Mail pushing method and system based on network marketing platform

Abstract

The invention relates to the technical field of data processing, in particular to a mail pushing method and system based on a network marketing platform. The method comprises the following steps: in response to establishment of a mail pushing task, performing filtering verification on the mail pushing task to judge whether the mail pushing task meets a preset requirement; in response to the fact that the mail pushing task passes the filtering verification, distributing a sending domain name according to the interactive representation of a recipient corresponding to the mail pushing task in a preset period; and determining the execution time of the mail pushing task, and executing the mail pushing task by using the sending domain name according to the execution time. According to the invention, reputation decline can be reduced, the risk that subsequent mails are intercepted is reduced, and the mail delivery effect is improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to an email push method and system based on an online marketing platform. Background Technology

[0002] As online marketing platforms are increasingly used for brand promotion and content collaboration, platform operators typically need to send collaboration invitations, task notifications, and related operational instructions to a large number of content creators or marketing targets via email. Due to the low cost and wide reach of email, existing online marketing systems generally use scheduled tasks or batch scheduling to achieve centralized email delivery.

[0003] In existing technologies, email push processes typically include steps such as task creation, scheduled triggering, and unified sending, and information delivery is completed based on preset email templates and sending configurations. To simplify system implementation, existing solutions often employ fixed or simple polling sending domain strategies, and determine whether sending is allowed based on several rules before sending. However, in practical applications, such solutions usually only judge based on whether the sending conditions are met, lacking differentiation of different recipients' historical interaction behavior, and failing to dynamically adjust the sending strategy according to factors such as recipient email reply status and activity differences.

[0004] In large-scale email delivery scenarios, existing technologies lack specificity in domain name selection due to varying response levels among recipients. This can easily lead to the mixing of high-response and low-response recipients, negatively impacting the reputation of the sending domain and reducing overall email delivery rates. Furthermore, common sending rules in existing email systems are often fixed in the business process via code. Adding or adjusting filtering conditions often requires modifying the existing logic, resulting in insufficient system scalability and maintenance flexibility.

[0005] On the other hand, most existing email push systems rely on fixed time points to trigger sending tasks, failing to fully consider the distribution of recipients' active time and fluctuations in system operation. When scheduled tasks are triggered in a concentrated manner or when the system experiences short-term delays, it can easily lead to concentrated email sending or a shift in sending timing, which not only increases the instantaneous load on the system but may also affect email open rates and response times.

[0006] In summary, existing email sending processes based on online marketing platforms are prone to problems such as declining reputation of the sending domain and email delivery delays, which affect the effectiveness of email delivery. Summary of the Invention

[0007] To reduce issues such as domain reputation degradation, concentrated delayed email delivery, and template parameter abuse during email sending, and to improve email delivery rate and system stability, this application provides an email push method and system based on an online marketing platform.

[0008] Firstly, this application provides an email push method based on an online marketing platform, employing the following technical solution: An email push method based on an online marketing platform includes: in response to the creation of an email push task, filtering and verifying the email push task to determine whether the email push task meets preset requirements; In response to the email push task passing the filtering verification, the sending domain name is assigned according to the interaction performance of the corresponding recipient within a preset period of the email push task. Determine the execution time of the email push task, and execute the email push task using the sending domain name according to the execution time; The method of allocating sending domains based on the interaction performance of the recipients corresponding to the email push task within a preset period includes: obtaining historical interaction data between the recipients corresponding to the email push task and the marketing platform within the preset period, calculating the historical response rate of the target recipients; determining whether the historical response rate meets a preset ratio threshold; in response to meeting the preset ratio threshold, allocating any sending domain from a preset first type of sending domain pool to the email push task; otherwise, allocating any sending domain from a second type of sending domain pool to the email push task, wherein the average reputation of the sending domains in the first type of sending domain pool is greater than the average reputation of the sending domains in the second type of sending domain pool.

[0009] First, the email push task is filtered and verified. Before the email is actually sent, the compliance of the email content and the recipient status are checked. This can intercept potential illegal or low-quality email push tasks at the system entry point, reduce invalid sending behavior from the source, reduce the risk of the sending domain being complained about or blacklisted, and at the same time reduce the initial computing pressure on the server.

