How to improve new user retention through APP downloads? 6 tips!

This article shares with you some tips on improving new user retention through APP downloads, which can be used as a reference for application download placement.

There are several ways to optimize APP retention, which can be roughly divided into: optimizing channel traffic; finding the key behaviors of high-retention users to optimize products; creating a points/coin mall to continuously improve user stickiness and activity; and consolidating content to provide core value for users.

There are also some conventional operational methods, such as Push, RTA activation, etc. Here we mainly share how to optimize new user retention from the channel delivery side.

1. Optimize channel sources

On the premise that the testing budget is sufficient, various mainstream channels can be rolled out to test the delivery, including mobile phone manufacturers such as Huawei, Xiaomi, OPPO, and VIVO, and information flow channels such as Toutiao, Guangdiantong, and Kuaishou, and channels with good retention and high user quality can be selected for delivery.

Other small channels, such as Meizu, Samsung, Qutoutiao, WIFI Master Key, etc., can be appropriately selected if there is a demand for expansion, but the output effect and user quality are generally not as good as mainstream channels.

It is important to note here that both stores and information flow channels have incentive resource positions, such as the installation gifts of OPPO/Huawei mobile app stores and the incentive mall position of Qutoutiao. Their customer acquisition costs and next-day retention and seven-day retention are very good, but the data such as fifteen-day retention, thirty-day retention and user usage time are very poor.

These users are obviously profit-seeking. They are not attracted by the content and functions of the products themselves. After taking advantage of the products, they leave. The second type are alliance resource positions in information flow and manufacturer store channels. Users are relatively scattered and strongly profit-driven. There is a general retention bias, and the possibility of fake traffic cannot be ruled out.

With a limited budget, you can focus on mobile phone manufacturer stores and search channels , bring in more natural traffic by optimizing store titles, descriptions, pictures and comments, and obtain more free resources through first launches and business relationships. Paid promotion can focus on brand words and product words; at the same time, increase the accuracy of traffic through search channels; information flow channels can also be considered as appropriate, and by communicating with the media about the cost and retention of competing products, select one or two channels for testing.

Note: If you find that the retention rate of a certain channel is very poor, you should investigate the reasons. For example, the retention rate difference of Qutoutiao can be inferred that the target audience of Qutoutiao is relatively low-end, and most of them are in fourth-, fifth- and sixth-tier cities. Why is the retention rate of low-end users low? On the contrary, the content of the APP homepage requires a higher level of appreciation, so we can optimize the display of the homepage content to suit each person, rather than pausing this channel.

2. Personalized material delivery

Match people and materials . For example, if the product is iQiyi Suike and it invested in an animated drama on Douyin, the corresponding group should be people under 22 years old. Summarize the attributes of the materials and deliver them to the corresponding people, or study the users first and then make the corresponding materials. The selling points of the materials can be refined according to the advantages of their own products, or they can refer to the high-quality materials of competitors, which can effectively improve the delivery effect and user retention.

After the material is produced, a large number of AB tests are required. We can split a material into several dimensions for controlled variable testing, such as material scenes, product selling points, BGM, landing pages, duration and rhythm, etc., to find out which dimension affects users the most, and then focus on breakthroughs.

Let me introduce another product function of the current information flow channel: Deeplink lifting, which is also called scene restoration in some companies and single book lifting in the novel industry. To put it simply, users download the APP through some kind of delivery material, and after activating the application, this content is the first thing displayed to users.

For example: I was browsing Douyin and saw the material of "Snowy Swordsman", which was an advertisement placed by iReader. I downloaded and activated iReader, and the novel "Snowy Swordsman" was displayed first when I entered the APP. What you think is what you get, which improves user experience and retention. The key to product logic is to associate the mkey and content ID value in the feedback monitoring link. After the user activates the APP, the server will obtain the content ID through the specific mkey value and then send it to the client to achieve this.

