First, why do we need to do paid forecasting?
Based on the user’s recent in-app behavior and attribute characteristics, combined with machine learning technology, users with different probabilities of paying in the next week are selected from the active users in the past week, which is the payment prediction.
At present, the liquidity of a product is an important consideration to measure whether it can move towards benign development. Therefore, obtaining income has become one of the most critical links in the process of product operation. In the second half of the Internet era, traffic competition is becoming increasingly fierce, so it is particularly important to dig out the existing users. Through payment prediction, the user groups that will have payment behaviors in the next week are identified in advance from the users, and targeted marketing strategies are formulated for these users, such as the configuration of time-limited discount packages and other user operation behaviors, to further promote payment transformation.
Second, how to promote the generation of users’ payment behavior?
The high cost of pulling new products makes enterprises pay more and more attention to the full value of the life cycle of each stock user and guide them to pay and convert continuously, which can effectively improve the profitability of products. The constant payment behavior also reflects the stickiness of users. Highly engaged users can help a product find more potential users, which can help a product achieve a virtuous monetization cycle.
With the in-depth operation of products, all kinds of APPS have gradually explored the transformation mode suitable for their own industry characteristics. E-commerce and game apps are the most typical.
(I) E-commerce apps
For e-commerce apps, the tense atmosphere created by in-app activities such as time-limit reduction and on-time buying will usually stimulate potential users who hesitate to pay to make final payment.
(2) Game apps
For game apps, the purchase of props is the main form of payment conversion. In addition to attracting the preferential activities of props, pushing payment reminders at key points in the level will greatly promote the payment conversion of players.
Third, how to combine paid forecast to do revenue growth?
Different categories of apps have different profit models. Taking game apps as an example, we introduce in detail how to combine payment forecasting to help products increase revenue.
Currently, the main form of payment conversion in games is the purchase of items, but a lot of revenue is generated through in-app video AD clicks. We will start with these two dimensions and detail the specific application of payment forecasting.
L The principle of paid forecasting for growth
To put it simply, the principle of payment prediction is to select the users who are likely to pay in the next week from the circle of players based on their previous in-game behaviors, such as the app version number, the number of sessions in the last 7 days and other attributes and behavioral data through an algorithm model.
By default, the system divides the active users into high (>70%), medium ((20%,70%)) and low (≤20%) groups according to the probability of payment events occurring within the next week. Of course, you can also customize the probability range according to the actual game and the characteristics of the players, for example, define the probability range of 50-70% of the middle and high probability of paying, and then customize the probability range.
* The illustration shows the paid forecast details page
After the delineation of the population, the next is the process of fine operation with growth services.
L High probability people focus on in-app revenue
High-probability payers are potential recharge players, and their probability of paying is much higher than other groups. Keywords such as “recharge limited time discount” and “limited props release” are pushed, including “special benefits” and “gift promotion”, which can reduce the psychological game in the process of payment. At the same time, words with a strong sense of urgency like “limited time” and “limited time” create a tense atmosphere and further promote the generation of user payment behavior.
Currently, the high-probability payers generated by the prediction can be directly used for the audience segmentation of Huawei push service and in-app messages. For example, when users browse the details of card drawing activities, through the background configuration of in-app messages, users are reminded of the limited time drawing activities of rare props in the form of popup window for those with high probability of paying. In this way, users with strong purchase intention can further stimulate the conversion of payment through limited time activities.
It can also send top-up and discount activities to users in the form of message push, and make targeted push based on users’ intention to pay, so as to achieve twice the result with half the effort.
The above hierarchical group operation mode is very convenient. When the scope of message push is configured in the background, direct circular selection can realize the targeted push.
* The figure shows the predicted audience configuration for the push service
L Low probability people focus on advertising revenue
For the low-probability payers, it is even more difficult to dig out the payment opportunity from them. Adjusting the revenue model and changing the profit thinking will help us dig out more of their value. The other most common monetization model for games is in-app advertising.
Many game developers worry that too much AD placement will affect the game experience of players, and the effect of the delivery is not good enough, but it will also bring the risk of losing players. How about if you’re running out of energy at the end of a key level of the game, but you don’t want to buy a game item, then the chance to come back to life after watching a 15s motivational video feels very useful? This is the clever placement of in-game ads.
Skillfully use the hearts of the players in the game, and attract the low-probability paying users to click on the incentive video in the game with key game props such as gold coins and physical strength. In this way, the click-through rate of the advertisements in the game will be greatly increased, and the advertising revenue will increase accordingly.
By applying the predicted low probability payers to the filtering conditions configured remotely, the incentive videos can be displayed in a targeted manner without affecting the gaming experience of other users.
* Figure shows the prediction criteria filtering interface for remote configuration
The above is an introduction to the application of paid forecast in the actual scenario of product operation. You can also combine the actual characteristics of the current application to find other revenue breakthrough points for paid forecast groups and develop differentiated and diversified operation strategies.
Welcome to experience and use the forecasting service. For more information, please click on the official website.
For more details on HMS Core, please refer to >> Huawei Developer Alliance website
To participate in the developer discussion go to CSDN or Reddit >> download demo and sample code on Github or Gitee >> Resolve integration issues go to Stack Overflow
The original link: developer.huawei.com/consumer/cn…
Original author: Pepper