How can APP use growth hacking methods to achieve 400,000 growth?

This article uses the growth hacker methodology and follows the data-driven operational thinking principles to conduct an overall review and analysis of a home furnishing APP. The author aims to use data analysis tools to dissect an Internet+ product and find out the driving logic of the kernel layer, presentation layer and business layer through the performance of the data layer. Let this product with the characteristics of going from 0 to 1 help us find the common characteristics of explosive growth of entrepreneurial products in a rational, scientific and rigorous case analysis form, and strengthen and remind us of the thinking principles of data-driven operations.

1. Product and User Overview

1. Product Overview

Product forms: Master side (Android version, IOS version), merchant/personal side (Android version, IOS version, hereinafter referred to as merchant side). Other product forms are not within the scope of this analysis.

Time period: July 1, 2017 - April 1, 2019

Data trends during the period: cumulative users 0-420,000, active users 0-15,000

Product flow chart:

2. User Profile

Understanding your user attributes, including basic attributes and device attributes, will help us develop product and operation strategies based on user data.

I won’t talk about the user portrait attributes that come with the system. I will only mention some valuable information that I have seen from the data layer:

(1) The cumulative user ratio of the master-side APP Android and IOS is 7:3, while that of the merchant side is 3:5.

Judging from this, most of the masters working in the home aftermarket use Android phones and their income is not high. For the master side, Android should be given priority in selecting the order of product iteration versions, Push push, channel distribution and promotion, and operational activity design.

In a fission activity of this product, we encountered a long picture QR code of h5 under iOS, which can be directly recognized by long pressing to complete the sharing fission, while the Android system must re-make a matching picture before it can be directly recognized. The opposite is true on the merchant side.

(2) Geographical distribution: As an O2O home furnishing product, understanding the geographical distribution of users is very important both for overall operational decisions and refined operations.

What was quite surprising during this review was that Hubei did not make it into the top five, whether on the master side or the merchant side. This also indirectly shows that the current platform's North Star indicator (explained later) is still mainly completed online.

3. User Conduct

Frequency of use, page visits, page paths and sources, etc., are the most basic dimensions for understanding user behavior.

Because this is an overview, we only analyzed the average single usage duration and daily startup times:

Average single usage time of the master terminal

Average single usage time of the merchant end

Average daily startup times of the master terminal

Average daily startup times of merchant terminals

Conclusion: There is a serious polarization in the average single-time usage time on the master side. It surges during the peak season when the number of orders is large or there are operational activities. It shows a slow downward trend overall, but the number of daily starts has increased significantly. This is related to the addition of operating methods such as master check-in, daily lottery, and points mall to product iterations, but how to increase the total usage time through master community and other content operations to complete the transition from tool to platform product is a problem that needs to be addressed in the next stage.

On the merchant side, the average single usage duration is relatively stable, and the sensitivity to routine operational activities is low (the fun of merchant-side operations will be introduced later). The overall trend is downward, indicating that merchants use the product with a strong purpose, the product UE is gradually being optimized, the order placement process is more simplified, and the user operation time is shortened. The number of daily startups also fluctuates in a box shape. The next step is how to increase the number of daily startups of merchants and enhance brand loyalty. This is a problem that needs to be addressed in the next stage.

It should be noted that when pulling the average daily usage time, the master end did not pull it according to the original cycle from July 1, 2017. This is because when I started pulling data according to this time period, I found that the historical peak of the entire cycle was between September 19 and September 30, 2017, which was much higher than other time periods. Moreover, both the average daily usage time and the number of daily startups are abnormal.

I conducted a focused analysis of this abnormal data, and then compared it with the changes in the cumulative number of users and inquired about the operation situation at the time, and finally solved the mystery:

The cumulative number of users tells us the truth: it turned out that in September 2017, when the APP product had just been launched and switched from the previous H5 format, there were very few initial users, only a few dozen, who came from the field promotion of operations staff. At the same time, in order to attract the participation of seed users, subsidies were used to stimulate the masters' quotations. As a result, the only dozens of masters at the time competed for quotation subsidies, resulting in a large amount of usage time and opening times.

The regression to the mean on September 30 was because, first of all, the total number of masters increased significantly in terms of the cumulative number of users, with the number increasing several times or even ten times each day in the following days, which lowered the average usage time and various indicators. At the same time, the platform stopped the quotation subsidy activities.

Therefore, when we do data analysis in our daily life, we should conduct special and detailed analysis and processing of abnormal points, otherwise it will lead to deviations in the overall data analysis.

