AI-based intelligent outbound marketing analysis system

Abstract

The present invention discloses an AI-based intelligent outbound marketing analysis system, which belongs to the technical field of outbound call systems. The AI-based intelligent outbound marketing analysis system includes a number screening module, an information push module, a dialing strategy module, and an outbound call module. The outbound call module is connected to a secondary call screening module, and the secondary call screening module depicts a character portrait. The secondary call screening module is connected to a nesting module, and the user is screened twice through the nesting module. The outbound call module performs call answering through an AI module, and the AI ​​module is connected to a knowledge base. The conversation content is analyzed twice through the secondary call screening module, keywords are extracted, and the user's intention is assessed through the character portrait through the nesting module, and the secondary call marketing content is adjusted. Potential customers are further screened out, which is convenient for improving the order conversion rate.

Description

Technical Field

[0001] The present invention belongs to the technical field of outbound call systems, and specifically relates to an AI-based intelligent outbound call marketing analysis system. Background Art

[0002] Outbound calls refer to the automatic outbound calls made by a phone through a computer, and the recorded voice is played to the user through the computer. It is an indispensable part of the modern customer service center system that integrates computer and telephony. The outbound marketing system is a platform specially designed based on the needs of market customers and the different needs of different enterprises and manufacturers in various industries.

[0003] For example, a method and device for configuring outbound call rules proposed in CN109636302A include: obtaining user tags configured according to user behavior data; classifying users according to the configured user tags; configuring an outbound call identifier for each user classification and configuring the outbound call rules of the user classification according to the user tags in the user classification. Through the technical solution of the present disclosure, outbound call rules that meet the user characteristics are configured for different types of users, and outbound calls are made to users according to the configured outbound call rules, thereby achieving targeted outbound calls to users, which can improve the call connection rate of the agents and the work efficiency of the agents. However, when configuring the outbound call speech rules according to the user type, there will be certain errors in the data, and during the conversation, AI customer service can easily determine high-intent customers, which is convenient for switching to manual customer service to quickly place orders, but it is difficult to further divide and distinguish the user's intentions, and it is impossible to quickly screen out potential target users based on the conversation. Summary of the invention

[0004] The purpose of the present invention is to provide an AI-based intelligent outbound marketing analysis system to solve the problems encountered by the existing outbound marketing analysis system during use as mentioned in the above background technology.

[0005] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: an AI-based intelligent outbound marketing analysis system, comprising a number screening module, an information push module, a dialing strategy module, and an outbound call module, the outbound call module being connected to a secondary call screening module, the secondary call screening module portraying a character portrait, the secondary call screening module being connected to a nesting module, and performing secondary screening of users through the nesting module, the outbound call module performing call answering through an AI module, the AI ​​module being connected to a knowledge base, the outbound call module being connected to a manual customer service intervention module, the manual customer service intervention module being connected to an emotion monitoring module for monitoring the emotions of manual customer service, and the emotion monitoring module and the outbound call module being both connected to a fuse module.

[0006] Preferably, the number screening module is used to retrieve numbers from a database, screen the numbers, eliminate empty numbers, wrong numbers, and obtain the number's location. The number screening module can be optionally connected to a user portrait library, and the user portrait label, number, and number's location are passed to the dialing strategy module through the number screening module.

[0007] Preferably, the information push module is used to send information to the user as a reminder before dialing, and send the service personnel's phone number or website link to the user after the call is completed.

[0008] Preferably, the dialing strategy module selects an outbound call time period for dialing based on the assigned number and number location, and further based on the user portrait, and sends the user portrait label, number, and number location to the AI ​​module.

[0009] Preferably, the AI ​​module extracts relevant words from the knowledge base according to the user portrait tag, number, number location, and dialing time period, and communicates with the user through the outbound call module.

[0010] Preferably, the outbound call module includes a speech analysis module, a recording module, and a sentiment analysis module. The outbound call module transmits the analysis results of the speech analysis module and the sentiment analysis module to the AI ​​module. The AI ​​module retrieves or switches the speech from the knowledge base according to the analysis results. The recording module is used to record and transcribe the call, and send the call content to the knowledge base for storage.

[0011] Preferably, the speech analysis module analyzes and understands the user's answer, and the emotion analysis module is optionally independent of the speech analysis module to capture and analyze the user's voice, intonation, and speaking speed.

[0012] Preferably, the secondary call screening module obtains the conversation content from the recording module, performs secondary analysis and retrieval on the user's speech after the call, extracts key words, and creates a character profile of the user, and optionally retrieves the personality analysis results of the emotion analysis module from the outbound call module.

