Recently, we are working on optimizing the success rate of Push push. I would like to share with you a related article that I compiled earlier.

An APP without message push function cannot be called an APP. Message push is a user operation tool commonly used by product and operation personnel. The purpose of push notifications is to:

1) To notify that something has happened. Such as successful payment deduction reminder, delivery reminder and so on.

2) Notify the platform of what is available, encourage users to view or participate, and increase platform activity, conversion, and retention. Such as a piece of information or a discount event.

APP push can be divided into two categories depending on whether there are pre-trigger conditions:

1) Business push: After a certain business logic is triggered, the program automatically completes the push. As mentioned above, when the program detects that an order has generated a waybill number or has been shipped, it will proactively send a push notification to the user.

2) Marketing push, without any pre-trigger conditions, is a non-business push sent by operations staff for the purpose of promoting user conversion, activity, and retention. For example, in a lottery event, the purpose is to guide users to open the APP to participate.

For different mobile phone operating systems, the message push processes of Android and iOS are different, which are introduced below.

1. Android push process

Developers generally use third-party push platform services directly to complete this. Of course, you can also choose to develop a message push solution yourself. This is a highly technical and complex project. Large companies can try it, but small and medium-sized enterprises should forget about it. From the perspective of technology investment, labor costs, implementation speed and final results, a professional third-party push service provider is a better choice.

The message push process of Android is as follows:

1) Operation personnel determine the push targets and content according to the operation objectives. The push objects here can be of different dimensions, such as device ID, account, tag, etc. The number of objects pushed is called the "total number of sends".

2) After the push object and content are given to the third-party push platform, they are not sent directly, but the validity of the device corresponding to the sending object is screened. The filtered valid devices are called "valid device count".

3) The third-party push platform further determines which devices among the valid devices have established a persistent connection with the push server. The prerequisite for sending messages is that "the device is connected to the Internet and establishes a long connection with the push server." Here, the devices with long online connections are called "number of sending devices".

4) Send messages to devices that have established long connections online and deliver the messages to the devices. The number of devices to which the message is delivered is called the number of devices delivered.

5) The message sent to the device in step 4 is further routed to the APP through the "package name" and displayed. The number of messages delivered to the APP is called the “number of messages delivered to the APP”.

6) After seeing the message, the user either clicks or ignores it. We define the number of times a message is clicked as “click count”.

At this point, a message is considered pushed.

2. Analysis of factors affecting conversion and conversion rate in each link of Android push

Knowing the process of message push, we can analyze the factors affecting conversion in each link and analyze how to improve the conversion rate of each link.

As shown in the figure above, the message push path is a funnel, and there is loss in each link. This means that there is room for improvement in every aspect. Next, let’s look at the influencing factors of each link and how to improve it.

1. Determine the recipient and content

“Total number of shipments” refers to the number of recipients selected by the operator. For example, select 100w.

1) The choice of recipients has a global impact on conversion rate.

For example, if the app has not been launched for 7 days and the app has not been launched for 30 days, the former is better than the latter in terms of valid devices, long online connections, delivery devices, and routing to APP display.

For example, pushing "exclusive activities for new users" to new users will have a higher click-through rate than pushing it to all users.

Therefore, when choosing the recipients, try to be as precise as possible and do segmented push according to different dimensions and tags. An important basic work for segmented push notifications is to label different types of users. Classify users through rich label dimensions. For example, by age, interests and hobbies, user level, number of transactions, new or old, registration time, retention, etc.

Unless the message is targeted at all users, reduce the amount of push notifications. On the one hand, full push notifications will frequently disturb users, and on the other hand, most users will receive content that they are not interested in, which may cause users to uninstall your app.

2) Copywriting mainly affects user clicks . For the same push recipients, attractive copywriting can get more clicks and views. This is mainly related to the copywriting ability of the operation personnel, so no further analysis is done here.

2. Identify effective equipment

"Number of valid devices" refers to the number of valid devices in the sending object. For example, 90w.

