Electricity sales personnel partner training method and system
By refining the themes in real scenes and using large models to conduct natural human-computer dialogues, the problem of poor simulation customers in the existing telemarket training system is solved, and more realistic customer simulation and more efficient training results are achieved.
Patent Information
- Application Number
- CN202411732348.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-29
- Publication Date
- 2025-03-18
AI Technical Summary
There are two major problems when simulating customers: one is that it requires a large amount of reply content to be written manually, resulting in a single repetition of the content and a poor sense of situation in the practice; the other is that it can only use the students' current speech information, and cannot extract the information contained in earlier human-computer dialogues, which makes it difficult to realize complex plot branches and poor ability to simulate customers.
By refining the real scene agent with all the discussion content of the customer, we get the theme, and use the big model to arrange the prompt words, conduct natural human-computer dialogues under the theme, and simulate real customers.
It realizes the simulation of real customers and customer groups, allowing the agent to practice skills in actual combat, reduce training costs, enhance training results, and improve the ability and training effects of simulated customers.
Smart Images
Figure CN119648490A_ABST
Abstract
Description
Background Art
[0002] In the field of telephone sales and telephone customer service, agents are the key factor in the operation and development of enterprises, and it is very important to do a good job in agent training. In the early days, agent training was completely manual, and the training content was consolidated and the training effect was tested through training between agents and coaches, and between agents. With the development of computer technology and artificial intelligence, various forms of training systems have emerged to help agents memorize training materials and help companies evaluate the effectiveness of agent training.
[0003] A core function of the training system is to simulate customers and simulate sales or service processes. Specifically, the "customer" asks questions, the agent answers the questions, and the "customer" asks new questions based on the agent's answer. The existing technical implementation solution is to design courseware, in which customer questions and possible answers corresponding to the questions are set; then continue to set questions for each possible answer until the end. Most of the existing solutions are based on NLP models to determine which preset speech the student's speech is similar to, and write reply content for each speech and design plot branches. There are two major problems with the existing solutions. First, the reply content needs to be written manually, which requires a lot of manpower, and the content is monotonous and repetitive, and the sense of practice is poor; second, it can only use the student's current speech information, and cannot extract the information contained in the earlier human-computer dialogue, so complex plot branches cannot be achieved. These two shortcomings lead to the poor ability of the existing training system to simulate customers, making it difficult for agents to be immersed in the situation and achieve good training results. Summary of the invention
[0004] The present application provides a training method and system for telemarketing personnel, which is used to simulate real customers and customer groups, allowing agents to practice their skills in actual combat, reduce training costs, and enhance training effects.
[0005] In a first aspect, a method for training telemarketers is provided, comprising the following steps:
[0006] Distill all the conversations between agents and customers in real scenarios to get the theme;
[0007] Use the big model to arrange prompt words, conduct natural human-computer dialogue under the topic, and simulate real customers.
[0008] In the above technical solution, themes are obtained by extracting all the contents discussed between agents and customers in real scenarios; prompt words are arranged using a large model, and natural human-computer dialogue is conducted under the theme to simulate real customers; real customers and customer groups are simulated to allow agents to practice their skills in actual combat, reduce training costs, and enhance training effects.
[0009] In a specific implementation plan, the steps of extracting all the content discussed between the real scene agent and the customer to obtain the theme include:
[0010] Segment all the discussions between agents and customers in real scenarios to obtain segmented discussion content;
[0011] The discussion content of each segment is refined to obtain the theme.
[0012] In a specific implementation manner, if a segmented discussion content includes multiple topics, then:
[0013] Divide the segmented discussion content into multiple topics; or
[0014] Combine multiple said topics into one big topic.
[0015] In a specific implementation scheme, the output results of the large model include the content of the reply to the student, whether the topic is ended, the student's speech evaluation, and the summary of the conversation content.
[0016] In a specific implementation scheme, it also includes: arranging the topics according to the output results of the large model.
