A recall strategy screening method and related device

Abstract

This application relates to the field of data processing technology, and provides a method and related apparatus for selecting recall strategies, which selects suitable recall strategies for specific users of an application to improve the recovery of users with low activity. The method mainly includes: determining the number X of recall strategies in the recall strategy set; dividing the sample user group into X sub-sample user groups, where X is a positive integer greater than 0; for each of the X sub-sample user groups, pushing the same recall strategy from the recall strategy set to sample users within the same sub-sample user group; calculating the recall rate for each sub-sample user group based on the number of users responding to the recall strategies in the recall strategy set within each sub-sample user group; and selecting target recall strategies that meet a preset standard for a target recall rate, where the target recall rate is one or more of the recall rates, and the target recall strategy is one or more of the recall strategies.

Description

Technical Field

[0001] This application belongs to the field of data processing technology, and in particular relates to a recall strategy screening method and related apparatus. Background Technology

[0002] For internet service providers (ISPs) of online applications, the longer a user spends on their application, the more dependent that user is on the application, and the higher the likelihood of the ISP gaining potential revenue from that user. For example, users who spend more time on an application are generally more interested in the application's paid services, and the application is more likely to facilitate a transaction for that user to make a purchase. Therefore, in order to increase user stickiness to their applications, ISPs usually need to proactively push content that can increase user interest, thereby achieving the goal of users spending more time on their applications. Recall strategies, as a plan to push specific content to users, are an effective means of winning back inactive users and increasing user dwell time on the application.

[0003] However, if internet service providers implement a recall strategy for users through the background of their applications, it may cause harassment if users are not interested in the strategy. In severe cases, it may lead to users uninstalling the application and causing user churn. If internet service providers do not actively implement recall strategies for users with low activity levels, their applications may gradually fade from users' lives, reducing users' dependence on the application and gradually decreasing the probability of users making purchases for their paid services.

[0004] It is evident that identifying suitable recall strategies for specific users has become a pressing technical problem that needs to be solved. Summary of the Invention

[0005] The purpose of this application is to provide a recall strategy screening method and related apparatus, which aims to screen out suitable recall strategies for specific users of an application, so as to increase the probability of recovering users with low activity and increasing the duration of user stay in the application.

[0006] Firstly, this application provides a recall strategy screening method, including:

[0007] Determine the number X of recall strategies in the recall strategy set, where X is a positive integer greater than 0, and the recall strategy is an implementation plan for pushing specific content to users;

[0008] The sample user group is divided into X sub-sample user groups, where the number of users in each sample user group is greater than or equal to X.

[0009] For X sub-sample user groups, push the same recall strategy from the recall strategy set to sample users in the same sub-sample user group;

[0010] The recall rate for each sub-sample user group is calculated based on the number of users responding to the recall strategies in the recall strategy set within each sub-sample user group.

[0011] Select target recall strategies that meet preset criteria for target recall rates, wherein the target recall rate is one or more of the recall rates, and the target recall strategy is one or more of the recall strategies.

[0012] Optionally, before dividing the sample user group into X sub-sample user groups, the method further includes:

[0013] A specific number of sample users are extracted from the set of users classified into the same target level as the sample user group.

[0014] Optionally, before extracting a specific number of sample users from the set of users categorized into the same target level, the method further includes:

[0015] Retrieve user behavior data for each user in the application's overall user set within the most recent time period;

[0016] Input the user behavior data of each user into a preset user churn prediction model to obtain the user churn probability of each user in the overall user set.

[0017] According to the preset grading criteria, each user in the overall user set is divided into a corresponding user churn level, resulting in Y user churn levels, where Y is a positive integer greater than 0, and the target level is one of the Y user churn levels.

[0018] Optionally, after selecting the target recall strategies that meet the preset criteria for target recall rates, the method further includes:

[0019] The target recall strategy is executed on all users in the same target level user set.

[0020] Optionally, before executing the targeted recall strategy on all users in the same target level user set, the method further includes:

[0021] Calculate the natural recall rate of the remaining users in the user set of the same target level after removing the sample users;

[0022] Determine whether the target recall rate is greater than the natural recall rate;

[0023] If the target recall rate is greater than the natural recall rate, then the target recall strategy will be executed on all users in the user set of the same target level.

[0024] If the target recall rate is less than or equal to the natural recall rate, a prompt will be made to input a new recall strategy.