The information flow also has a function called dynamic material delivery. For example, iQiyi and Bytedance can be connected with an API, and the advertising delivery backend can be connected to iQiyi's massive content. During the advertising delivery process, Toutiao can first lock the tags of the users who are about to be delivered, and then send the corresponding material request to iQiyi. iQiyi will select the corresponding material for delivery and display. This method is more accurate in matching people and materials. At present, this functional platform has not been fully developed, so we can wait and see.

3. DMP and crowd portrait

The DMP of information flow channels such as Toutiao and Kuaishou has two main functions. One is to create a new population, where you can select your own target users for delivery based on the platform's own user tags; the other is to upload the population, upload the IMEI or IDFA of old users, and package them into a population package for delivery. If the population size is not large, it can be expanded, and it is recommended to be less than 4 times.

The larger the target audience, the less accurate it will be. For more vertical products, using DMP for delivery is a good way to improve user accuracy.

There are many dimensions in crowd portraits . The background of Bytedance’s delivery includes region, gender, age, interests, network, etc. Generally, the client will also have its own user analysis background. Taking iQiyi as an example, the angles of user analysis can be celebrity preference, IP content preference, viewing time, consumption concept, life outlook, etc.

We can conduct population portrait analysis on converted users or long-retained users, and then make refined targeting when delivering ads to improve user quality and retention.

4. Delivery time

There are three points to note. First, understand the high-activity periods of the APP, and then increase the channel delivery during this period.

Taking short video software as an example, noon, evening and weekends are generally the high-activity periods. The high-activity periods of most other APPs are also in the above three time periods. During this period, users have plenty of time to deeply experience the product, explore the true value of the product, and are more likely to stay.

Second, if it is delivered at night, due to the length of the content, the next-day retention of new users will generally be higher. The probability of finishing the app when you open it at night is lower than that in the morning or morning, so the probability of opening it the next day is higher.

Third, the secondary retention rate on Sundays is generally lower, which is normal because Monday is a working day and a small number of users may not open the APP, but open it again on the next Saturday or Sunday, so the long-term retention rate is not affected and may be higher than on working days.

5. Double bid for the second stay

Mainstream platforms such as Toutiao, Kuaishou, and Guangdiantong have double-bids , and media such as WIFI and Qutoutiao also have them, but the algorithms are not yet mature.

Take Kuaishou as an example. For all advertisers who send back deep conversion data such as next-day retention data (mainly for Internet services, games, and e-commerce industries), payment data (mainly for the game industry), completion data (financial industry), and credit data (financial industry) to Kuaishou, and if the account sends back more than 5 pieces of this type of data, the account will automatically open the use permission of "deep conversion goal" optimization.

This function introduces deep conversion optimization based on the original ocpc activation/form. When the customer accurately and in real time sends the deep conversion data (payment, next-day process, credit, completion) back to Kuaishou (return connection document: https://yiqixie.com/d/home/fcAA-2e87at1BgxhXBAjOQueu), the system model will continue to assist in optimizing the "deep conversion target" on the basis of ensuring the achievement of the "target cost", and try to reduce the deep conversion target cost, but it will not guarantee the achievement of the specific deep conversion target cost.

Upgrading from optimizing based on the activated population as the delivery model to optimizing based on the secondary retention population as the model will make delivery more precise and can effectively improve secondary retention.

6. Direct search

Direct search: When users search for keywords in the search box, they will be redirected directly to the video playback details page, skipping the original search results page. This has the following advantages:

● "Search Direct" gives users clear directions to quickly reach content, increasing user playback penetration

● Add recommendation paths to solve the problem that some information flow channels do not support the integrated connection method. Through direct search and playback, users' content interest tags can be obtained, which can immediately trigger the recommendation strategy.

● Simplify the search operation steps. The original search has a result page, and one more path means one more loss

The above advantages can slightly improve the retention of new users, so you can try them.

Author: Dai Shaohua

Source: Dai Shaohua