Revelation on user growth in 2020!

At a time when traffic is becoming increasingly scarce, we must not only find growth clues through growth experiments, but also find the best growth methods through experimental verification, and improve growth efficiency through tool chain and process standardization, which can ultimately improve the efficiency and stability of the scale growth process.

There were many growth cases in 2019. I think there are three that deserve more attention:

• The first one: Tmall’s Double 11 transaction volume reached 268.4 billion, setting a new record;
• The second one: Luckin Coffee, which went public in April 2019. It took only 18 months from its founding to its listing. According to the Q3 financial report released in November, the company had 3,680 stores, 30.7 million cumulative transaction users, and a profit of 186 million from store operations.
• The third one: GSX, which went public in June 2019. Its Q3 financial report showed that GSX had a net income of RMB 550 million, a year-on-year increase of 419.5%. It is known as "the first Chinese K12 online education company with large-scale profitability."

This article wants to discuss that behind these three cases, they all point to the same growth trend: the tool chain and standardized practices behind achieving scale growth.

What is Growth Tool Chain and Process Standardization

We all know that the Taobao and Tmall apps have implemented customized product recommendations for each individual. After you search for something, activities or brands related to the product will appear in the recommended position on the homepage. During the Double Eleven period, hundreds of millions of images were generated. Where did these images come from?

Alibaba has a Luban system, which provides a large number of template materials and can automatically design posters, greatly saving manpower. Isn’t it amazing?

What about Luckin Coffee? From its founding to now, it has opened 3,680 stores in two years. How did it choose store locations and conduct production management? How to ensure the quality of coffee made by different people? In a recent article, it was said that the completion time for each Luckin Coffee order is set at 2 minutes. How can this be guaranteed?

Luckin Coffee relies on an order management system + automatic coffee machine + monitoring system. The order system can automatically distribute orders to stores. When a store has too many orders, it will be distributed to nearby stores. At the same time, the system data can also provide feedback on which area is suitable for new stores. The automatic coffee machine ensures the uniformity of coffee taste, and the monitoring system ensures the standardization of operating procedures.

Finally, let’s take a look at GSX. Peers in the education industry generally believe that GSX’s high profitability is due to the low-cost customer acquisition and high average order value conversion brought about by the “private domain traffic” model.

The sales conversion model of Who's Teaching can be summarized as attracting traffic and building groups by offering trial classes, information packages or fission methods, continuously converting and selling courses during the community service process, and following up with course sales after selling courses. After course sales, users can still make secondary conversions and repurchases in the group.

Many educational institutions are following this process, but the reason for Who’s Teaching’s success is that there are standardized operating methods in every link. For example, a template has been formed for the language used from the time users join the group to the time they convert.

The so-called tool chain and standardized practices behind large-scale growth refer to the development of tools or establishment of standardized execution processes for the operating methods of each link in the growth model after finding an effective growth method, so as to improve growth efficiency and expand the scale of growth.

To implement tool chains and standardized practices, we need to build a growth architecture around an effective growth model.

Specifically, the model is as follows:

Under this model, there are four layers in total, which are also the process of a growth team from building a growth model to conducting growth experiments and then to scale growth:

First layer, growth model layer

This step requires building a growth model based on the North Star Indicator and designing indicators for each link of the growth model. This is a process of breaking down the North Star Indicator into executable indicators. Executable indicators are one of the criteria used to measure whether the growth experiment is effective.

The second layer, growth hypothesis layer

Propose growth hypotheses based on growth models and actionable metrics through data analysis.

The third layer, growth experiment layer

The growth experiment is divided into four major steps, namely product development, user introduction, user engagement, and revenue conversion. Each step includes many small specific actions, such as the impact of operational posters, copywriting, channels, and promotion timing on the results, as well as the impact of product UI, button position, and guiding copywriting on the results.

The fourth layer: growth tools and process standardization

The role of this layer is to improve the efficiency of experiment launch and result verification through various standardized tools and processes. The relationship between this layer and the experimental layer is mutual guidance. The experiments conducted by the experimental layer will fill the verification results into the tool layer, and at the same time, the needs of the experiments will drive the production of tools.

