of the users used it for more than 70 minutes. Not long after that, some competing products emerged (the Vbulletin team, the largest forum system at the time, developed a mobile app intended to solve the same problem), but they all fell far behind us before long. This was because the entire team followed the MVP idea and focused on repeatedly polishing the smooth experience of reading and posting posts based on user feedback, which gained very good user reputation and led the market, and also obtained investment from a famous Silicon Valley investment institution.

Key data - target population profile

In addition, during the start-up phase, you can connect to some third-party application monitoring SDKs to understand the portrait of the initial user group, and indirectly verify whether the user group is consistent with the characteristics of the assumed target user group. Common ones are demographic attributes (gender, age, education, and region).

Examples:

In early April this year, I talked with the product manager of a domestic fitness APP. The APP was originally a fitness and step counting tool App. In the early stage of the product, the next-day retention of new users was at the industry average level. When observing the portrait of the target user group, it was found that there were significantly more female users than male users, and the retention rate of female users was significantly higher than that of male users. Therefore, we decided to tilt our product strategy towards female users, focusing on functions and content recommendations for women's fitness, fat loss, and beauty. The overall next-day retention rate of the product increased by nearly 100% compared to before.

Similarly, I recently served an internal client of Tencent. They developed a new product intended for young people, but found that the age distribution of its users was mostly teenagers and the elderly:

This happened to be related to their user channels . They originally had a product for teenagers and the elderly. In order to bring the first batch of users to the product, they directly diverted users from the old product, but found that they were not the target users of the product.

Key data - retention rate

When the current users meet the characteristics of the target audience, the core focus is on indicators such as the retention rate, usage time/frequency, and user stickiness of these users. Here we will discuss the retention rate in detail.

There are many dimensions of retention rate (7 days, two weeks, 30 days, etc.). Choose according to the characteristics of the product. If the product itself meets niche and low-frequency needs, the retention rate should be two weeks or even 30 days. A high retention rate means that users recognize the value of the product and become dependent on it. Generally speaking, the hypothesis can be verified. Usually a retention rate below 20% is a more dangerous signal.

This article introduces a data-driven leading indicator model that can guide product design by finding leading indicators, thereby improving retention rates. Let’s first look at the definition of leading indicators. Leading indicators refer to a product behavior of new users in the early stages of using a product. There is a very high linear correlation between this indicator and the user retention rate indicator, which can predict whether the user will remain in the product.

Using the formula I summarized to describe it, it is roughly as follows:

Positive prediction probability (%): indicates that if a user performs this behavior, the possibility of the user remaining active can be predicted.

Negative prediction probability (%) : If the user does not perform the behavior, it can be predicted that the user will not remain active.

Finally, the credibility of the leading indicator = positive prediction probability x negative prediction probability . Let's look at the case directly.

Case

Take the previous forum social app as an example, and assume that "users add more than 7 friends within 10 days before registration" is a leading indicator, then we calculate a set of data:

Among them, if the user adds more than 7 friends in the first 10 days, the possibility of their retention in 30 days is 99%; if the user adds less than 7 friends, the possibility of their not retaining in 30 days (loss) is 95%, and the credibility of the comprehensive indicator is 0.9405.

Similarly, calculate the credibility of the following two leading indicators:

Finally, we get the comparison:

The above data are just hypothetical data. In reality, we need to compare more than a dozen or even twenty behavioral indicators to find the behavior with the highest credibility.

The first item in this model, "a new user adds more than 7 friends within 10 days of registration," is a classic Facebook "aha moment." The so-called "aha moments" are the moments when users realize the core value of the product, which is our "leading indicator."

(Screenshots of Facebook and Instagram recommended friends)

In addition, leading indicators should meet the following conditions:

2. Rapid Growth Stage

After the initial stage of product polishing, the product has a good retention rate, and now the product begins to enter a period of spontaneous growth. During the product stage of spontaneous growth, it is still necessary to pay attention to data such as user retention, user time, and changes in user portraits , but the focus can be on the management of the entire user life cycle, focusing on the entire funnel analysis from new user growth, activation, triggering "aha moments" to stable and active product users.

New user growth and activation

There are generally two ways to increase and activate new users. The first is to build the virality coefficient of the product and let the product grow spontaneously. The book "Lean Operations Data Analysis" mentions several interesting categories of user virality:

Native virality, which is the way to attract new users through the app’s own invite-friend function;

Word-of-mouth virality , that is, through word-of-mouth communication, users actively become new users through search engines ;

Artificial virality means encouraging users to invite others through artificial intervention, such as incentives such as rewarded invitations.

