How to build a complete growth model?

Today we will talk about three aspects: what is a growth model, the benefits of a growth model, and the steps to build a growth model.

01 What is a growth model

In a nutshell, the growth model is the result of breaking down the North Star indicator.

02 The benefits of building a growth model

1. Facilitate the search for growth ideas

Take the growth model of NetEase Cloud Music as an example.

NetEase Cloud's North Star Indicator is the total daily listening time. The total daily listening time = the total number of daily listening users * the average daily listening time of users. The total daily listening users can be broken down into the number of new listening users per day + the average number of daily listening users of old users...; through this step-by-step breakdown, the growth model is established.

For example, the daily app download volume is composed of data from channels A, B, C, and D. If we find that a certain channel has high investment costs but low download volumes, then we should focus on optimizing this channel.

For example, if we find that the proportion of first-time listeners on NetEase Cloud Music is too low, we should analyze the reasons and focus on optimization.

2. You can predict the impact of segmentation metrics on the North Star metric

Once the model is established, improvements in the underlying indicators will indirectly affect the value of the North Star indicator.

Assuming we increase the proportion of continuous listening from 5% to 10%, the average daily number of old users listening to music will increase from 15,000 to 30,000 users (300,000 existing users * 10% continuous listening).

If our boss asks us to increase the number of new music users from 5,000 to 10,000 per day, we can make a simple prediction on how much we need to increase the number of app downloads, first-time visits, and the proportion of first-time music users to meet our boss's requirements.

03 Steps to construct a growth model

1. Define your North Star Metric

First, we need to identify the North Star Metric for our product. Friends who don’t understand the North Star indicator can refer to my previous article, and I will not elaborate on it here. Only with the North Star indicator can we build our growth model around it. Here we define NetEase Cloud Music’s North Star Indicator as “Total Listening Time”

2. Draw the user's core conversion path

The so-called user core conversion path records the main steps that users go through from knowing nothing about a product to experiencing the core value of the product.

The core conversion path for NetEase Cloud products is: download the app → visit the app for the first time → listen to music for the first time → continue listening to music

3. Assembling the Growth Model

In the field of growth, there are about four growth models, namely AARRR model, full-chain funnel model, factor decomposition model, and full quantitative model. We will introduce them one by one below.

1) AARRR model

The AARRR model is like a funnel, from top to bottom: acquisition, activation, retention, monetization, and recommendation.

Acquisition refers to the channels through which you acquire users and how many users you acquire;

Activation refers to whether users have experienced the core value of the product and whether they have formed a positive interaction with the product;

Retention refers to the means you use to keep users and whether users will continue to use your product;

Monetization refers to how you monetize and how much profit you can get;

Recommendation refers to how to get users to spread your product spontaneously, and the majority of users are willing to spread it.

Taking NetEase Cloud Music as an example, its North Star indicator is the total listening time.

• Mainly acquire users through paid promotion and brand marketing;
• Listening to a song for the first time is an indicator of NetEase Cloud activation;
• Continuous listening to songs is its retention indicator;
• It mainly obtains revenue through value-added services (collecting membership fees), e-commerce, advertising, games, etc.;

Users generally recommend products by sharing music; the core indicators of acquisition, activation, retention, monetization, and recommendation are found, and the entire growth model is established.

The AARRR model is the simplest of the four growth models. For companies that are new to the growth field, this model is suitable as the first version.

The AARRR model is easy to build and can provide a general understanding of the factors that affect the North Star indicator; however, the disadvantage is that this model does not explain the relationship between the various factors, and it is a qualitative model that can only roughly estimate the data.

2) Full chain funnel model

The model that can use a multiplication formula to express the relationship with the North Star indicator is the full-chain funnel model. This model can not only decompose the factors that affect growth, but also find the corresponding segmentation indicators and values, and express their relationship with a simplified formula.

Let’s take NetEase Cloud Music as an example. The total listening time = app download volume * first-time visit ratio * first-time listening ratio * continuous listening ratio * average listening time.

3) Factorization Model

The factor decomposition model is an extension of the full-chain funnel model. The factor model not only has multiplication, but also addition, which can expand the segmented indicators in more detail. However, compared with the full-chain funnel model, it is more difficult to build and requires more data.

Let’s take NetEase Cloud Music as an example. NetEase Cloud’s North Star Indicator is the total daily listening time. The total daily listening time = total daily listening users * average daily listening time of users. The total daily listening users can be broken down into daily new listening users + daily average number of old users listening to music… By breaking down the process step by step, the growth model is established. This growth model makes it easier for us to identify growth opportunities.

Not only that, we can also predict how much impact the improvement of a certain indicator will have on the indicators at the previous level. For example, if the percentage of first-time users increases from 50% to 60%, the number of new users per day will increase from 5,000 to 5,940 (30,000*60%*33%).

4) Fully quantitative model

The so-called fully quantitative model refers to decomposing the factors that affect growth and the corresponding segmented indicators, and combining all indicators in Excel to calculate the North Star indicator.

The fully quantitative model is the most sophisticated growth model that can not only observe historical trends and predict an array of future North Star indicators, but also conduct hypothetical analysis and quantify the impact of changes in different indicators on the North Star indicator. However, this model is time-consuming and laborious to assemble and maintain.

Let’s take NetEase Cloud Music as an example. Let’s assume that the North Star Indicator of NetEase Cloud Music is “monthly active users”. To build a fully quantitative model, you need to create at least three worksheets in Excel. (The following data is virtual, for the convenience of calculation, only the first 3 months are shown)

The first worksheet is the number of new active users per month. Here you need to list all the channels in Excel, such as brand marketing, paid promotion, and user recommendations.

Secondly, list the number of downloads brought by each channel each month. For example, the number of users brought by the brand marketing channel in the first month is 1,000, 1,500 in the second month, and 2,000 in the third month.

Then calculate the monthly new downloads. For example, the monthly new downloads in January are 1,000 from brand marketing, 2,000 from paid promotion, and 3,000 from user recommendations.

Next, write out the activation ratio for each month, assuming that the first month is 50%, the second month is 60%, and the third month is 70%.

Finally, calculate the monthly new active users by multiplying the monthly new downloads by activation rate.

The second worksheet is the user retention rate for each month from 1 to 12 months. Here we assume that the retention rates in the first three months are 30%, 20%, and 10% respectively.

The third worksheet calculates the number of monthly active users based on the first two worksheets.

• The number of active users in January is 3,000, because the number of new active users in the first month (Table 1) is 3,000.
• The number of active users in February was 4,800: this included 3,000 new active users in the first month, of which only 30%, or 900, remained in the first month; and also included 3,900 new active users in the second month.
• The number of active users in March was 7,330, including: only 20% (3,000*20%=600) of the 3,000 users in the first month stayed in the second month, and 780 of the 3,900 users in the second month stayed (3,900*20%); another 5,950 active users were added in March.

In this model, we only show data from 1-3 months with very few variables. This is just to make it easier for everyone to understand. In practice, it will be more complicated. We do not recommend that teams that are new to growth adopt a fully quantitative model immediately.

So far, I have finished talking about how to build a growth model. I would like to wish my friends a happy Chinese New Year and see you next year.

Author: Fu Xiaohu (Little Tiger)

Source: Little Tiger Speak