[0017] In a specific implementation scheme, the step of arranging the topics according to the output results of the large model is specifically as follows:
[0018] The topic for the next call is selected based on the output result of the large model, and the summary of the previous call is brought into the topic of the next call to complete the arrangement of the topic.
[0019] In a specific implementation manner, the arrangement of the topics adopts dynamic connection rules.
[0020] In a specific implementation scheme, the dynamic connection rules include:
[0021] Selecting the topic according to the output result of the large model;
[0022] Insert the summary of the conversation content into the selected topic to complete the direct and smooth connection of the topic.
[0023] In a specific implementation manner, the large model is a large language model.
[0024] In a second aspect, a training system for telemarketers is provided, comprising:
[0025] The theme module is used to extract all the content discussed between agents and customers in real scenarios and obtain the theme;
[0026] The dialogue module is used to use the large model to arrange prompt words, conduct natural human-computer dialogue under the topic, and simulate real customers.
[0027] In the above technical solution, themes are obtained by extracting all the contents discussed between agents and customers in real scenarios; prompt words are arranged using a large model, and natural human-computer dialogue is conducted under the theme to simulate real customers; real customers and customer groups are simulated to allow agents to practice their skills in actual combat, reduce training costs, and enhance training effects. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 A flowchart of a method for training telemarketers provided in an embodiment of the present application;
[0029] Figure 2 A structural block diagram of a training system for telemarketers provided in an embodiment of the present application;
[0030] Figure 3 A schematic diagram of the subject summary, arrangement and simulation robot solution provided for the embodiments of the present application. DETAILED DESCRIPTION
[0031] The present application is further described in detail below through the accompanying drawings and embodiments. Through these descriptions, the characteristics and advantages of the present application will become clearer and more specific.
[0032] The word "exemplary" is used exclusively herein to mean "serving as an example, embodiment, or illustration." Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments. Although various aspects of the embodiments are shown in the drawings, the drawings are not necessarily drawn to scale unless otherwise noted.
[0033] In addition, the technical features involved in different embodiments of the present application described below can be combined with each other as long as they do not conflict with each other.
[0034] To facilitate understanding of the telemarketing personnel training method and system provided in the embodiment of the present application, its application scenario is first explained. The telemarketing personnel training method and system provided in the embodiment of the present application are used to simulate real customers and customer groups, allowing agents to practice skills in actual combat, reduce training costs, and enhance training effects. A core function of the training system is to simulate customers and simulate sales or service processes, specifically, "customers" ask questions, agents answer questions, and "customers" ask new questions based on the agent answers. The existing technical implementation scheme is to design courseware, in which customer questions and possible answers corresponding to the questions are set; then continue to set questions for each possible answer until the end. Most of the existing schemes are based on NLP models to determine which preset speech the student's speech is similar to, and write reply content for each speech and design plot branches. There are two major problems with the existing schemes. First, the reply content needs to be manually written, which requires a lot of manpower, and the content is single and repetitive, and the sense of practice is poor; second, only the current speech information of the student can be used, and the information contained in the earlier human-computer dialogue cannot be extracted, so complex plot branches cannot be realized. These two shortcomings result in the poor ability of existing training systems to simulate customers, making it difficult for agents to be immersed in the situation and achieve good training results. To this end, the present application embodiment provides a method and system for telemarketing personnel to practice skills in real combat, so as to simulate real customers and customer groups, and reduce training costs and enhance training effects. The following is a detailed description of the embodiments in conjunction with specific drawings.
[0035] refer to Figures 1 to 3 , Figure 1 A flowchart of a method for training telemarketers provided in an embodiment of the present application; Figure 2 A structural block diagram of a training system for telemarketers provided in an embodiment of the present application; Figure 3 A schematic diagram of the subject summary, arrangement and simulation robot solution provided for the embodiments of the present application.