[0010] Secondly, after the email push task passes the filtering verification, the sending domain name is dynamically allocated based on the recipient's interaction performance within a preset period. By quantifying the recipient's historical response rate, recipients with different interaction qualities are distinguished. This allows email push tasks with a high probability of interaction to use sending domain names with higher credibility, while tasks with a low probability of interaction use relatively secondary sending domain names. In this way, without changing the overall sending volume, the possibility of high-quality sending domain names losing credibility due to low-quality recipients is reduced, and the probability of emails being blocked due to sending domain names is also reduced, further improving email delivery performance.

[0011] Furthermore, by determining the execution time of the email push task and using the allocated sending domain name to send the email at that time, a correspondence is established between the use of the sending domain name and the recipient's response capability. This not only protects the stability of sending resources but also improves the actual delivery rate and reply probability of emails, effectively alleviating the problems of declining domain reputation and low email delivery rate in existing technologies.

[0012] Optionally, filtering and verification of email push tasks may include at least: detecting whether there are sensitive words in the email body and / or email title, and detecting whether the recipient is a blocked target of the sender.

[0013] Sensitive word detection can identify high-risk expressions at the email content level that are likely to trigger spam rules or user complaints, thereby blocking related tasks in time before sending and reducing bounce messages, complaints or bans caused by inappropriate content; by detecting whether the recipient is a person blocked by the sender, it can avoid continuing to push information to people who clearly do not want to receive emails.

[0014] Optionally, the number of times the target recipient made a valid reply within a preset period can be obtained from the historical interaction data between the recipient and the marketing platform, as a reply count value; the total number of emails sent by the marketing platform to the target recipient can be obtained, as a sending count value; and the ratio of the reply count value to the sending count value can be used as the recipient's historical reply rate.

[0015] The ratio of the number of valid replies generated by the recipient within a preset period to the total number of emails sent is calculated, thus transforming the recipient's interaction performance into a quantifiable metric.

[0016] Optionally, the credibility of the sending domain name is negatively correlated with the total number of user complaints generated during the use of the domain name.

[0017] The average reputation of domains in the first category of the mailing domain pool is higher than that in the second category, structurally achieving hierarchical management of mailing resources. Simultaneously, establishing a negative correlation between domain reputation and the total number of user complaints gives reputation evaluation clear technical meaning and a statistical basis. This allows mailing tasks for high-interaction recipients to prioritize the use of mailing domains with lower complaint rates and higher stability, while low-interaction tasks use relatively secondary domains.

[0018] Optionally, the recipient's historical interaction behavior data within a preset period can be obtained, and the interaction behavior data includes at least one of the following: email opening time, link clicking time, and reply sending time. Analyze historical interaction data to determine the recipient's optimal active window, and select a time point within the optimal active window as the execution time.

[0019] By acquiring historical interaction data of recipients within a preset period and determining the optimal active window based on this data, the email execution time no longer depends on a fixed point in time, but matches the actual behavior habits of the recipients. On the one hand, this can increase the probability of emails being viewed and replied to in a timely manner, improving delivery effectiveness. On the other hand, since different recipients have different behavior habits, their corresponding execution times are also different, enabling the batching of email push tasks and further reducing server pressure.

[0020] Optionally, historical interaction data can be analyzed to determine the recipient's optimal active window, including: dividing the preset period into multiple sub-periods and dividing the sub-periods into multiple time periods according to the time dimension; Count the frequency of various interactive behaviors that occur within the same time period across multiple sub-cycles; Based on the preset weights and frequencies of various interactive behaviors, the activity score for each time period is calculated, and the time period with the highest activity score is taken as the optimal activity window.

[0021] The system divides a preset period into multiple sub-periods, and further into time periods. Frequency statistics and weighted scoring are performed on interactive behaviors within different time periods to construct a time-based activity assessment model. By setting preset weights for different types of interactive behaviors, the system can distinguish the relative importance of behaviors such as replying, clicking, and opening in activity assessment, avoiding biases caused by simple counting.

[0022] Optionally, historical interaction data can be analyzed to determine the recipient's optimal active window, including: dividing the preset period into multiple sub-periods and dividing the sub-periods into multiple time periods according to the time dimension; Based on the dates corresponding to the sub-cycles, the historical interaction behavior data is divided into holiday datasets and weekday datasets; Calculate the recipient's activity level on holidays and weekdays respectively, and determine the recipient's activity preferences accordingly; Based on activity preferences, activity scores for each time period are calculated in the corresponding dataset to determine the optimal activity window.