The conversion rate of this link is called "effective device rate", effective device rate = number of effective devices / total number of sends * 100% = 90/100 * 100% = 90%

The factors that affect the effective equipment rate are as follows.

1) Users uninstall the app, affecting the number of valid devices. There are many reasons why users uninstall apps, one of the most important being that push messages frequently disturb users. As shown in the figure below, annoying push notifications are the biggest reason why users uninstall apps.

To reduce the number of users uninstalling apps due to disturbing messages, you can start from two aspects.

On the one hand, you can control the frequency of push notifications. You can learn from the restrictions on message push frequency imposed by WeChat subscription accounts and services. For non-social and non-information applications, it is more appropriate to use them 1-2 times a week. Currently, most applications are used once a day, which is not restrained enough. The original intention was to stimulate users to launch the application and improve retention, but the result is counterproductive and drives users away.

On the other hand, the matching between pushed content and user needs needs to be improved. The content to be pushed should be what the recipient wants to see or can arouse his interest. That is, different content should be pushed to different objects as mentioned in step 1, and precise push should be implemented.

2) The user turns off the push function of the APP.

What we can do here is to effectively guide users to turn on APP push. There are two ways to guide them to turn on the push.

One type is when users download a new APP and launch it for the first time, they are informed of the benefits of enabling push reminders and are guided to enable push services.

One type is to guide users to turn on push reminders in the system settings when it is detected that the user's APP has not turned on the push service, using appropriate locations and scenarios.

3) The APP failed to register the unique identifier of the third-party platform.

What does this mean. The premise for all push notifications is that the APP registers a unique ID on a third-party push server, usually by calling an interface to register when the APP is started. If the registration fails, it will be judged as an invalid device. Registration failure may be caused by the user turning off the push function of the APP, or it may be caused by the failure of the interface call.

To improve the success rate of registering a unique identifier, you can guide users to enable the APP push service (see the previous point) and monitor the registration success rate of the registration interface.

If registration failure is detected, you can register again through the retry mechanism; if large-scale failures are detected, you need to check whether there is a problem with the interface, an integration error, or a new online release that affects the interface. The purpose of monitoring is to detect problems in advance.

4) The APP does not establish a long connection with the third-party push server within a specified period (for example, 3 months).

This is one of the conditions for judging the effectiveness of the device, and it may be different for different push platforms.

This is related to the retention of the app. If the app uninstall volume is high and the retention is poor, there will be fewer devices that establish long connections with third-party push services within a certain period of time.

3. Identify online long connections

"Number of devices sending push notifications" refers to the number of long connections established between valid devices and the third-party push platform server. Such as 50w.

The conversion rate of this link is called "device delivery rate". Device delivery rate = number of long-connected online devices/number of valid devices*100%=50/80*100%=62.5%

The long-term online connection of a device is related to three conditions.

1) The network environment is stable and good. This is the user's network environment, and developers and third-party push platforms have no control over it. What can be done is to reconnect through the retry mechanism after the user's network environment is good.

2) The push service is running. This means that the service of the third-party push platform is running on the user's mobile phone, which requires that the user is using an APP that has integrated the third-party push service. This requirement is too high, which means that as long as the APP is closed or running in the background, the message will not be received, and the effect is very poor. For this situation, third-party push platforms generally have a "long-term mutual protection mechanism" to improve the message delivery effect.

"Long-term mutual protection" means that when any APP on the user's mobile phone that has integrated a third-party push platform is opened, the push service can be started and push notifications can be received even if your app is not opened.

Simply put, assuming that the three applications A, B, and C on the user's mobile phone have integrated the SDK of a third-party push platform (such as Umeng), then through technical means, as long as one of the three apps A, B, and C is opened and becomes an active app, it can be guaranteed that the messages of the other two apps (even if they have not been opened) can also be sent through the channel of this opened app.

However, long-term mutual protection is ineffective in some deeply customized Android systems (such as Xiaomi and Huawei). Manufacturers are increasingly restricting the application of this technology based on power and data consumption considerations.