[0025] Optionally, calculating the natural recall rate of the remaining users excluding the sample users in the user set of the same target level includes:

[0026] Obtain the number of naturally recalled users among the remaining users who meet the recall criteria in the most recent time period;

[0027] Calculate the organic recall rate, which is equal to the number of organically recalled users divided by the number of remaining users.

[0028] Optionally, after executing the target recall strategy on all users in the same target level user set, the method further includes:

[0029] Calculate the overall recall rate of all users in the user set with the same target level;

[0030] Determine whether the overall recall rate is greater than the natural recall rate;

[0031] If the overall recall rate is greater than the natural recall rate, then the target recall strategy is stored as the preferred recall strategy in association with the target level.

[0032] If the overall recall rate is equal to or less than the natural recall rate, then stop executing the target recall strategy for all users in the same target level user set.

[0033] Secondly, this application provides a recall strategy screening system, including:

[0034] A determining unit is used to determine the number X of recall strategies in the recall strategy set, where X is a positive integer greater than 0, and the recall strategy is an implementation plan for pushing specific content to users.

[0035] A partitioning unit is used to divide a sample user group into X sub-sample user groups, wherein the number of users in each sample user group is greater than or equal to X.

[0036] The push unit is used to push the same recall strategy from the recall strategy set to sample users in the same sub-sample user group for X sub-sample user groups.

[0037] The calculation unit is used to calculate the recall rate of each subsample user group based on the number of responding users who respond to the recall strategies in the recall strategy set in each subsample user group.

[0038] A filtering unit is used to filter out target recall strategies that meet preset criteria for target recall rates, wherein the target recall rate is one or more of the recall rates, and the target recall strategy is one or more of the recall strategies.

[0039] Optionally, the system further includes:

[0040] The extraction unit is used to extract a specific number of sample users from a set of users classified into the same target level as the sample user group.

[0041] Optionally, the system further includes:

[0042] The acquisition unit is used to acquire user behavior data for each user in the overall user set of the application in the most recent time period.

[0043] The input unit is used to input the user behavior data of each user into a preset user churn prediction model to obtain the user churn probability of each user in the overall user set.

[0044] The segmentation unit is further configured to classify each user in the overall user set into a corresponding user churn level according to a preset level segmentation standard, thereby obtaining Y user churn levels, where Y is a positive integer greater than 0, and the target level is one of the Y user churn levels.

[0045] Optionally, the system further includes:

[0046] An execution unit is used to execute the target recall strategy on all users in the user set of the same target level.

[0047] Optionally, the system further includes:

[0048] The calculation unit is also used to calculate the natural recall rate of the remaining users in the user set of the same target level after removing the sample users;

[0049] The judgment unit is used to determine whether the target recall rate is greater than the natural recall rate;

[0050] A triggering unit is configured to, if the target recall rate is greater than the natural recall rate, trigger the execution of the target recall strategy for all users in the user set of the same target level.

[0051] The reminder unit is used to remind the user to input a new recall strategy if the target recall rate is less than or equal to the natural recall rate.

[0052] Optionally, when the computing unit calculates the natural recall rate of the remaining users excluding the sample users in the user set of the same target level, it is specifically used for:

[0053] Obtain the number of naturally recalled users among the remaining users who meet the recall criteria in the most recent time period;

[0054] Calculate the organic recall rate, which is equal to the number of organically recalled users divided by the number of remaining users.

[0055] Optionally, the system further includes:

[0056] The computing unit is also used to calculate the overall recall rate of all users in the user set of the same target level;

[0057] The judgment unit is also used to determine whether the overall recall rate is greater than the natural recall rate;

[0058] A storage unit is used to associate and store the target recall strategy as a preferred recall strategy with the target level if the overall recall rate is greater than the natural recall rate.

[0059] The stopping unit is configured to stop executing the target recall strategy for all users in the same target level user set if the overall recall rate is equal to or less than the natural recall rate.

[0060] Thirdly, this application provides a computer device, comprising:

[0061] Processor, memory, bus, input / output interfaces, wireless network interface;

[0062] The processor is connected to the memory, the input / output interface, and the wireless network interface via a bus;

[0063] The memory stores a program;

[0064] When the processor executes the program stored in the memory, it implements the recall strategy screening method described in any one of the first aspects above.

