How to attract new users and increase growth?

Because there is no growth, growth is imminent!

According to the report of QuestMobile on China's mobile Internet, BAT users both exceed 1 billion, the total number of users and time spent have stagnated, and short videos have become the biggest winner. The scale of domestic mobile Internet industry users in social, video, e-commerce, financial management, travel and other fields has exceeded 500 million. However, the overall number of users and duration have also reached their peak. From 2019 to date, the number of active users has only increased by 2 million, and the average daily duration per person has only increased by 18 minutes.

Overall, the number of users of the three BAT series has exceeded 1 billion, and the four BATT (Toutiao) series accounts for 70% of the user time on the entire network. The competition for existing users has become the new normal; the user end and the sinking market have become the only high-growth breakthrough; the application end and short video time growth account for more than 60%, becoming the only winner.

So how to grow? What role can big data AI play in growth? This article will explain them one by one.

5 Steps to Mathematical Intelligence and Mental Growth

1. There are already too many festivals to count, so stop creating more.

Let's first roughly count the festivals that have invested heavily in advertising for product operations throughout the year. The following figure shows:

Users also get tired sometimes. There are so many festivals, and Alibaba can’t even handle them all, so how can it celebrate others? Relying on creating festivals to achieve growth seems a bit lackluster, so it is not recommended to create any more festivals.

2. Fission

People within the 5th ring road choose cost-effectiveness among expensive ones, while people outside the 5th ring road choose quality among cheap ones.

Fission is particularly suitable for users who are looking for a bargain. In other words, Pinduoduo relies on fission because it has found the right group of people, namely fourth- and fifth-tier cities and towns, so fission requires user group division.

User group division means that what is shared in the group that buys Porsche is Tiffany's love story, while what is shared in the group that buys Pinduoduo is the quality of Pinduoduo.

During the fission process, product operators may wish to understand the mindsets of users inside and outside the 5th ring: The demand within the 5th ring is to find cost-effective products among good ones, and the demand outside the 5th ring is to find good quality products among cheap ones.

Think about it, Pinduoduo sends you an activity saying that if you forward this link to a hundred people and ask them to help you bargain, you can get a product worth 50 yuan. What would you do within the 5th ring? Obviously not. Because your connection is more valuable than 50 yuan, and your time forwarding is more valuable than 50 yuan. So fission actually doesn't work for people like you, but it will work if he is someone outside the 5th ring.

It is more important to do a good job in user profiling and user segmentation and to deeply explore the needs of top and middle-level customers within the 5th ring road to increase the depth of usage. For example: What you need is high-quality and time-saving products and services, so what should be given to you is the Daily Fresh version of JD Plus, which is also a membership system. Fission can be used for those who are greedy for small profits.

3. Find the best sellers

Do hot-selling marketing through content and channels, only hot-selling products can attract new customers.

For example, retail platforms such as MissFresh promoted a product called "Three Boxes of Crayfish for RMB 19" on Weibo this summer, which had a better effect in attracting new customers.

4. Use hot products instead of red envelopes

Red envelopes can be used, but they are not a leverage, at least not on the front line.

Take myself as an example. I won’t buy something just because someone gives you a red envelope of 5 yuan. Obviously not. Therefore, if the main customer base of an app is in first- and second-tier cities, it should pursue more: when users have this demand, the first thing they think of is me, and that means it will be a hit.

5. Do e-commerce in niche areas (you can refer to the three tricks of Xinmeida)

Let me first introduce New Meituan’s three key strategies, namely: 1. clear store positioning; 2. attractive products; and 3. a closed evaluation loop.

No matter what kind of e-commerce you are doing, including platform-level e-commerce, vertical segmented e-commerce, new retail e-commerce, social e-commerce or short video e-commerce, you must do user reviews.

Unless a restaurant in a shop in a first-tier city is already a chain store (that is, a recognized one by everyone), the restaurant must clearly define its own cuisine, otherwise it will be difficult to survive. Therefore, in first-tier cities, stores must clearly state their purpose, products must be attractive, and evaluations must form a closed loop. For example, if you open a store and name it Birch Forest, believe it or not, the conversion rate will definitely be poor, because users will not know whether you are serving Sichuan cuisine or local cuisine. Therefore, if you want to engage in e-commerce in a niche field, you must learn these three skills.

To sum up, the three tricks in the above five steps are to use clear user portraits to recommend accurate products to help users make quick decisions and place orders, thereby increasing the stock growth of average order value. Hot-selling products use the attractiveness of products and user portraits to attract new users and achieve incremental growth.

Case Study

Here is an example of a daily priority new retail product:

Taking this example, nowadays, developing sinking markets, social e-commerce, new retail, and big data-enabled retail are not only urgent needs but also popular. Secondly, e-commerce brings together almost all the elements of AI big data, including data collection, data governance, data visualization applications, etc., which require the use of cloud computing, algorithm models, etc.

Let’s first look at the picture of the product details page:

After looking at the above two original pictures, I believe that judging from the user's mind, it is difficult for the user to have the desire to shop. Reasoning from the user's mind and the pictures of baked pastries that we often see in the circle of friends, we can modify the above two pictures.

User mind promotes modification steps:

• Use good reputation to promote condiments, strengthen the Yangzhi Ganlu in circle 1, and use Yangzhi Ganlu and other methods to recommend consumers to buy grapefruit, that is, use Yangzhi Ganlu as a final push to get customers to order grapefruit;
• The recommendation content in circle 2 is too crude. This kind of recommendation is only limited to the case where the user is hesitant about which grapefruit to buy. However, the recommendation is not only for this function. The real purpose of the recommendation here is to increase the average order value.
• Based on user mindset, it is recommended that the content in circle 3 be deleted directly.

Use digital intelligence to promote modification steps:

Now, whether it is developing the sinking market, whether it is serving existing users or incremental users, we have to be scientific, that is, use digital and intelligent means to achieve rapid and sustainable growth.

Continuing with the above example:

Step 1: Gather user portraits to generate a data set

1. Find the right "people", such as the gender, age, consumption level, occupation, etc. of people who like grapefruit;
2. Find the right "person" with interest tags, such as shopping, education, audio and video, games, financial management, etc.
3. Find the right "person" and customize the labels, such as customized label A, customized label B, etc.
4. Find the right "scenario", behavior tags, recently active apps, and recently visited scenarios;
5. Find the correct "scene", scene label, airport, business district, cinema, scenic spot, custom scene, etc.

Step 2: Data Governance

1. Choose the right wide table
2. Choose the right database

Step 3: Algorithm Model

1. From the perspective of business functions, clustering is used to find similar users who have common preferences for similar products;
2. Use association rule algorithms to mine the products that should be recommended, and use intelligent big data algorithms to calculate the content that should be recommended in circle 2 in the figure above.

Step 4: Use different data sets to test the accuracy of the algorithm model

1. test;
2. Collect data feedback and improve algorithm models.

When users cannot remember the recreated festivals, and when red envelope fission is costly and inefficient in attracting new customers, should our product operation technology focus more on hot products, the mental positioning of product user groups, and the use of technologies such as big data user portraits to build more detailed and down-to-earth operations?

Author: Lian Shilu

Source: LineLian