[0023] By dividing historical interaction data into holiday datasets and weekday datasets, and analyzing the differences in recipient behavior under different life or work conditions, the activity values ​​for holidays and weekdays are calculated separately, and the recipient's activity preferences are determined accordingly. This avoids mixing data with large differences in behavior patterns, thereby improving the accuracy of determining the optimal activity window.

[0024] Optionally, the activity value of the recipient during holidays is calculated, including: counting the number of various interactive behaviors generated in each sub-period of the holiday data; for any sub-period, the local interaction volume is obtained by summing the product of the number of various interactive behaviors and its preset weight coefficient, and the average of the local interaction volume of each sub-period is taken as the activity value of the holiday.

[0025] We perform weighted statistics on the interaction behaviors within each sub-period of the holiday data set and take the average as the holiday activity value. This makes the assessment of holiday activity reflect the overall trend, smooths the impact of occasional high-frequency or low-frequency behaviors on the results, and improves the stability of activity preference judgment.

[0026] Optionally, the email push task is executed using the sending domain name based on the execution time, including: constructing a sending time window based on the execution time; executing the email push task in response to the current real-time system time being within the sending time window, otherwise returning the task to the queue to wait; wherein, the task within the sending time window is sent earlier or later within the window period depending on the server load.

[0027] A sending time window is established, and email pushes are executed or delayed within the window based on the real-time system status, making the push process time-flexible. On the one hand, pushes are only executed when the system time is within the sending time window, ensuring that email sending matches the recipient's active period; on the other hand, sending emails earlier or later within the window period based on server load can effectively reduce system pressure caused by excessive instantaneous concurrency.

[0028] Secondly, this application provides an email push system based on an online marketing platform, employing the following technical solution: An email push system based on an online marketing platform includes: a memory storing computer program instructions, which, when executed by a processor, implement the email push method based on the online marketing platform described above.

[0029] The above-mentioned email push method based on a network marketing platform is generated into a computer program and stored in a memory so that it can be loaded and executed by a processor. Thus, a system is created based on the memory and processor for easy use.

[0030] This application has the following technical effects: This application initially reduces the workload of email push tasks by filtering and verifying them, while also decreasing the progress of non-compliant email push tasks. Then, it quantifies the recipient's historical response rate, dynamically classifies sending resources into different reputation levels, and prioritizes high-reputation domains to serve highly active users, protecting domain reputation and improving delivery rates while reducing server load. Attached Figure Description

[0031] Figure 1 This is a flowchart illustrating an email push method based on an online marketing platform, according to an embodiment of this application.

[0032] Figure 2 This is a flowchart of an email push method based on a network marketing platform according to an embodiment of this application, mainly used to illustrate the filtering and verification logic.

[0033] Figure 3 This is a flowchart of step S2 in an email push method based on a network marketing platform according to an embodiment of this application. Detailed Implementation

[0034] This application discloses an email push method based on an online marketing platform, which aims to solve technical problems such as low delivery rate and uneven server pressure distribution in large-scale email marketing.

[0035] Reference Figure 1 A method for sending emails based on an online marketing platform, comprising steps S1-S3.

[0036] S1: In response to the creation of an email push task, filter and verify the email push task to determine whether the email push task meets the preset requirements.

[0037] Understandably, the email push process begins with the creation of an email push task, which is proactively created by the marketing platform operator or brand promoter.

[0038] After the email push task is created, in order to achieve automated push, the system is configured with a timed task triggering module, which is a component used to issue execution instructions at a preset time frequency. This module generates trigger signals at regular intervals, driving the task distribution and scheduling unit to distribute the email push task to the task storage buffer queue in an orderly manner for subsequent processing.

[0039] After an email push task is created, it primarily includes the following information: email title, email body, and recipient information. The email push task is validated by invoking multiple pre-defined filters. For example, a content feature filter performs semantic scanning on the email title and body. In this example, the filter identifies highly sensitive promotional advertising terms such as "super low price," "replica of big brands," and "up to 90% off" in the body. Because these terms match and conform to the behavioral characteristics of spam, the system classifies it as suspected spam and triggers interception or manual review. In some embodiments, the filter chain may also include an identity status filter, such as a blacklist filter, used to verify whether the target recipient's contact information is on the sender's blocked list.