[0065] Fourthly, this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the recall strategy screening method as described in any of the preceding first aspects.

[0066] Fifthly, this application provides a computer program product that, when executed on a computer, causes the computer to perform the recall strategy screening method as described in any of the first aspects above.

[0067] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0068] This application determines the number X of recall strategies in the recall strategy set, where X is a positive integer greater than 0, and a recall strategy is an implementation plan for pushing specific content to users. To test the effectiveness of all recall strategies initially, the application only tests them on a sample user group, dividing the sample user group into X sub-sample user groups, where the number of users in each sub-sample user group is greater than or equal to the number of recall strategies X. Then, for each of the X sub-sample user groups, the same recall strategy from the recall strategy set is pushed to sample users within the same sub-sample user group, achieving effective recall across different sub-sample user groups. Different recall strategies are tested in user groups; then, based on the number of users responding to the recall strategies in the set of recall strategies in each sub-sample user group, the recall rate of each sub-sample user group is calculated to know the implementation effect of each recall strategy; target recall strategies corresponding to the target recall rate that meet the preset criteria are selected, where the target recall rate is one or more of the recall rates, and the target recall strategy is one or more of the recall strategies, to obtain the target recall strategy that meets the implementation effect of the preset criteria. This target strategy can improve the probability of recovering users with low activity and increase the time users stay in the application. Attached Figure Description

[0069] Figure 1 This is a schematic flowchart of an embodiment of the recall strategy screening method of this application;

[0070] Figure 2 This is a schematic flowchart of another embodiment of the recall strategy screening method of this application;

[0071] Figure 3 This is a schematic diagram of an embodiment of the recall strategy screening system of this application;

[0072] Figure 4 This is a schematic diagram of another embodiment of the recall strategy screening system of this application;

[0073] Figure 5 This is a schematic diagram of the structure of one embodiment of the computer device of this application. Detailed Implementation

[0074] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0075] It should be noted that the recall strategy in this embodiment refers to an implementation plan for pushing specific content to users of a certain application. For example, recall strategies include, but are not limited to, proactively issuing coupons to users, proactively pushing high-quality content (content similar to or complementary to content viewed by the user), and one or more combinations thereof. However, the same recall strategy may have different effects on different users. If an internet service provider executes a recall strategy for a user through its application background, and the user is not interested in the strategy, it may cause harassment, and in severe cases, lead to the user uninstalling the application, resulting in user churn. If an internet service provider does not proactively execute a recall strategy for users with low activity levels, its application may gradually fade from the user's life, reducing the user's dependence on the application and gradually decreasing the probability of the user making a purchase through its paid services. Therefore, this embodiment needs to collect and store several recall strategies in advance to form a recall strategy set, so as to select a suitable recall strategy for a specific user group through the methods of the following embodiments of this application.

[0076] Please see Figure 1 An embodiment of the recall strategy screening method of this application includes:

[0077] 101. Determine the number of recall strategies X in the recall strategy set, where X is a positive integer greater than 0, and the recall strategy is an implementation plan for pushing specific content to users.

[0078] This step first determines the number X of recall strategies in the recall strategy set that needs to be detected in this embodiment. Each recall strategy is an implementation plan that pushes specific content to the user through the application's background. For example, recall strategies include, but are not limited to, one or more combinations of the following: proactively issuing coupons to users, proactively pushing high-quality content (content similar to or complementary to content viewed by the user).

[0079] 102. Divide the sample user group into X sub-sample user groups, where the number of users in each sample user group is greater than or equal to X.

[0080] Step 101: The recall strategies in the recall strategy set should not be directly applied to all users of the application to avoid incorrectly implementing recall strategies on unsuitable users, leading to user churn. A small subset of application users should be selected beforehand as a sample user group to test X recall strategies. Based on the test results, decisions can be made to apply certain recall strategies to a larger scale of similar users. For this reason, this step divides the sample user group into X sub-sample user groups, with the number of users in each sample user group being greater than or equal to X. This allows for one-to-one testing of all X recall strategies in Step 101 with each of the X sub-sample user groups. Typically, the number of users in each sample user group is several times or even tens of times greater than the number of recall strategies to reduce the impact of individual sample users on the final recall rate.

[0081] 103. For X sub-sample user groups, push the same recall strategy from the same recall strategy set to sample users in the same sub-sample user group.