Under this architecture, standardization of growth tools and processes is the underlying foundation for us to achieve scale growth. Standardization of tools and processes can effectively speed up the efficiency of growth experiments and guide execution actions under large-scale growth.

The role of growth tool chain and process standardization

We use two cases to illustrate the role of growth tool chain and process standardization:

Case 1: How Xianyu improves growth efficiency from the perspective of software development

Xianyu is a second-hand trading platform that has conducted many growth experiments around user retention and ordering. Through data analysis, we found that if users repeatedly browse the same product but do not place an order, the conversion rate of the order will increase if there is a red envelope.

Therefore, Xianyu’s growth team proposed an experimental plan. If a user repeatedly enters the product details page but does not place an order, a push notification will be sent to remind the user to receive a red envelope.

This experiment needs to test three parameters. The first is how many times "repeatedly", the second is the amount of the red envelope, and the third is the time point of sending the push.

We conducted many experiments around these three parameters. However, in the beginning, in terms of technical development, we wrote code on a case-by-case basis. It might take three weeks from development to function launch, and after launch, we had to make continuous adjustments based on the data.

To improve experimental efficiency, Xianyu established a rule engine based on event streams:

This rule engine is applied to the above case, and the code implementation is as follows:

With such a rule engine, we can efficiently solve growth experiments with similar needs, and all actions generated by related events can be solved by this rule engine.

This is a rule engine in a laboratory environment. In actual business scenarios, due to the limitations of business rules, it is found that this rule engine cannot implement backtracking, which will cause problems in actual use.

Xianyu's technical team proposed a new solution and continued to optimize it, eventually implementing a recommendation method based on the EPL engine. During scale growth, it can support demands such as "in the rental business, if a user browses four different listings, further growth actions will be generated."

Case 2: How Baby Play English improves growth efficiency from the perspective of process standardization

Baby Play English is an educational institution that has built a growth model based on social marketing methods using private domain traffic. Through small courses with low customer unit price, users are attracted to the community for 10-day course learning and service, and large-course marketing conversion is carried out during the service process.

In terms of growth execution, the community marketing team of Baby Play English is divided into four groups: event operation, content operation, training group, and product operation. Event operation and content operation conduct growth experiments around each link of the community conversion chain to find the best activities and copywriting. The training group is responsible for training part-time sales and standardizing the community management actions of part-time sales.

Baby Play English has built a sophisticated social marketing tool around the 10-day period when users participate in the trial classes. Since it is a part-time sales model, more support needs to be given to sales. How precise is it? Let's take a look at the specifications of the group building process:

(1) Post to Moments before creating a group

The 24th training camp of "Baby Plays English" has started warming up. At 15:00 today, the secretary will invite you to join the warm-up group before the class. Please agree to the secretary's invitation to join the group in time! This week there will be three days of pre-class warm-up for English enlightenment and a lecture on English enlightenment by Tsinghua scholar Lan Xin. Hurry up and invite your friends to join the group to listen. Just scan the QR code to join the group!

Illustrations: 3 course schedules + 3 course introduction pictures + welfare lecture picture + lecturer introduction picture + recruitment picture, put your own QR code in picture 789!

(2) Private message before creating a group:

Hello, parents. The 24th intensive English training camp for your baby that you have signed up for has started today. Please accept the secretary’s invitation to join the group in time!

How to establish a growth tool chain and process standardization

The core of large-scale growth is to establish a tool chain and process standardization. So how to establish a tool chain and standardized process?

Step 1: Sorting out the overall process of scale growth

Currently, there are two schools of thought on growth:

• One is the "inspirational group", who rely on hacker-like creativity to achieve growth. The inspirational group has no rules to follow, they just think of whatever they think of.
• The other type is the "experimentalists". The overall process is divided into formulating North Star indicators - breaking down North Star indicators - formulating executable indicators - finding growth clues through data analysis - proposing growth hypotheses based on growth clues - conducting growth experiments on growth hypotheses - verifying and correcting based on experimental results - using experimental methods on a larger scale to achieve growth.