2. Growth Hacker Method: The Best Points, North Star Indicators, and Growth Equation

First of all, let me explain that this article only uses the thinking model of growth hacker methodology to do a system review. The reason for choosing growth hacker is that this entrepreneurial product, which is not from a large company or a wealthy family, has been striving to achieve explosive growth at low cost along the way; and the main method of growth hacker is based on data collection and analysis, a process of quickly designing experiments and verifying conclusions; using super technical implementation capabilities to efficiently put operational ideas into practice, quickly trial and error, and iterate, which is also the purpose of this article to strengthen and remind data-driven operational thinking principles.

I believe that growth hacking is not just a methodology, but more about the mindset, behavioral code, management model and staffing that guide product operations. Today, when the traffic dividend period has ended, growth hacker thinking should be one of the core thinking that product operators must have.

The prerequisite for applying the growth hacking methodology is that the product is good enough - that is, the P/MF product has sufficient market match, the user base is large enough and users can be acquired quickly. In this regard, this home furnishing product can meet the basic requirements.

Next, I will use the growth hacker methodology combined with data-driven operational thinking to review this product:

1. User delight (Ahaha moment)

(1) Merchant side

I think the client's pleasure lies in two different moments, and the main pleasure is: in the initial stage, within 3 minutes of placing the order, multiple masters will quote prices and the fees are lower than offline (including the subsidies given by the platform).

Another small bonus is that after hiring a technician through the platform, you can rest assured without having to worry about after-sales issues - that is, when problems arise during the installation phase, the platform will intervene to help resolve the after-sales issues.

(2) Master side

I think there is only one cool point: after using this APP, you can receive new orders or make more money every day.

This experience is similar to that of a taxi-hailing app. When we call for a taxi in the snow, the best thing is that as soon as we place the order, a safe and reasonably priced driver will come to pick us up. The driver’s reminder – “You have a new order~” is enough to make him feel happy.

2. North Star Indicator

Data layer: active users, that is, active.

Business layer: order completion volume, i.e. conversion. Of course, the proportion of the North Star indicator changes in different product cycles. During the introduction period from 2017 to April 2018, the North Star indicator was mainly the number of active users. After entering the development period, it was mainly the number of orders issued and completion status. The team KPI indicators and personnel assessments at each stage are also guided and divided around this core North Star indicator.

3. Growth Equation

Number of active merchants x average order volume x number of active masters x average order volume x average unit price x order completion rate = order completion volume growth

4. AARRR Model

This is also the main workflow model in growth hacking, which is actually what we often call "Acquisition", "Activation", "Retention", "Referral" and "Revenue".

I have written a lot about these in my previous articles, so I will not expand on them here. The whole process is shown in the figure:

3. AARRR Model: Attracting New Customers

New customers during the period

1. Highest point

First, let’s look at the new customer acquisition on the master’s side, as shown in the figure above: except for a few low points during the Spring Festival and off-season holidays, there are several noteworthy high points for new customer acquisition.

What caught my attention the most and what I spent a lot of time researching was April 18, 2018. On that day, both Android and iOS reached their all-time highs. By checking the channel source and time period details, we can see that Android almost all comes from the default channel, and the time is concentrated within an hour around 12 o'clock (Apple only has AppStore).

From this, I can almost rule out the possibility that the "surge" on that day was caused by the advertising or the traffic explosion on a certain online platform.

At the same time, I also specifically asked the product manager to check the backend order status for the day, which was the final conversion target for the entire platform. There was no abnormality on that day:

From this, it can be seen that the most likely operation on that day was to deliver the default product package directly to users.

At first I thought it was a concentrated large-scale ground promotion, and an operations staff member also mentioned this when I asked him. Later I thought that in the user scenario of field promotion, it is unlikely that the time would grow so concentratedly.

It is a pity that the various channel packages were not activated at that time, and all came from the default channels, so it was impossible to distinguish the traffic channels of the APP.

From all the possibilities that I collected in April and on that day, I concluded that this special moment of attracting new customers was caused by the action of mass SMS activation on a batch of big data collected centrally. (The graphics and reasons on the merchant side are the same, so I won’t go into details.)

2. Second highest point

The second highest point was from November to the end of December 2018. This period was also the time when the anniversary celebration and Double 11 activities led to a significant increase in average daily orders and the launch of the product coupon center and gold coin mall.

The period of the Double 12 event is also the North Star indicator, and the number of orders reached a historical peak. Therefore, there is no need to elaborate on the reasons for attracting new customers at this second highest point. The development of various online activities, the superposition of traffic sources and the updating of product functions led to this second highest point.