[0082] The X recall strategies in step 101 are pushed one-to-one with the X sub-sample user groups in step 102, that is, the same recall strategy from the same set of recall strategies is pushed to the sample users in the same sub-sample user group.

[0083] 104. Calculate the recall rate for each subsample user group based on the number of users responding to the recall strategies in the recall strategy set within each subsample user group.

[0084] Specifically, the recall rate for each subsample user group is equal to the number of users in that subsample user group who responded to the recall strategy divided by the total number of users in that subsample user group.

[0085] 105. Select target recall strategies that meet preset criteria for target recall rates. Target recall rate is one or more of the recall rates, and target recall strategy is one or more of the recall strategies.

[0086] It is understood that this embodiment pre-stores preset standards for uniformly evaluating the recall rates of all recall strategies in step 104. For example, preset standards might include: the recall rate of a recall strategy must be greater than or equal to 50% (a value set according to actual needs, this is just an example), or selecting the recall strategy with the highest recall rate from the set of recall strategies. It is understood that preset standards can be set according to actual needs; no specific setting is made for the preset standards here. The recall rate that meets the above preset standards in this step is considered the target recall rate, and the recall strategy corresponding to this target recall rate is called the target recall strategy. These target recall strategies that meet the implementation effects of the preset standards can likely win back users with low activity levels and increase user dwell time in the application, and can be used for larger-scale applications with specific user groups of the same type.

[0087] Please see Figure 2 Another embodiment of the recall strategy screening method of this application includes:

[0088] 201. Obtain user behavior data for each user in the overall user set of the application in the most recent time period.

[0089] In this embodiment, the "application" refers to software (APP) with a certain user base. This step analyzes the overall user set of the application and obtains the user behavior data of each user in the overall user set for the most recent time period. The most recent time period can be set according to actual needs, such as the most recent week, the most recent month, etc. The user behavior data mainly includes the user's usage of the application, content browsing, and conversion; for example, usage can include: number of APP launches, APP browsing duration, APP active time periods, etc.; content browsing includes: number of case studies viewed, number of diary entries viewed, number of effect image views, number of page views, number of element exposures, number of element clicks, etc.; conversion includes: whether there is a service purchase, whether there is a customer added on WeChat, whether a contract is signed, etc. The user behavior data of each user for the most recent time period can be obtained from the user's user log.

[0090] 202. Input the user behavior data of each user into the preset user churn prediction model to obtain the user churn probability of each user in the overall user set.

[0091] In order to predict the probability of user churn for each user's behavior data obtained in step 201, this step requires the use of a preset user churn prediction model. This user churn prediction model is a trained machine learning algorithm model that can analyze user behavior data to determine the probability of a corresponding user turning into a churning user.

[0092] For the training process of the aforementioned machine learning algorithm model, churned users can be defined first. For example, users who uninstall the app can be defined as churned users, or users who have been inactive for 7 or more days / 14 days can be defined as churned users. Then, some historical user behavior data from user logs can be selected as feature variables for modeling. A suitable machine learning algorithm can be selected for model training, such as Random Forest or GBDT. When the AUC value (range 0 to 1) of the trained machine learning algorithm model reaches 0.8 or higher, it is considered a good user churn prediction model. Of course, the closer the AUC value of the user churn prediction model is to 1, the better its accuracy in predicting the probability of user churn from user behavior data. When the user churn prediction model performs well, a prediction model file for the user churn prediction model can be generated, usually a model prediction file with the extension .pkl, for use in this step.

[0093] 203. Based on the preset level classification criteria, each user in the overall user set is classified into the corresponding user churn level, resulting in Y user churn levels, where Y is a positive integer greater than 0.

[0094] After step 202, knowing the churn probability of each user in the overall user set of the application, this step then categorizes each user in the overall user set into a corresponding churn level according to a preset grading standard, resulting in Y user churn levels. For example, all users of the application can be divided into three user groups: high, medium, and low. Users with a churn probability > 0.66 are considered high-probability churn users, users with a churn probability < 0.33 are considered low-probability churn users, and the rest are medium-probability churn users. Of course, depending on actual needs, all users of the application can also be divided into 4 or 10 groups, etc. There is no limit to the number of user churn levels into which each user in the overall user set is classified.

[0095] 204. Extract a specific number of sample users from the user set that is divided into the same target level as the sample user group. The target level is one of the Y user churn levels.