The indicator we focus on here is called the "viral coefficient". Students who are interested can learn more about it on their own.

New user download -> activation -> 'Aha Moments' -> product stable and active

After the product begins to enter the spontaneous growth period, it is necessary to pay attention to the user life cycle from new users to active users (after retention) to core users, and refine and detail the key indicators of each process.

Case

Taking the previous forum social APP as an example, when a new user enters the product, they will see a welcome page (as shown in the lower left picture). After registering and logging in, they will see the product homepage (as shown in the lower right picture). Most apps have a similar process:

The process from a new user entering the App welcome page to eventually becoming a core user is roughly as follows: New user (exploring and discovering the value of the product) -> spectator (gradually recognizing the value of the product and having a certain sense of participation) -> producer (recognizing the value of the product and actively participating):

According to the popular classification, the product life cycle (PLC) is divided into the start-up phase, growth phase, maturity phase, and decline phase . In each phase of the product, the weight and focus of data analysis are different. Let's talk about it by phase and case study:

At this point, the indicators of user behavior at each stage are broken down:

New Users & Discoverers:

Welcome page bounce rate

New user registration rate

New user onboarding process conversion rate

Initial feed page bounce rate

Search result conversion rate

Push permission activation rate

Bystanders (passers-by):

The average number of sections followed by each user

The average number of users each user follows

Average number of likes/shares per active user

Number of feed card displays

Feed card clicks

· Click-through rate of subscription content push

Content Producers:

Average number of posts per active user

Average number of photos and videos posted by each active user

Average time each user spends in the forum

· Distribution of active users’ behaviors in the forum

The refined breakdown of behavioral indicators in the early and middle stages of the user life cycle helps the product to continuously refine details during the rapid growth period of the product, and continuously improves the user experience from new users to core users . At the same time, after the data at each node is improved and stabilized, the product operations staff will start to carry out various promotions and publicity campaigns to expand the scale and occupy the market.

3. Maturity

With the rapid growth of users and continuous improvement of products, the focus of data operations begins to shift from the first half of the user life cycle (attraction, activation, retention) to the second half (churn, return) around the time the product enters the mature stage.

Here is a data template Daily Net Change (applied from John Egan@Pinterest) that is used during the growth and maturity stages. Different from focusing only on DAU and MAU data, focusing only on the increase or decrease in the number of active users is often just pleasing yourself. This model can help you intuitively observe the factors that drive user growth or changes in the user base, and use a graph to show the product's new additions, return traffic, and retention.

Net Change = New Users + Returning Users – Lost Users.

New users means how many new users joined that day

Returning users refer to how many old users have not used the app for 28 consecutive days and started using it again today.

Lost users refer to the number of existing users who last used the app 28 days ago.

Loss and Return

In the process of paying attention to churn and return, the data will reveal a change in the current user base. For a specific analysis of the reasons for churn, please refer to the following process:

The core idea is to determine the cause of user loss through qualitative follow-up + data verification as the main means, change the product operation strategy to prevent user loss or bring back users, and promote return.

In addition, for some stable delivery channels, common improvement methods may have limited effect on improving conversions. In this case, more refined channel analysis can be performed to optimize and improve ROI:

Examples:

Improve ROI

4. Recession

Eventually, the product enters the decline stage. Generally, there are two ways to do this before entering the decline stage:

1. Scale

This often occurs in the retail industry. For example, if a massage and health care store is opened and receives positive reviews within a certain range, a chain franchise model can be started when the product is mature. By rapidly and widely expanding the market, a brand effect can be formed to form a barrier, and the risk of recession can be resisted.

2. Ecological

When a product is growing or close to perfection, a single product may easily have problems with overly vertical demand and users being unable to form dependence. New products with synergistic capabilities can be developed to build a complete product ecosystem, so that users who cannot be satisfied or have lost interest in the current product can be directed to the new product and become new users of the new product. At the same time, users of the new product can also be directed back to the old product on the new product, forming a mutually dependent chain between products, and ultimately users can circulate effectively to form an ecosystem.

The author of this article @黄岳浩 is compiled and published by (Qinggua Media). Please indicate the author information and source when reprinting!

Product promotion services: APP promotion services, advertising platform, Longyou Games