[0013] Preferably, the nested module is pre-set with a character portrait of the target user and corresponding label keywords. If the keywords or labels extracted in the secondary call screening module trigger the character portrait content in the nested module, the trigger information and the character portrait content are sent to the dialing strategy module. The dialing strategy module selects a time period for a secondary call based on the character portrait content.

[0014] Preferably, the fuse module is used to be triggered when the user conveys negative information, low-intention information, or when the manual customer service conveys negative information, announces the ending words, and forcibly hangs up. The emotion monitoring module is used to monitor the emotions of the manual customer service and trigger the fuse module. The manual customer service intervention module is used to switch to the manual customer service when the speech analysis module captures the intention words.

[0015] Compared with the prior art, the present invention has the following beneficial effects:

[0016] The second call screening module conducts a second analysis of the conversation content, extracts keywords, and uses nested modules to create character portraits to assess user intentions. The second call marketing content is adjusted to further screen out potential customers, thereby increasing the order conversion rate.

[0017] The outbound call module can extract multi-dimensional information about users, mine users’ semantics and emotions based on conversation information, further capture keywords, and further divide customers into high-intent, medium-intent, and low-intent groups based on the number of overlaps with the intended keywords or tags of the nested modules.

[0018] The manual customer service intervention module is further set up to intervene in the outbound call dialogue status, so as to capture high-intent customers in the dialogue, quickly communicate detailed content, and help to close the order. A fuse module is also set up to conduct two-way monitoring to ensure that the dialogue content can be carried out in high quality according to the predetermined script content, which can not only protect the emotions of customer service, but also prevent the dissatisfaction of manual customer service from affecting customer relations. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic diagram of the present invention.

[0020] In the figure: 1. Number screening module; 2. Information push module; 3. Dialing strategy module; 4. Outbound call module; 5. AI module; 6. Secondary call screening module; 7. Nesting module; 8. Fuse module; 9. Manual customer service intervention module; 10. Emotion monitoring module; 11. Knowledge base. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] See also Figure 1The present invention provides a technical solution: an AI-based intelligent outbound marketing analysis system, including a number screening module 1, an information push module 2, a dialing strategy module 3, and an outbound module 4. The outbound module 4 is connected to a secondary call screening module 6, which depicts a character portrait. The secondary call screening module 6 is connected to a nesting module 7, and the user is screened twice through the nesting module 7. The outbound module 4 performs call answering through an AI module 5. The AI ​​module 5 is connected to a knowledge base 11. The outbound module 4 is connected to an artificial customer service intervention module 9. The artificial customer service intervention module 9 is connected to an emotion monitoring module 10 for monitoring the emotions of the artificial customer service. The emotion monitoring module 10 and the outbound module 4 are both connected to a fuse module 8.

[0023] In this implementation scheme, since there are differences in character portrait analysis through the database and personal data is private information, the user's intention is assessed through the second call screening module 6, and high-intention, medium-intention, and low-intention customer groups can be divided, which is convenient for adjusting the second call marketing content and further screening out potential customers, so as to improve the order conversion rate.

[0024] The artificial customer service intervention module 9 further provided intervenes in the outbound call dialogue status, so as to capture high-intent customers in the dialogue, quickly communicate the detailed content, and help to close the order.

[0025] A fuse module 8 is also provided for two-way monitoring to ensure that the content of the conversation can be conducted in high quality in accordance with the predetermined script. This can not only protect the emotions of the customer service, but also prevent the dissatisfaction of the manual customer service from affecting the customer relationship. The fuse module 8 monitors the customer's speech content and the manual customer service's speech content. When a keyword is triggered, the ending remarks are automatically broadcast to forcibly close the conversation.

[0026] Specifically, the number screening module 1 is used to retrieve numbers from the database, screen the numbers, eliminate empty numbers and wrong numbers, and obtain the number's location. The number screening module 1 can optionally be connected to a user portrait library, and the user portrait label, number, and number's location are passed to the dialing strategy module 3 through the number screening module 1.

[0027] In this embodiment, the database here refers to the number database, and the user portrait library may optionally include information such as gender, age, consumption preferences and even occupation corresponding to the user number. Since the user portrait library is often difficult to obtain, and the data is likely to have certain errors compared with the individual's consumption intentions, and may even involve user privacy issues, it is not used here as the main reference source of data, but only as an optional reference analysis item.

[0028] The number screening module 1 screens out empty numbers and wrong numbers, and passes some numbers that are difficult to determine whether they are empty numbers or wrong numbers to the dialing strategy module 3, so that it can be dialed during non-working hours, such as after 5 pm.

[0029] Specifically, the information push module 2 is used to send information to the user as a reminder before dialing, and send the service personnel's phone number or website link to the user after the call is completed.