[0040] In conjunction with example 2, in response to any filter determining that the task does not meet the sending requirements and returning a signal to refuse sending, the exception handling logic is triggered. Detailed reasons for rejection are recorded, such as "target address is blacklisted" or "current domain sending frequency is too high," and the subsequent process of the email push task is terminated, with no further domain routing or actual sending. This pluggable design ensures that the system can quickly adapt to different countries' anti-spam laws (such as the EU General Data Protection Regulation / GDPR) by adding specific compliance check modules without refactoring the underlying logic of the entire push system.

[0041] S2: In response to the email push task passing the filter verification, the sending domain name is assigned according to the interaction performance of the corresponding recipient within the preset period of the email push task.

[0042] Reference Figure 3 Step S2 includes steps S21-S22.

[0043] S21: Obtain historical interaction data between the recipients and the marketing platform corresponding to the email push task within a preset period, and calculate the historical response rate of the target recipients.

[0044] In this embodiment, the calculation of the target recipient's proximity is obtained. Historical interaction data within the day refers to the recipient's behavior of opening, reading, or replying to emails sent by the online marketing platform, and calculates their historical response rate metric.

[0045] The reason for constructing a historical response rate metric is that by quantifying the recipient's level of responsiveness to past push notifications, the interaction quality can be accurately assessed, thus serving as a basis for allocating high-value sending resources (such as more reputable domain addresses). The specific calculation formula is described below: ; In the formula, Historical response rate to the target recipient; This represents the recipient's past The number of emails received and replied to within one day; This represents the total number of emails sent to this recipient within the same period, or the total number of emails received by the recipient. It's important to note that if the recipient is a new user, there may be no historical interaction data. In such cases, the recipient's historical response rate is set to 0.

[0046] It is understandable that when the historical response rate increases, it means that the recipient is interacting more actively with the platform and is a high-quality user, thus leading to an increase in the calculated historical response rate.

[0047] S22: In response to a historical response rate greater than or equal to a preset ratio threshold, any one of the first type of sending domain names in the preset sending domain name pool is assigned to the email push task; otherwise, any one of the second type of sending domain names in the sending domain name pool is assigned to the email push task.

[0048] In this embodiment, the preset ratio threshold is set to... If a recipient's historical response rate is greater than or equal to a preset threshold, the recipient is determined to be a high-response recipient, and a domain name is selected from the first type of sending domain name pool (i.e., a set of domain names with high credibility) and assigned to the email push task. If the historical response rate is lower than the preset threshold, a domain name will be randomly matched from the preset second type of sending domain name pool for the email push task. Regarding the domain name's credibility, the total number of user complaints (report spam) generated by the domain name within a preset period is counted, and the ratio of this number to the total number of successful deliveries is taken as the complaint rate. The difference between 1 and the complaint rate is taken as the domain name's credibility.

[0049] In this embodiment, the preset ratio threshold is set to When the threshold is set too low, a large number of email delivery tasks to low-quality recipients will consume valuable sending domains, increasing the risk of damage to the sending domain's reputation. Conversely, if the threshold is set too high, it will lead to insufficient utilization of valuable sending domains. Therefore, controlling the ratio within a certain range is crucial. This dynamic matching mechanism, based on interactive data, maximizes the delivery effectiveness of high-quality email domains while protecting their reputation. It ensures that emails to high-quality recipients are delivered through a more stable network channel, effectively improving the final delivery rate and thus facilitating cooperation between senders and recipients.

[0050] S3: Determine the execution time of the email push task, and execute the email push task using the sending domain name according to the execution time.

[0051] In the step of determining the execution time of the email push task, the recipient's historical interaction behavior data within a preset period is first obtained. The interaction behavior data includes at least one of the following: email opening time, link clicking time, and reply sending time.

[0052] In this embodiment, the recipient's information is retrieved within a preset period (such as near). Historical interaction data within (days), including but not limited to email opening time, link click time, and reply sending timestamp.

[0053] Then, the historical interaction data is analyzed to determine the recipient's optimal active window; the time point within the optimal active window is selected as the execution time.

[0054] In one embodiment, historical interaction data is analyzed to determine the recipient's optimal activity window; the time point selected within the optimal activity window as the execution time can be determined in the following way: The preset period is divided into multiple sub-periods. In this embodiment, the sub-period length is 24 hours, that is, one day is one sub-period. Then, the sub-period is divided into multiple time periods by hour, and the frequency of interaction behavior data within the same time period of multiple sub-periods is counted to determine the optimal active window.