[0096] In this embodiment, it is generally assumed that users within the same churn level have similar churn probabilities. Since users with similar churn probabilities are identified by the churn prediction model based on user behavior data, users within the same churn level are likely to exhibit similar user behavior and therefore are also likely to be recalled using the same effective recall strategy. Based on this understanding, this step extracts a specific number of sample users from the user set divided into the same target level (same churn level) as a sample user group. The smaller the sample user group represents of the total number of users in the same target level, the better, to reduce the impact of the test on the overall target level; conversely, the larger the sample user group, the better, to reduce the excessive influence of individual sample users on the recall rate calculation of the recall strategy. In practical applications, a balance can be achieved based on actual needs.

[0097] 205. Determine the number X of recall strategies in the recall strategy set, where X is a positive integer greater than 0, and the recall strategy is an implementation plan for pushing specific content to users.

[0098] The execution of this step is the same as described above. Figure 1 Step 101 in the embodiment is similar, and the repeated parts will not be described again here.

[0099] It should be noted that recall strategies in the recall strategy set can be added, combined, deleted, or modified as needed.

[0100] 206. Divide the sample user group into X sub-sample user groups, where the number of users in each sample user group is greater than or equal to X.

[0101] The execution of this step is the same as described above. Figure 1 Step 102 in the embodiment is similar, and the repeated parts will not be described again here.

[0102] 207. For X sub-sample user groups, push the same recall strategy from the same recall strategy set to sample users in the same sub-sample user group.

[0103] The execution of this step is the same as described above. Figure 1 Step 103 in the embodiment is similar, and the repeated parts will not be described again here.

[0104] 208. Calculate the recall rate for each subsample user group based on the number of users responding to the recall strategies in the recall strategy set within each subsample user group.

[0105] The execution of this step is the same as described above. Figure 1 Step 104 in the embodiment is similar, and the repeated parts will not be described again here.

[0106] 209. Select target recall strategies that meet the preset criteria for target recall rates. The target recall rate is one or more of the recall rates, and the target recall strategy is one or more of the recall strategies.

[0107] The execution of this step is the same as described above. Figure 1 Step 105 in the embodiment is similar, and the repeated parts will not be described again here.

[0108] 210. Calculate the natural recall rate of the remaining users in a user set that excludes the sample users for the same target level.

[0109] It is understandable that in the actual implementation of each application, users defined as churned (users who uninstalled the app, or users who have been inactive for 7 or more days, etc.) may return to use the application without any recall strategy and meet the application's activity criteria. Similarly, users in a user set at a certain target level may also meet the recall criteria in the recent time period without any recall strategy implemented; in this embodiment, such users are called naturally recalled users. Therefore, after extracting a specific number of sample users from the user set at the same target level in step 204, this step obtains the number of naturally recalled users who meet the recall criteria in the recent time period from the remaining users in the user set at the same target level excluding the aforementioned sample users, and then calculates the natural recall rate, where the natural recall rate is equal to the number of naturally recalled users divided by the number of remaining users. In practical applications, the recall criteria in this step can be a decrease from a higher user churn level to a lower user churn level; or, the recall criteria can be that user behavior data meets the activity criteria in the recent time period, etc.

[0110] 211. Determine whether the target recall rate is greater than the natural recall rate. If the target recall rate is less than or equal to the natural recall rate, proceed to step 212; if the target recall rate is greater than the natural recall rate, proceed to step 213.

[0111] It should be noted that the target recall strategy obtained in step 209 is calculated through testing on a subsample user group. Normally, the target recall rate of the subsample user group implementing the target recall strategy should be greater than the natural recall rate for the strategy to be considered effective. Therefore, this step further determines whether the target recall rate is greater than the natural recall rate. If the target recall rate is less than or equal to the natural recall rate, it proves that the target recall strategy is less effective than not implementing it at all. If the target recall rate is greater than the natural recall rate, it proves that the target recall strategy has some effect, and the difference between the target recall rate and the natural recall rate can indicate the quality of the effect; the larger the difference, the better the effect.

[0112] 212. Prompt you to enter a new recall strategy.

[0113] If step 211 determines that the target recall strategy is not as effective as not implementing it, in order to obtain a better recall strategy than natural recall, this step can reflect the input of a new recall strategy, thereby triggering the execution of step 205, and iterating through the test to find a more controllable solution than natural recall.