For this situation, developers also have their own solution - aggregate push, which is to integrate multiple push services. For example, Xiaomi Push, Huawei Push, and Umeng Push. When it is judged to be a Xiaomi device, Xiaomi Push will be used. When it is judged to be a Huawei device, Huawei Push will be used. For other devices, Umeng Push will be used.

Currently, some paid push service providers provide aggregated push services, and developers can complete the access in one go.

3) The push service connects to the server of the third-party push platform.

When the push service is running, it is possible that the device is not connected to the third-party push server, causing it to be offline. This is mainly a technical implementation. Please pay more attention to this when integrating third-party push.

4. Delivery equipment

"Number of devices delivered" refers to the number of devices to which messages have been sent. Such as 45w.

The conversion rate of this link is called "device delivery rate", which means the number of devices to which messages are delivered/the number of devices with long connections online. *100%=45/50*100%=90%

The number of devices delivered is mainly affected by network reasons, which may cause the message to fail to be sent to the device, such as a network disconnection (which may also cause the long-term connection channel to be disconnected). Generally speaking, the "number of devices delivered" and the "number of devices sent" are very close, usually above 98%.

5. Routing to APP display

"APP delivery number" refers to the number of messages displayed on the APP and visible to users. Such as 40w

The conversion rate of this link is called "APP delivery rate". APP delivery rate = number of APP deliveries/number of device deliveries*100%=40/45*100%=88.89%

After the message is delivered to the device, the factors that affect the routing to the APP are:

1) The APP was deleted, resulting in routing failure. This has been mentioned before and is related to APP retention. The key is how to improve APP retention, which I will not go into here in depth.

2) Some deeply customized Android systems (such as MIUI) have certain restrictions. For example, the ROM limits the process communication between apps, which will cause routing failure . In this type of situation, see if the third-party push platform provides compatibility solutions for some systems.

6. User clicks

"Number of clicks" refers to the number of times a message is clicked after it is visible to APP users. Such as 5w.

The conversion rate of this link is called "click-through rate". Click-through rate = number of user clicks/number of apps delivered*100%=5/40*100%=12.5%

When the message is finally displayed on the APP, everyone is definitely looking forward to the user clicking on it to view it. But looking at the entire market, the click-through rate of APP messages is relatively low, and 8% is considered very good.

The factors that affect user click-through rate mainly include:

1) The matching between the pushed content and the pushed object. As mentioned above, if you push football news to basketball fans, the click-through rate will definitely be low.

2) The content and appeal of the copy. This mainly depends on the copywriting level of the operator. Short, concise and powerful copywriting can greatly promote user clicks. I won’t go into details here.

The above analyzes 6 steps and 5 conversion rate indicators. However, as an operator, you generally don’t pay attention to such fine granularity.

Usually, operations personnel focus on "overall delivery rate" and "click rate".

Overall delivery rate = number of messages delivered to the APP/number of valid devices*100%=40/80*100%=50%. From the above analysis, we can see that there are many factors that affect the overall delivery rate, and many of these factors are related to third-party push platforms. When the push objects and content are the same, the overall delivery rate can reflect the advantages and disadvantages of the third-party push platform.

3. iOS message push process

Developers generally use a third-party push platform + APNs to complete it.

First, use a third-party push platform to hand over the push object and content to APNs (Apple Push Notification service, Apple's official push notification solution), and APNs will complete the rest, as shown in the following figure.

For the iOS platform, push data can only obtain the number of successful APNs delivery, but not the various data running within APNs. It is impossible to effectively analyze the push conversion links and conversion rate indicators.

But overall, because iOS uses the official unified push service, the conversion rate from the successful number of APNs delivery to the final APP display is much higher than that of Android. We can assume that after the successful number of APNs delivery, APNs will be successfully sent.

I will not go into details about iOS push here.

Improving the conversion rate of message push is a delicate job that requires breaking down each link and node for targeted analysis and improvement in order to achieve results.

I hope the above analysis will be of some help to you.

Author: Yue Laosan

Source: Product Notes