[0055] For any time period within any sub-period of the preset period, calculate its comprehensive activity score, and take the time period with the highest comprehensive activity score as the optimal activity window.

[0056] In the formula, This represents the overall activity score over a given time period. Indicates the first Preset weights for interactive behaviors, This indicates that each sub-period occurs within the same time period. The number of interactive behaviors can also be understood as the number of times a certain type of interaction occurs within the same time period corresponding to multiple sub-periods. The number of interactive behaviors; in this embodiment, the weight of the reply behavior is set to Click behavior set Open behavior set ; This indicates the total number of interaction behavior categories.

[0057] After obtaining the optimal active window, the execution time is taken from a specific time point within the optimal active period. In this embodiment, the execution time is taken from the initial time of the optimal active period. For example, the optimal active period for a certain recipient is calculated as follows: to Then the execution time is Finally, based on the preset dates of this marketing campaign, calculate the specific time points falling within this active period and set them as the execution time. For example, the execution time can be set to the day before the campaign. .

[0058] In another embodiment, historical interaction data is analyzed to determine the recipient's optimal active window; the time point selected within the optimal active window as the execution time can also be obtained through the following methods.

[0059] Similarly, the preset period is first divided into multiple sub-periods. In this embodiment, the sub-period length is 24 hours, meaning one day is one sub-period. Then, the sub-period is divided into multiple time periods based on the hourly dimension. Next, the sub-periods are categorized based on the corresponding dates. For example, in this embodiment, historical interaction data from the recipient's past 90 days is collected, and the sub-periods are divided into holidays and weekdays.

[0060] After classifying the data, we get: Holiday Dataset: Contains interaction records for statutory holidays and weekends.

[0061] Weekday dataset: Contains interaction records from Monday to Friday (excluding statutory holidays).

[0062] This classification method takes into account the differences in the recipient's behavioral patterns under different life circumstances, avoiding the mixing of two completely different behavioral characteristics, and providing underlying data support for subsequent accurate profiling.

[0063] Subsequently, activity scores for holidays and overall activity scores for weekdays were calculated to determine the recipient's activity preferences.

[0064] Specifically, the formula for calculating holiday activity value can be expressed as: ; In the formula, This indicates the recipient's activity level during holidays; This indicates the number of self-periodic events in the holiday data set, which can also be understood as the total number of holiday days. Indicates the first Preset weight coefficients corresponding to class-specific interactive behaviors; Represents the first in the holiday data set Within each sub-cycle The number of times a class interaction occurs; This indicates the total number of interaction behavior categories.

[0065] The formula for calculating holiday activity value can be expressed as: ; In the formula, This indicates the recipient's daily activity level. This indicates the total number of working days within the preset period; Indicates the first Preset weight coefficients corresponding to class-specific interactive behaviors; : indicates the first in the weekday dataset Within a day, the The number of times a class interaction occurs; This indicates the total number of interaction behavior categories.

[0066] If the activity level on a holiday is higher than that on a weekday, the recipient's activity preference is determined to be a holiday. A holiday activity score is calculated based on the recipient's historical interaction data for holidays, and the execution time is determined based on this score and the recipient's holiday preference. Conversely, if the activity level on a holiday is less than or equal to that on a weekday, the recipient's activity preference is determined to be a weekday. A weekday activity score is calculated based on the recipient's historical interaction data for weekdays, and the execution time is determined based on this score and the recipient's holiday preference.

[0067] Similar to the calculation of the overall activity score, the frequency of interactive behavior data in multiple sub-periods within the holiday dataset is statistically analyzed within the same time period to determine the optimal activity window.

[0068] For any given time period, calculate its overall activity score, and select the time period with the highest overall activity score as the optimal activity window.

[0069] In the formula, The activity score during holidays indicates the time period. Indicates the first Preset weights for interactive behaviors, This indicates that the events in each sub-period of the holiday dataset occur within the same time period. The number of interactive behaviors can also be understood as the number of times multiple sub-periods in the holiday dataset occur within the same time period. The number of interactive behaviors; in this embodiment, the weight of the reply behavior is set to Click behavior set Open behavior set ; This indicates the total number of interaction behavior categories.