Compared with the inspirational growth method, the experimental growth method is more traceable, scientific and reasonable. As for the growth process of experimentalists, we can divide this process into three major stages: implementation of growth experiments - verification of experimental results - large-scale growth.

For example, a used car platform has an annual budget of more than 100 million yuan for elevator advertising. In order to achieve better advertising effects, they conducted separate experiments on the advertising areas, posters, copywriting, and the process of poster-guided registration.

By placing posters in different communities and using the number of people who scanned the QR code on the posters and the number of new visitors in the area during the same period as evaluation dimensions, we can screen out more accurate user channels and extract user portraits. Next, we put posters of different styles but the same copy in different buildings in the same community to verify the poster with better effect. Then we verified the posters with different copy again, and finally found the best placement area and poster style, and increased the placement efforts.

Focusing on the three links of experiment implementation, verification of results and large-scale growth, we need to find the execution process under each link, sort out the execution process and abstract the key links in this process, and create tool chains and process standardization for the key links.

Generally speaking, these three links include these tools and standardized processes:

In the past year, I have surveyed the growth teams of nearly a hundred companies for work reasons. Many teams are trying to promote the standardization of tool chains and processes, but they also encounter many problems.

Common problems include the system’s inability to address personalized needs, standardized processes being developed but no one using them when they are rolled out on a large scale, and the process of developing tools and processes taking up very high manpower costs.

The process of growth experimentation itself is the process of trying to find new ways of growth. This process itself is very complicated. So how do we develop reliable tools and formulate standardized processes?

I’ve found several common traits among growth teams that have successfully improved performance:

1. Development of tools and formulation of standardized processes are only started after successful experimental verification

This contains two meanings:

• One meaning is that after the experiment is successful, when promoting growth methods to more people, there needs to be standardization. For example, a clear channel portrait should be given to the channel team, and several activity templates and activity H5s should be given to students in activity operations.
• The second meaning is that if the same requirement is encountered multiple times during the experiment, tool development and standardization are needed to improve efficiency.

2. MVP thinking is also needed when developing tools and formulating standardized processes

Tools and standardized processes are not achieved overnight. The demand for tools and standardized processes comes from the people who actually use them. The person in charge of specific work in the experimental environment is the best person to organize the tool requirements and standardized processes. After development, it is also necessary to test whether the tools and processes are easy to use in a small range and continuously iterate.

3. Developing tools and establishing standardized processes starts with defining the problem and clarifying the boundaries

Each tool and process only solves a specific problem. There should be reasonable expectations and definitions on what problem to solve and to what extent. Tools and processes can only solve problems with how things are done, but cannot solve problems with people. Don't mix people and things together.

For example, for a product whose main target users are Internet professionals, in order to attract new users, a hypothesis is proposed to attract traffic through case analysis articles of products/activities. The first problem to be solved becomes how to mass-produce such articles.

There are two problems that need to be solved here: one is to find people who can write articles; the other is to require these people to write articles that meet the requirements. How can this be standardized?

We must separate people from things. People are a process of screening. We can attract people under conditions of high temptation, but how to guide users to write articles that meet the requirements requires us to provide tools.

Next, they analyzed the most widely read product/activity case analysis articles on the market and found that these articles achieved three points: first, the logical structure of the article was clear; second, the case background was detailed; and third, the analysis viewpoints were sharp.

Can these 3 points be solved? The third point is that we cannot guarantee that everyone can come up with sharp opinions, so is this important? They conducted user interviews and found that for case analysis articles, 70% of users' attention was on the specific story of the case, and 30% was on the opinions. More users said that if they wanted to see opinions, they would read industry analysis and sharing by experts.

Then, it was easy to produce the tool. They created SOPs based on the different information focuses contained in product case analysis and activity case analysis articles, and standardized the article structure and required content.

4. Tools and processes are not guiding principles, but requirements for execution results and handles for the execution process.

How do you understand this sentence? The example of producing the case analysis article SOP above is also used to illustrate.

There are many ways to write articles, but most of them are principles, such as finding well-known products, providing case data, etc., but these are guiding principles. Effective tools and processes are requirements for results.