[0114] 213. Execute the target recall strategy for all users in the same target level user set.

[0115] If step 211 determines that the targeted recall strategy is effective, even better than natural recall, then this step can further execute the targeted recall strategy on all users in the same target user group to maximize the probability of recalling churned users at the same target level.

[0116] 214. Calculate the overall recall rate of all users in a user set with the same target level.

[0117] After executing the target recall strategy on all users in the user set with the same target level in step 213, responses from users who responded to the target recall strategy can be received. At this time, the number of responding users in the user set with the same target level can be counted, and then the overall recall rate of all users in the user set with the same target level can be calculated, where the overall recall rate is equal to the number of responding users divided by the total number of users in the user set with the same target level.

[0118] 215. Determine whether the overall recall rate is greater than the natural recall rate. If the overall recall rate is greater than the natural recall rate, proceed to step 216; if the overall recall rate is less than or equal to the natural recall rate, proceed to step 217.

[0119] Further evaluation is performed on the overall recall rate of the target recall strategy applied to the user set with the same target level in step 214. The overall recall rate is compared to the natural recall rate to verify the actual effectiveness of the target recall strategy on the same user set with the same target level. If the overall recall rate is greater than the natural recall rate, it proves that the target recall strategy has indeed maintained its effectiveness in the subsample user group. Furthermore, if the overall recall rate is significantly greater than the target recall rate in step 209, it proves that the effectiveness of the target recall strategy has been underestimated. If the overall recall rate is not significantly different from the target recall rate in step 209, it proves that the target recall strategy has a very objective effect in the subsample user group. If the overall recall rate is greater than the natural recall rate but much smaller than the target recall rate in step 209, it proves that the effectiveness of the target recall strategy has been overestimated. If the overall recall rate is less than or equal to the natural recall rate, it proves that the application of the target recall strategy to all users in the same user set with the same target level has actually failed.

[0120] 216. Store the target recall strategy as the preferred recall strategy in association with the target level.

[0121] If step 215 determines that the overall recall rate is greater than the natural recall rate, it proves that the targeted recall strategy has indeed maintained the test results in the subsample user group. This targeted recall strategy is a recall strategy that can be implemented multiple times or over a long period of time. In particular, a targeted recall strategy with an overall recall rate much greater than the target recall rate in step 209 proves that the effect of the targeted recall strategy has been underestimated. This step stores the targeted recall strategy as a preferred recall strategy associated with the target level, so that it can be recommended as a preferred recall strategy for decision-makers when users at the target level need to implement a recall strategy.

[0122] 217. Stop executing the target recall strategy for all users in the same target level user set.

[0123] If step 215 determines that the overall recall rate is less than the natural recall rate, it proves that the target recall strategy for all users in the same target level user set has actually failed. The target recall strategy should not be executed again for all users in the same target level user set, and should be stopped to avoid further exacerbating user churn.

[0124] The above embodiments describe the recall strategy screening method of this application. The recall strategy screening system of this application is described below. Please refer to [link / reference]. Figure 3 An embodiment of the recall strategy screening system of this application includes:

[0125] The determining unit 301 is used to determine the number X of recall strategies in the recall strategy set, where X is a positive integer greater than 0, and the recall strategy is an implementation plan for pushing specific content to users.

[0126] The partitioning unit 302 is used to divide the sample user group into X sub-sample user groups, wherein the number of users in the sample user group is greater than or equal to X.

[0127] Push unit 303 is used to push the same recall strategy from the recall strategy set to sample users in the same sub-sample user group for X sub-sample user groups.

[0128] The calculation unit 304 is used to calculate the recall rate of each subsample user group based on the number of responding users in each subsample user group who respond to the recall strategies in the recall strategy set.

[0129] The filtering unit 305 is used to filter out target recall strategies that meet preset criteria for target recall rates, wherein the target recall rate is one or more of the recall rates, and the target recall strategy is one or more of the recall strategies.

[0130] The operations performed by the recall strategy screening system in the application embodiment are the same as those described above. Figure 1 The operations performed in the embodiments are similar and will not be described again here.

[0131] Please see Figure 4 Another embodiment of the recall strategy screening system of this application includes:

[0132] The determining unit 401 is used to determine the number X of recall strategies in the recall strategy set, where X is a positive integer greater than 0, and the recall strategy is an implementation plan for pushing specific content to users.