3. The third highest point

The third highest point was from May to June 2018. This period is very interesting because according to historical data, this should be a low season. However, if we look at the product version update records, we can roughly understand what happened:

It turns out that in May and June, we launched the "Gold Medal Master", "Invitation Gifts", "Task System" and daily lucky draw features. I have written about their importance to operations in my previous articles.

Therefore, it can be concluded that the third highest point in attracting new customers has a strong relationship with product updates.

Moreover, I will also mention in the promotion part later that the highest peak of active IOS users on the master side was not the highest and second highest points mentioned just now, but on May 11th. This also verifies the saying that "whether users are active depends mainly on the product."

4. Analysis of new customer acquisition methods

In addition to the detailed analysis of the three major high points in the cycle, I also combined the growth hacker methodology, product life cycle theory and several major growth intervals on the graph for analysis.

This case can maintain an overall high growth trend. I think the following new customer acquisition methods are worth learning:

(1) Market matching of the overall copywriting

This is also one of the two matches that growth hackers must achieve first in order to expand the scale of customer acquisition, that is, the degree to which the description of the product's advantages can impress the target customers.

The most representative of this is Steve Jobs’ statement: “Put 1,000 songs in your pocket.” Similarly, in the growth period before April 2018, this case proposed "5 quotes in 3 minutes" and "If you want to find a master, go to ***" to the merchant side; and proposed "more rewards on the platform, more money for the master", "income doubling plan" and so on to the master side. These policies were intuitive and clear, and achieved valuable seed user growth in the introduction period.

(2) Matching degree between channels and products

That is, the effectiveness of the marketing channel you choose in promoting products to target users. You need to analyze the user's behavior type to select the corresponding customer acquisition channel and monitor the channel source results for screening. After this case entered the development stage, it also effectively monitored various channels, thereby discovering that SEO/SEM, third-party e-commerce, etc. later became the core channels for completing the North Star indicators.

During this process, we are constantly making new attempts and conducting optimization experiments.

(3) Design user invitations and viral loops

First, we identified the active seed users at a certain stage. After screening the user attributes through different dimensions, we put the potential "super users" in different periods at the top priority of user operations, using methods including: first-order discounts, recharge coupons, induced sharing, points redemption, inviting friends to get red envelopes, targeted event push, event invitations, game fission, etc.

In communication studies, K = the number of invitations each user sends to his friends * the conversion rate of those who receive the invitations into new users.

By combining marketing methods, when k>1, the user base will grow like a snowball. Imagine how Dropbox achieved dozens of times growth at that time? This is because of the use of the trick of inviting friends to enjoy larger capacity, including the use of hotmial's tail signature invitation and PayPal's case of giving away $10 for registration.

(4) Experiment, experiment, and experiment again until you find a trick that works best

Just as the main reason for the 4.18 high point mentioned above was the batch SMS activation of centralized procurement data, in the process of this case, there were countless similar operational actions that seemed insignificant or even low-level, including Airbnb's path to success, which was brought to the extreme through continuous optimization experiments.

Don’t think that some methods are outdated. The case of Airbnb has fully demonstrated that every possible means will be used to achieve high growth at low cost.

IV. AARRR model to promote activation

Activity of the master side during the period

Merchant activity during the period

1. Chef's side

It is similar to the analysis of attracting new users and will not be repeated here. What is interesting is that the peak point of the outbreak on May 11 mentioned above received enhanced verification at the active point.

Thanks to the launch of "Gold Medal Master", "Gifts for Invitations", "Task System" and daily lucky draw features in May and June, the number of active IOS users on the master side reached its highest peak. The activity of the entire master side has also experienced a process from small platform A in the introduction period, to small platform B in the transition period, and then to small platforms C and D in the current development period.

2. Merchant side

This time we will conduct a targeted analysis of the user behavior mentioned in the previous article. It is mentioned above that the average single usage time on the merchant side is relatively stable and less sensitive to routine operational activities. So what are the pleasures of merchant operations based on daily activity?

It turns out that the daily activity graph on the merchant side is very different from that on the master side, with an explosive growth peak at the end of November to December. The reason for this is the launch of the coupon center, Double 11, Double 12 activities, etc.

Therefore, from the data analysis here, we can once again confirm the merchant pleasure mentioned at the beginning: multiple masters will quote within 3 minutes of placing the order and the fee is lower than offline (including the subsidy given by the platform).

At this stage, the masters are already active enough, and with the simplification of the product ordering process, all that remains is to use the annual Double 11 and Double 12 as detonation points to provide merchants with coupons and promotions to reduce ordering costs, thereby creating an explosive growth in activity.