In the SOP for case analysis articles produced by this company, the first step requires the author to submit a topic application form. The topic template requires some key information to be stated. Taking the product case analysis article as an example, it includes: product launch time, number of users, latest version release time and functions, the content of the last 10 version updates, user behavior flowchart, user points rules, etc. At the same time, it provides channels for finding materials and analysis methods for each content.

According to the content specified in this template, the author needs to analyze the relevant content of this product. If this information is valid, it can ensure that the content of this article is detailed. At the same time, channels are provided for the information sources of each content. For content that does not have ready-made information, such as the "User Behavior Flowchart", method instructions are provided, which provide a starting point for the author.

The reason why many teams have poor results when developing tool and process standardization is that they do not design in-depth into the actual user scenarios. For many non-standard execution processes, it is necessary to participate in the front-line work and summarize.

Let’s take a consulting company as an example. Consulting companies generally need to do a lot of market research, some of which are questionnaire surveys and some are interview-based surveys. So how do you ensure the effectiveness of the research?

We will find that the research template of consulting companies will generally ask you what the process is for doing this thing, and then ask you what you did in each step of this process, and finally ask you what is the time-consuming thing in the process of doing this thing.

With this set of questions, we can get the complete process of a person doing something as well as the pain points and difficulties. Therefore, within the consulting company, new researchers will be required to prepare questions according to these three levels.

What is needed for large-scale growth in tool and process standardization is not just a set of high-end systems. Of course, data analysis tools and collaborative office tools are needed, but what is more needed is to provide tools that go deep into the execution process.

Among the many students who are doing events that I have interviewed, the lack of technical development resources is the biggest pain point. Sometimes, in order to follow the hot spots, the hot spots have passed and the events have not been completed. Some companies have an activity middle platform, which provides H5 templates for common activities such as raffles, bargaining, and treasure collection. Operators only need to change the activity background and rules, which greatly improves efficiency.

Why is it said that growth requires tooling and standardization?

Over the past year or so, I have surveyed more than 100 students working in growth. During the survey, I found that if growth is to be implemented, there will be three typical pain points.

Many students who are doing growth in startups have almost no growth team. They are just one person working alone and still rely on selling "ideas" to achieve growth. The company lacks basic tools such as data analysis, or the data statistics are very rough and cannot meet the needs of refined analysis.

For example, in a certain overseas e-commerce company, there is a product manager who is specifically responsible for growth. His main job is to design functions such as check-in and sharing to improve retention.

In some medium-sized companies, the core work of the growth team is to build growth models, conduct data analysis to find growth clues, and find the optimal growth method through A/B testing. The problem these teams encounter is that their efficiency is relatively low. Usually a round of growth experiments takes 2-3 months, and what restricts the efficiency of the experiment is the execution process.

In addition to the above two types, another typical pain point is that after verifying the experimental idea, there is no way to expand on a large scale and it is highly dependent on a certain resource.

Taking the "private domain traffic" model as an example, the private domain traffic model is to attract users to WeChat groups through external delivery or fission, and guide users to make purchases or downloads through group operations. If this model is implemented in the experimental environment, at least three issues need to be considered in order to achieve scale growth:

• The first one is whether a channel quality assessment mechanism has been established and how to ensure the conversion rate of the delivery channel is stable;
• Second: whether the community service process and tools have been established, and how to ensure the operational effectiveness of different operators;
• The third one is whether a subsequent service process has been built to extend the user life cycle of community conversion traffic.

The role of the growth tool chain and process standardization is to support the implementation of growth experiments and the large-scale execution of growth strategies from the perspective of internal management, and to continuously find the best solution through experimental verification. This is actually somewhat similar to the role of the "middle platform".

At a time when traffic is becoming increasingly scarce, we must not only find growth clues through growth experiments, but also find the best growth methods through experimental verification, and improve growth efficiency through tool chain and process standardization, which can ultimately improve the efficiency and stability of the scale growth process.

Author: Zhang Xiaohuai

Source: Agile Growth Research Lab