[0030] In this embodiment, the user is reminded of information before dialing. When the nested module 7 sends a secondary dialing instruction to the dialing strategy module 3, the dialing strategy module 3 can send an information reminder again through the information push module 2 before making an outbound call through the outbound call module 4.

[0031] The information here can be information push from the APP or text messages.

[0032] Specifically, the dialing strategy module 3 selects an outbound call time period for dialing based on the assigned number and the number's location, and further based on the user portrait, and sends the user portrait label, number, and number location to the AI ​​module 5.

[0033] In this embodiment, the dialing strategy module 3 selects a dialing time period according to the assigned number and the location of the number.

[0034] Specifically, the AI ​​module 5 extracts relevant words from the knowledge base 11 according to the user portrait label, number, number location, and dialing time period, and communicates with the user through the outbound call module 4.

[0035] In this embodiment, the knowledge base 11 is used to store the speech scripts and corresponding voice information, and can also record and store the conversation information recorded by the outbound call module 4 to facilitate case analysis, and can also provide marketing information search function to manual customer service.

[0036] Specifically, the outbound call module 4 includes a speech analysis module, a recording module, and a sentiment analysis module. The outbound call module 4 passes the analysis results of the speech analysis module and the sentiment analysis module to the AI ​​module 5. The AI ​​module 5 retrieves or switches the speech from the knowledge base 11 according to the analysis results. The recording module is used to record the call and transcribe it into text, and send the call content to the knowledge base 11 for storage.

[0037] In this embodiment, the speech analysis module and the emotion analysis module here use corresponding convolutional neural network analysis models to ensure the response speed of voice dialogue, which are fast but have low accuracy.

[0038] Specifically, the speech analysis module analyzes and understands the user's answer, and the sentiment analysis module is optionally independent of the speech analysis module to capture and analyze the user's voice, intonation, and speaking speed.

[0039] In this embodiment, the emotion analysis module may be optionally independent of the speech analysis module, and is used to capture and analyze the user's voice, intonation, and speech speed, and to analyze and judge the user's emotions.

[0040] Specifically, the secondary call screening module 6 obtains the conversation content from the recording module, performs secondary analysis and retrieval on the user's words after the call is over, extracts key words, and creates a character profile of the user, and optionally retrieves the personality analysis results of the emotion analysis module from the outbound call module 4.

[0041] In this embodiment, the secondary call screening module 6 here can select a more accurate analysis model, and can be optionally connected to a learning database, which can perform machine self-learning based on the dialogue library, and establish more accurate analysis data based on the CNN convolutional neural network algorithm to capture keywords.

[0042] Specifically, the nested module 7 is pre-set with the character portrait of the target user and the corresponding label keywords. If the keywords or labels extracted in the secondary call screening module 6 trigger the character portrait content in the nested module 7, the trigger information and the character portrait content are sent to the dialing strategy module 3. The dialing strategy module 3 selects a time period for a secondary call based on the character portrait content.

[0043] In this embodiment, the nested module 7 is provided with user personas of the target consumer group corresponding to the marketing product, and user personas of the corresponding high-intention group. The keywords screened out by the secondary call screening module 6 can be retrieved in the nested module 7. According to the number of keywords or tags retrieved in the nested module 7, the users are divided into high-intention customers, medium-intention customers, and low-intention customers. The corresponding number user personas and the high, medium and low-intention customer groups to which they belong are passed to the dialing strategy module 3 for secondary dialing or even multiple dialing. Here, the low-intention customer group can be set to not dial.

[0044] Specifically, the fuse module 8 is used to be triggered when the user conveys negative information, low-intention information, or when the manual customer service conveys negative information, announces the ending words, and forcibly hangs up. The emotion monitoring module 10 is used to monitor the emotions of the manual customer service and trigger the fuse module 8. The manual customer service intervention module 9 is used to switch to the manual customer service when the voice analysis module captures the intention words.

[0045] In this embodiment, the negative information here can be uncivilized language analyzed by the voice analysis module, and the low-intention information includes information that the user says no, refuses, etc. to clearly refuse sales promotion, and the intention words can be words such as good and can.

[0046] The emotion monitoring module 10 is used to monitor the emotion of the manual customer service. The manual customer service intervention module 9 here can also make a character portrait of the customer service so as to allocate customers in a targeted manner.