[0133] The partitioning unit 402 is used to divide the sample user group into X sub-sample user groups, wherein the number of users in the sample user group is greater than or equal to X.

[0134] Push unit 403 is used to push the same recall strategy from the recall strategy set to sample users in the same sub-sample user group for X sub-sample user groups.

[0135] The calculation unit 404 is used to calculate the recall rate of each subsample user group based on the number of responding users in each subsample user group who respond to the recall strategies in the recall strategy set.

[0136] The filtering unit 405 is used to filter out target recall strategies that meet preset criteria for target recall rates, wherein the target recall rate is one or more of the recall rates, and the target recall strategy is one or more of the recall strategies.

[0137] Optionally, the system further includes:

[0138] Extraction unit 406 is used to extract a specific number of sample users from a set of users divided into the same target level as the sample user group.

[0139] Optionally, the system further includes:

[0140] Acquisition unit 407 is used to acquire user behavior data of each user in the overall user set of the application in the most recent time period;

[0141] Input unit 408 is used to input the user behavior data of each user into a preset user churn prediction model to obtain the user churn probability of each user in the overall user set;

[0142] The segmentation unit 402 is further configured to classify each user in the overall user set into a corresponding user churn level according to a preset level segmentation standard, thereby obtaining Y user churn levels, where Y is a positive integer greater than 0, and the target level is one of the Y user churn levels.

[0143] Optionally, the system further includes:

[0144] The execution unit 409 is used to execute the target recall strategy on all users in the user set of the same target level.

[0145] Optionally, the system further includes:

[0146] The calculation unit 404 is also used to calculate the natural recall rate of the remaining users in the user set of the same target level after removing the sample users;

[0147] The judgment unit 410 is used to determine whether the target recall rate is greater than the natural recall rate;

[0148] Triggering unit 411 is used to trigger the execution of the target recall strategy for all users in the user set of the same target level if the target recall rate is greater than the natural recall rate;

[0149] The reminder unit 412 is used to remind the user to input a new recall strategy if the target recall rate is less than or equal to the natural recall rate.

[0150] Optionally, when the calculation unit 404 calculates the natural recall rate of the remaining users excluding the sample users in the user set of the same target level, it is specifically used for:

[0151] Obtain the number of naturally recalled users among the remaining users who meet the recall criteria in the most recent time period;

[0152] Calculate the organic recall rate, which is equal to the number of organically recalled users divided by the number of remaining users.

[0153] Optionally, the system further includes:

[0154] The calculation unit 404 is also used to calculate the overall recall rate of all users in the user set of the same target level;

[0155] The judgment unit 410 is also used to determine whether the overall recall rate is greater than the natural recall rate;

[0156] Storage unit 413 is used to associate and store the target recall strategy as a preferred recall strategy with the target level if the overall recall rate is greater than the natural recall rate.

[0157] Stop unit 414 is used to stop executing the target recall strategy for all users in the same target level user set if the overall recall rate is equal to or less than the natural recall rate.

[0158] The operations performed by the recall strategy screening system in the application embodiment are the same as those described above. Figure 2 The operations performed in the embodiments are similar and will not be described again here.

[0159] The computer device according to embodiments of this application is described below. Please refer to... Figure 5 One embodiment of the computer device in this application includes:

[0160] The computer device 500 may include one or more central processing units (CPUs) 501 and memory 502, wherein the memory 502 stores one or more application programs or data. The memory 502 is volatile or persistent storage. The program stored in the memory 502 may include one or more modules, each module including a series of instruction operations on the computer device. Furthermore, the processor 501 may be configured to communicate with the memory 502 and execute the series of instruction operations stored in the memory 502 on the computer device 500. The computer device 500 may also include one or more wireless network interfaces 503, one or more input / output interfaces 504, and / or one or more operating systems, such as Windows Server, Mac OS, Unix, Linux, FreeBSD, etc. The processor 501 can execute the aforementioned... Figures 1 to 2 The specific operations performed in the illustrated embodiment will not be described in detail here.