3. Activate conversion and churn funnels

Here I specifically made a conversion and loss funnel from the source, the download statistics of the channel to the final activity, which is worth studying by operations and marketing personnel:

Schematic diagram of the download data of each version of the product channel and the proportion of cumulative activated users of Umeng:

Activate conversion and churn funnels (Umeng statistics from July 1, 2017 to April 1, 2019)

5. Retention of AARRR Model

Shifu Android User Life Cycle Table

Merchant IOS user life cycle table

It is more intuitive to use the user life cycle chart in the user growth function of Umeng+ here. At the same time, I also made a simultaneous comparison with the 7/14/30 day retention rate, which is basically consistent with the conclusion shown in this graph.

At present, the overall product is still tool-oriented. Although it is gradually transitioning to a platform type by improving stickiness through functions such as community, points mall, academy, and question answering, the current proportion of silent users and lost users is relatively high.

(1) Master side

For the masters, attracting new users is a double-edged sword. While new users can be converted into users with high growth potential, half of them may also become users with high churn risk. Even when they reach the growth stage, a large proportion of them will become silent users depending on the situation. At the same time, once they are lost, the proportion of users with high recall potential is very small, and some may even uninstall the product directly.

It is understandable that the masters are more casual about such products and have little loyalty. It mainly depends on whether the platform can satisfy their pleasure. Once it fails, the possibility of returning again is very small.

(2) Merchant side

New users who are attracted more stably can be converted into users in the growth stage, and the user quality is relatively high. As long as they are used to using this product, they have the characteristics of high-value potential users. In addition, during the churn stage, if applied properly, the proportion of users with high recall potential is also very high.

It is understandable that merchants are more rational and purposeful about such products. They mainly look at whether the platform can satisfy their needs and is valuable to them in the early stages. They either make good use of it or wait until the right time to come back. It is very important to recall and re-enter the platform to recover lost customers.

(3) Recommendations

For silent and lost users on the merchant side, push notifications should be made in different groups, and copywriting that can satisfy the merchants should be adopted; on the master side, the effect of group push notifications is not good, and the focus should be on the construction of new channels and methods for attracting new users in the early stage and the product user experience.

Umeng+'s user life cycle table combined with group push to achieve targeted recall

6. Transformation of AARRR Model

The order growth curve and corresponding events in this case in 2018

The ultimate goal of operations is conversion (Revenue), and the funnel analysis method is commonly used.

The funnel model is used to analyze the loss and conversion at each step, and to analyze the changing trends and proportions of the number of users in different user groups from the beginning to the end of an event, so as to find optimization solutions. The funnel shows the final conversion rate and the conversion rate between each step, and is analyzed through trends, comparisons, and drill-down analysis. This method is widely used in the analysis of various key processes of the product.

As shown in the figure:

Demo analysis of the service event funnel on the master side

From the conversion rate of the first quotation → start of work → uploading of the finished drawings and final completion of the service, we can use such a funnel model to find out which link has the lowest conversion rate. At the same time, compare with the industry standard values. If they are not reached, analyze the specific reasons and then make targeted optimization and improvements.

At the same time, you can also do funnel analysis on the key events of the page. For example, if the conversion rate from the homepage to the order center is 80%, but the conversion rate from the order center to the final order is only 5%. Then, problems should be discovered by setting up step-by-step tracking points for the order center. For example, after entering the order center, there is a lack of reminders of the current step, and there are also problems with the UI design of the landing page and the UE of online image upload. After comparing with competing products and doing A/B testing, the final conversion rate increased several times.

In addition, there is no magic bullet for conversion. We can only do more user research, investigate user needs, and optimize service pricing based on the characteristics of each project. At the same time, in order to gradually increase the ARUP value, different strategies should be adopted for different user groups. Based on understanding user needs, we innovate and upgrade transformation products and services.

Other conversion methods besides the main conversion method

Summarize

Through this review of the home furnishing APP, I have revisited the essence of the growth hacker methodology and gained a sense of awe for data-driven operations.

All excellent product presentation layers must have carefully designed kernel and business layer driving logic inside. At the same time, an excellent data analysis tool gives us the eyes to discover, allowing us to glimpse the truth and gain insight into the mystery in a rational, scientific and rigorous way.

In the second half of the Internet era when traffic dividends disappear, we need the guidance of growth hacker methodology and follow the principles of data-driven operational thinking to achieve low-cost explosive growth from 0 to 1.

If one studies alone without friends, one will be ignorant and uninformed.

Author: Fu Rutao

Source: Lao Fu said operation