[0036] exist Figure 1 In the embodiment of the present application, a method for training telemarketers is provided, comprising the following steps:
[0037] Distill all the conversations between agents and customers in real scenarios to get the theme;
[0038] Use the big model to arrange prompt words, conduct natural human-computer dialogue under the topic, and simulate real customers.
[0039] In the above technical solution, themes are obtained by extracting all the contents discussed between agents and customers in real scenarios; prompt words are arranged using a large model, and natural human-computer dialogue is conducted under the theme to simulate real customers; real customers and customer groups are simulated to allow agents to practice their skills in actual combat, reduce training costs, and enhance training effects.
[0040] In a specific implementation plan, the steps of extracting all the content discussed between the real scene agent and the customer to obtain the theme include:
[0041] Segment all the discussions between agents and customers in real scenarios to obtain segmented discussion content;
[0042] The discussion content of each segment is refined to obtain the theme.
[0043] In a specific implementation manner, if a segmented discussion content includes multiple topics, then:
[0044] Divide the segmented discussion content into multiple topics; or
[0045] Combine multiple said topics into one big topic.
[0046] Specifically, the telemarketer training method shows unique flexibility in dealing with segmented discussion content, especially when dealing with segments containing multiple topics. The characteristics of this method can be elaborated as follows:
[0047] When telemarketers encounter a situation where the segmented discussion includes multiple topics during the training process, they can adopt the following two strategies to handle it flexibly:
[0048] 1. Divide the discussion into multiple topics:
[0049] Purpose: By breaking down the topics, telemarketers can focus more on the core content of each topic, thereby conducting more targeted exercises and discussions.
[0050] Implementation method: According to the logical structure of the content, keywords or the focus of discussion, the segmented content is divided into several independent theme paragraphs. Each theme paragraph revolves around a central topic, which is convenient for telemarketers to deeply understand and remember.
[0051] 2. Merge multiple topics into one big topic:
[0052] Purpose: In some cases, there may be an inherent connection or common goal between multiple topics. By combining these topics, telemarketers can grasp the core of the discussion from a broader perspective and improve their overall understanding and response capabilities.
[0053] Implementation method: Analyze the correlation between multiple topics, such as whether they belong to the same category of problems, whether they serve the same sales goal, etc. Then, integrate these topics into a more general theme, and conduct sparring and discussion around this theme.
[0054] Application scenarios:
[0055] Training for novice telesales personnel: For newly employed telesales personnel, training on segmented topics can help them gradually master sales skills and product knowledge and build up their self-confidence.
[0056] Improvement of advanced telemarketers: For telemarketers who already have a certain amount of experience, sparring with them through combined themes can challenge their thinking and adaptability, and further improve their sales level and professional quality.
[0057] It should be noted that:
[0058] 1. Choose strategies flexibly: During the actual coaching process, you should flexibly choose strategies to divide the segmented discussion content into multiple topics or merge them into one large topic based on the actual situation of the telemarketers and the complexity of the discussion content.
[0059] 2. Maintain content coherence: No matter which strategy you choose, ensure that the discussion content is coherent and logical to avoid confusion or misunderstandings among telemarketers during the exercise.
[0060] 3. Focus on practical simulation: During the training process, real sales scenarios should be simulated as much as possible to allow telemarketers to learn and grow in practice.
[0061] In the above technical solution, when processing segmented discussion content containing multiple topics, the sales skills and professional qualities of telemarketers can be effectively improved by flexibly choosing a strategy of dividing the content into multiple topics or merging them into one large topic.
[0062] In a specific implementation scheme, the output results of the large model include the content of the reply to the student, whether the topic is ended, the student's speech evaluation, and the summary of the conversation content.
[0063] In a specific implementation scheme, it also includes: arranging the topics according to the output results of the large model.
[0064] In a specific implementation scheme, the step of arranging the topics according to the output results of the large model is specifically as follows:
[0065] The topic for the next call is selected based on the output result of the large model, and the summary of the previous call is brought into the topic of the next call to complete the arrangement of the topic.