[0161] In the several embodiments provided in this application, those skilled in the art should understand that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0162] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0163] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A recall strategy screening method, characterized in that, include: Determine the number X of recall strategies in the recall strategy set, where X is a positive integer greater than 0, and the recall strategy is an implementation plan for pushing specific content to users; The sample user group is divided into X sub-sample user groups, where the number of users in each sample user group is greater than or equal to X. For X sub-sample user groups, push the same recall strategy from the recall strategy set to sample users in the same sub-sample user group; The recall rate for each sub-sample user group is calculated based on the number of users responding to the recall strategies in the recall strategy set within each sub-sample user group. Select target recall strategies that meet preset criteria for target recall rates, wherein the target recall rate is one or more of the recall rates, and the target recall strategy is one or more of the recall strategies; After selecting target recall strategies that meet preset criteria for target recall rates, the method further includes: The target recall strategy is executed on all users in the user set of the same target level. Before executing the target recall strategy on all users in the same target level user set, the method further includes: Calculate the natural recall rate of the remaining users in the user set of the same target level after removing the sample users; Determine whether the target recall rate is greater than the natural recall rate; If the target recall rate is greater than the natural recall rate, then the target recall strategy will be executed on all users in the user set of the same target level. If the target recall rate is less than or equal to the natural recall rate, a prompt will be made to input a new recall strategy.

2. The recall strategy screening method according to claim 1, characterized in that, Before dividing the sample user group into X sub-sample user groups, the method further includes: A specific number of sample users are extracted from the set of users classified into the same target level as the sample user group.

3. The recall strategy screening method according to claim 2, characterized in that, Before extracting a specific number of sample users from a set of users categorized into the same target level, the method further includes: Retrieve user behavior data for each user in the application's overall user set within the most recent time period; Input the user behavior data of each user into a preset user churn prediction model to obtain the user churn probability of each user in the overall user set. According to the preset grading criteria, each user in the overall user set is divided into a corresponding user churn level, resulting in Y user churn levels, where Y is a positive integer greater than 0, and the target level is one of the Y user churn levels.

4. The recall strategy screening method according to claim 1, characterized in that, The calculation of the natural recall rate of the remaining users in the user set of the same target level, excluding the sample users, includes: Obtain the number of naturally recalled users among the remaining users who meet the recall criteria in the most recent time period; Calculate the organic recall rate, which is equal to the number of organically recalled users divided by the number of remaining users.

5. The recall strategy screening method according to claim 1, characterized in that, After executing the target recall strategy on all users in the same target level user set, the method further includes: Calculate the overall recall rate of all users in the user set with the same target level; Determine whether the overall recall rate is greater than the natural recall rate; If the overall recall rate is greater than the natural recall rate, then the target recall strategy is stored as the preferred recall strategy in association with the target level. If the overall recall rate is equal to or less than the natural recall rate, then stop executing the target recall strategy for all users in the same target level user set.

6. A recall strategy screening system, characterized in that, include: A determining unit is used to determine the number X of recall strategies in the recall strategy set, where X is a positive integer greater than 0, and the recall strategy is an implementation plan for pushing specific content to users. A partitioning unit is used to divide a sample user group into X sub-sample user groups, wherein the number of users in each sample user group is greater than or equal to X. The push unit is used to push the same recall strategy from the recall strategy set to sample users in the same sub-sample user group for X sub-sample user groups. The calculation unit is used to calculate the recall rate of each subsample user group based on the number of responding users who respond to the recall strategies in the recall strategy set in each subsample user group. A filtering unit is used to filter out target recall strategies that meet preset criteria for target recall rates, wherein the target recall rate is one or more of the recall rates, and the target recall strategy is one or more of the recall strategies. The system also includes: An execution unit is used to execute the target recall strategy on all users in the user set of the same target level; The system also includes: The calculation unit is also used to calculate the natural recall rate of the remaining users in the user set of the same target level after removing the sample users; The judgment unit is used to determine whether the target recall rate is greater than the natural recall rate; A triggering unit is configured to, if the target recall rate is greater than the natural recall rate, trigger the execution of the target recall strategy for all users in the user set of the same target level. The reminder unit is used to remind the user to input a new recall strategy if the target recall rate is less than or equal to the natural recall rate.

7. A computer device, characterized in that, include: Processor, memory, bus, input / output interfaces, wireless network interface; The processor is connected to the memory, the input / output interface, and the wireless network interface via a bus; The memory stores a program; When the processor executes the program stored in the memory, it implements the recall strategy screening method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer storage medium stores instructions that, when executed on the computer, cause the computer to perform the recall strategy screening method as described in any one of claims 1 to 5.

Images