[0047] The method of applying the above AI-based intelligent outbound marketing analysis system specifically includes the following steps:

[0048] The number screening module 1 retrieves numbers from the database, screens the numbers, removes empty numbers and wrong numbers, and obtains the number's location. The number, number's location and user portrait library are transmitted to the dialing strategy module 3 through the number screening module 1. For some numbers that are difficult to determine whether they are empty numbers or wrong numbers, they are transmitted to the dialing strategy module 3 so that they can be dialed during non-working hours;

[0049] The information push module 2 sends information to the user to remind him before dialing;

[0050] The dialing strategy module 3 selects an outbound call time period for dialing according to the assigned number and the number's location, and sends the user portrait label, number, and number's location to the AI ​​module 5;

[0051] The AI ​​module 5 extracts relevant words from the knowledge base 11 according to the user portrait label, number, number location, and dialing time period, and communicates with the user through the outbound call module 4;

[0052] The outbound call module 4 transmits the analysis results of the voice analysis module and the sentiment analysis module to the AI ​​module 5. The AI ​​module 5 retrieves or switches the speech from the knowledge base 11 according to the analysis results. When the voice analysis module captures the intended words, the manual customer service intervention module 9 switches to the manual customer service. The fuse module 8 is triggered when the user transmits negative information or low-intent information or when the manual customer service transmits negative information, announces the ending words, and forcibly hangs up.

[0053] After the call is completed, the information push module 2 sends the service personnel's phone number or website link to the user;

[0054] The secondary call screening module 6 obtains the conversation content from the recording module, performs secondary analysis and retrieval on the user's speech after the call is over, extracts key words, and creates a character profile for the user, and optionally retrieves the character analysis results of the emotion analysis module from the outbound call module 4;

[0055] If the keywords or tags extracted in the secondary call screening module 6 trigger the character portrait content in the nesting module 7, the trigger information and the character portrait content will be sent to the dialing strategy module 3, and the dialing strategy module 3 will select a time period for a secondary call based on the character portrait content.

[0056] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. AI-based intelligent outbound marketing analysis system, characterized by: The invention comprises a number screening module (1), an information push module (2), a dialing strategy module (3), and an outbound call module (4); the outbound call module (4) is connected to a secondary call screening module (6); the secondary call screening module (6) depicts a character portrait; the secondary call screening module (6) is connected to a nesting module (7), and performs secondary screening on the user through the nesting module (7); the outbound call module (4) performs call answering through an AI module (5); the AI ​​module (5) is connected to a knowledge base (11); the outbound call module (4) is connected to an artificial customer service intervention module (9); the artificial customer service intervention module (9) is connected to an emotion monitoring module (10) for monitoring the emotion of the artificial customer service; the emotion monitoring module (10) and the outbound call module (4) are both connected to a fuse module (8); The number screening module (1) is used to retrieve numbers from a database, screen the numbers, remove empty numbers and wrong numbers, and obtain the number's location. The number screening module (1) is connected to a user portrait library. The user portrait label, number, and number's location are transmitted to the dialing strategy module (3) through the number screening module (1); The information push module (2) is used to send a message to the user as a reminder before dialing, and to send the user a service personnel telephone number or a website link after the call is completed; The dialing strategy module (3) selects an outbound call time period for dialing based on the assigned number and the number's location, and further based on the user profile, and sends the user profile tag, number, and number's location to the AI ​​module (5); The AI ​​module (5) extracts relevant speech from the knowledge base (11) according to the user portrait tag, number, number location, and dialing time period, and conducts a dialogue with the user through the outbound call module (4); The outbound call module (4) comprises a speech analysis module, a recording module and a sentiment analysis module. The outbound call module (4) transmits the analysis results of the speech analysis module and the sentiment analysis module to the AI ​​module (5). The AI ​​module (5) retrieves or switches the speech from the knowledge base (11) according to the analysis results. The recording module is used to record and transcribe the call, and send the call content to the knowledge base (11) for storage. The speech analysis module analyzes and understands the user's answer, and the emotion analysis module is independent of the speech analysis module and is used to capture and analyze the user's voice, intonation, and speaking speed; The secondary call screening module (6) obtains the conversation content from the recording module, performs secondary analysis and retrieval on the user's speech after the call is over, extracts key words, and creates a character profile of the user, and retrieves the personality analysis result of the emotion analysis module from the outbound call module (4); The nesting module (7) is pre-set with a character portrait of the target user and corresponding tag keywords. If the keywords or tags extracted in the secondary call screening module (6) trigger the character portrait content in the nesting module (7), the trigger information and the character portrait content are sent to the dialing strategy module (3). The dialing strategy module (3) selects a time period for secondary calling according to the character portrait content; The fuse module (8) is used to be triggered when the user transmits negative information or low-intention information, or when the manual customer service transmits negative information, to announce the ending words and forcibly hang up. The emotion monitoring module (10) is used to monitor the emotion of the manual customer service and trigger the fuse module (8). The manual customer service intervention module (9) is used to switch to the manual customer service when the speech analysis module captures the intention words.

Images