[0070] After obtaining the optimal active window, the execution time is taken from a specific time point within the optimal active period. In this embodiment, the execution time is taken from the initial time of the optimal active period. For example, the optimal active period for a certain recipient is calculated as follows: to Then the execution time is .

[0071] In one embodiment, the email body can be adjusted and sent directly after the system's real-time time reaches the execution time.

[0072] In another embodiment, a sending time window can be constructed based on the execution time. In response to the email push task entering the sending time window, the email body is adjusted and sent.

[0073] The sending time window is then constructed based on the execution time. In this embodiment, the sending time window is 1800 seconds before and after the execution time.

[0074] Subsequently, the system obtains the current real-time system time. Since the current real-time system time falls within the sending time window, the system considers the email push task executable and proceeds to the next email sending step. Tasks outside the sending time window are returned to the Redis queue to wait, thereby reducing the workload of email push tasks currently being executed synchronously and alleviating the instantaneous communication pressure on the server.

[0075] If the distance between the current system's real-time time and execution time is greater than If the timer is less than 1 second, it indicates that the target influencer's active window has not yet been reached. The system then triggers the task suspension logic, re-entering the task into the task queue for delayed scheduling. Email push tasks that have entered the sending time window can be sent up to 30 minutes in advance within the sending time window. The system can utilize the server's idle bandwidth to pre-process some queued tasks up to 30 minutes before the scheduled peak arrives.

[0076] If the server reaches its processing bottleneck, subsequent email push tasks will automatically queue and wait for the remaining time within the window. Email push tasks entering the sending time window can be delayed for up to 30 minutes within the sending time window. When the server reaches its processing bottleneck, subsequent tasks will automatically queue and wait for the remaining time within the window, and the system will not crash due to instantaneous request overload.

[0077] The purpose of adjusting the email body is to transform standardized email templates into personalized content tailored to specific recipients, and to incorporate tracking and security mechanisms.

[0078] Specifically, the personalization and security enhancement processing of email text mainly includes: sending address construction, nickname construction, and insertion of behavioral tracking features.

[0079] For sending address construction, the user-preset custom sender name identifier is first retrieved. Subsequently, the complete sender's email address is constructed using string concatenation logic. : In the formula, This represents the final generated sender email address; Represents the sender's username; This represents the sending domain name corresponding to the email push task.

[0080] For nickname construction, the main task is to replace the nickname in the email template body. The recipient's display name is extracted and used to replace the nickname variable in the template body. During the replacement process, to prevent special characters in the display name from disrupting the email template's rendering structure, a character escaping unit is invoked to perform regular expression escaping on the selected name, generating a safe name for replacement. For example, if the recipient's display name contains special punctuation, the escaping unit will process its characteristic symbols to ensure that it is treated as plain text content rather than program control instructions during the replacement process.

[0081] Behavioral tracking feature insertion refers to appending tracking feature labels to the end of the email body. Specifically, a hidden input label is generated, whose numerical attribute contains a unique identifier for the current email message. ;in, This represents the final generated email body content; Represents the content of the email body; This represents a hidden tracking tag that generates a unique identifier for the current email message. When the recipient interacts with the message, the system identifies this tag to accurately track email opening, clicking, and replying actions.

[0082] Finally, after completing the dynamic generation and security enhancement of the email content, the system enters the final push execution and behavior tracking phase. This step aims to securely deliver the email through the selected sending channel and establish a complete feedback data return mechanism to evaluate the push effect. The specific execution logic is as follows: First, identify the sending domain selected for the email sending task in step S2. Based on the domain's configuration information, call the corresponding push service interface to execute the actual push action.

[0083] After the email push action is completed, the lifecycle status of this email push task is updated in real time. Specifically, the sending status corresponding to the unique identifier of the email for this email push task is changed to "sent". The execution time and success feedback indicator in the push task details table are updated.

[0084] To ensure system stability, a communication rate adjustment module fine-tunes the instantaneous sending frequency to prevent triggering the recipient server's traffic limiting protocol. Simultaneously, the system utilizes an asynchronous message processing framework (such as RocketMQ) to establish a feedback listening channel. When the target recipient interacts (e.g., clicks or replies), the system identifies hidden tracking tags injected into the email body to precisely correlate the feedback behavior with the original push task.

[0085] The system collects and quantitatively evaluates recipient feedback behavior, calculating the latest interaction frequency of the target influencer. Specifically, based on newly generated reply records, the historical reply rate is recalculated. The updated data is written back to the storage unit, serving as the technical basis for subsequent push tasks when executing step S2, thereby enabling self-correction and continuous optimization of the push strategy.

[0086] This application also discloses an email push system based on an online marketing platform, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, an email push method based on an online marketing platform according to this application is implemented.

[0087] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0088] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for sending emails based on an online marketing platform, characterized in that, include: In response to the creation of an email push task, the email push task is filtered and verified to determine whether the email push task meets the preset requirements. In response to the email push task passing the filtering verification, the sending domain name is assigned according to the interaction performance of the corresponding recipient within a preset period of the email push task. Determine the execution time of the email push task, and execute the email push task using the sending domain name according to the execution time; The process of allocating sending domains based on the interaction performance of recipients within a preset period for email push tasks includes: obtaining historical interaction data between recipients and marketing platforms within a preset period for email push tasks, and calculating the historical response rate of target recipients; in response to a historical response rate greater than or equal to a preset proportion threshold, allocating any sending domain from a preset first-class sending domain pool to the email push task, otherwise allocating any sending domain from a second-class sending domain pool to the email push task, wherein the average reputation of sending domains in the first-class sending domain pool is greater than the average reputation of sending domains in the second-class sending domain pool.

2. The email push method based on a network marketing platform according to claim 1, characterized in that, Filtering and verifying email push tasks should include at least the following: detecting whether there are sensitive words in the email body and / or email title, and detecting whether the recipient is a blocked target of the sender.

3. The email push method based on a network marketing platform according to claim 1, characterized in that, Calculating the historical response rate of the target recipient includes: obtaining the number of times the target recipient made a valid response within a preset period from the historical interaction data between the recipient and the marketing platform, as the response count value; obtaining the total number of emails sent by the marketing platform to the target recipient, as the sending count value; and using the ratio of the response count value to the sending count value as the recipient's historical response rate.

4. The email push method based on a network marketing platform according to claim 1, characterized in that, The credibility of a domain name is negatively correlated with the total number of user complaints generated during the use of that domain name.

5. The email push method based on a network marketing platform according to claim 1, characterized in that, Determining the execution time of the email push task includes: obtaining the recipient's historical interaction behavior data within a preset period, wherein the interaction behavior data includes at least one of the following: email opening time, link clicking time, and reply sending time; Analyze historical interaction data to determine the recipient's optimal active window, and select a time point within the optimal active window as the execution time.

6. The email push method based on a network marketing platform according to claim 5, characterized in that, Analyze historical interaction data to determine the recipient's optimal active window, including: dividing the preset period into multiple sub-periods, and dividing the sub-periods into multiple time periods according to the time dimension; Count the frequency of various interactive behaviors that occur within the same time period across multiple sub-cycles; Based on the preset weights and frequencies of various interactive behaviors, the activity score for each time period is calculated, and the time period with the highest activity score is taken as the optimal activity window.

7. The email push method based on a network marketing platform according to claim 5, characterized in that, Analyze historical interaction data to determine the recipient's optimal active window, including: dividing the preset period into multiple sub-periods, and dividing the sub-periods into multiple time periods according to the time dimension; Based on the dates corresponding to the sub-cycles, the historical interaction behavior data is divided into holiday datasets and weekday datasets; Calculate the recipient's activity level on holidays and weekdays respectively, and determine the recipient's activity preferences accordingly; Based on activity preferences, activity scores for each time period are calculated in the corresponding dataset to determine the optimal activity window.

8. The email push method based on a network marketing platform according to claim 7, characterized in that, Calculate the recipient's activity value during holidays, including: counting the number of various interactive behaviors generated in each sub-period of the holiday data set; for any sub-period, summing the products of the number of various interactive behaviors and their preset weight coefficients to obtain the local interaction volume, and taking the average of the local interaction volumes of each sub-period as the activity value during holidays.

9. The email push method based on a network marketing platform according to claim 1, characterized in that, The process of executing an email push task using the sending domain name based on the execution time includes: constructing a sending time window based on the execution time; executing the email push task if the current real-time system time is within the sending time window, otherwise returning the task to the queue to wait; wherein, tasks within the sending time window are sent earlier or later within the window period depending on the server load.

10. An email push system based on an online marketing platform, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement an email push method based on a network marketing platform according to any one of claims 1-9.

Images