[0066] In a specific implementation manner, the arrangement of the topics adopts dynamic connection rules.
[0067] In a specific implementation scheme, the dynamic connection rules include:
[0068] Selecting the topic according to the output result of the large model;
[0069] Insert the summary of the conversation content into the selected topic to complete the direct and smooth connection of the topic.
[0070] Here, the summary can provide additional information, such as the car price; the topic has three types, the beginning topic, the middle topic and the ending topic; the middle topic and the ending topic need to add pre-information, such as the quotation topic needs to know the vehicle information.
[0071] Specifically, dynamic connection rules are an efficient and flexible training strategy that combines the output results of the artificial intelligence model with the actual needs of telemarketers, achieving a smooth transition of conversation content and precise connection of topics. Specifically, they include:
[0072] I. Core Elements of Dynamic Connection Rules
[0073] 1. Output results of the large model:
[0074] The large artificial intelligence model integrated in the telemarketing training system can simulate real customer conversation scenarios and output corresponding conversation content or suggestions.
[0075] These output results serve as the basis for selecting topics and inserting conversation content summaries, ensuring the targetedness and effectiveness of the training process.
[0076] 2. Choice of topic:
[0077] Based on the output results of the big model, the telemarketing training system can intelligently identify the stage of the current conversation and select appropriate topics for connection.
[0078] Topic types include opening topics, middle topics, and ending topics, and each type corresponds to different conversation content and goals.
[0079] 3. Insertion of dialogue content summary:
[0080] Under the selected topic, the system inserts the conversation content summary into the corresponding position to achieve a smooth transition of the conversation.
[0081] These summaries can include additional information, such as vehicle price, product features, etc., which helps telemarketers better understand customer needs and respond accordingly.
[0082] 2. How to connect the abstract with the theme
[0083] 1. The abstract provides additional information:
[0084] The summary contains key information from the conversation, such as the price of the car, which is crucial for telemarketers.
[0085] By inserting the summary into the subject line, telemarketers can have a clearer understanding of customer needs and formulate sales strategies accordingly.
[0086] 2. Three types of topics and their connection:
[0087] 2.1 Opening topic: Usually used to introduce a conversation and establish initial contact with the customer. In the opening topic, the telemarketer can introduce himself, the company or the product and try to arouse the customer's interest.
[0088] 2.2 Intermediate topic: It is the core part of the conversation, involving product introduction, price discussion, customer demand analysis, etc. In the intermediate topic, telemarketers need to flexibly adjust the content and strategy of the conversation according to customer feedback and needs.
[0089] For example, when a customer asks about the price of a vehicle, the system can recognize that this is a quotation topic and automatically insert a summary containing vehicle information and price. In this way, telemarketers can provide customers with accurate quotations and explanations based on this information.
[0090] 2.3 Closing topic: used to summarize the conversation and confirm customer needs and order information. In the closing topic, the telemarketer needs to ensure that the customer has a clear understanding of the products and services and guide the customer to complete the purchase decision.
[0091] Adding pre-information:
[0092] For middle and ending topics, the system will add preceding information as needed to ensure the coherence and accuracy of the dialogue.
[0093] For example, in a quotation topic, the system needs to know the vehicle information to generate an accurate quotation summary. Therefore, before inserting the quotation summary, the system will first ensure that the telesales personnel have obtained the relevant vehicle information.
[0094] 3. Advantages of Dynamic Connection Rules
[0095] Improve training efficiency: By intelligently identifying the dialogue stage and selecting the topic, the system can automatically adjust the training content and difficulty, thereby improving the training efficiency.
[0096] Enhance the capabilities of telesales personnel: By simulating real conversation scenarios and providing instant feedback, the system can help telesales personnel better master sales skills and product knowledge and improve their ability to deal with customers.
[0097] Optimize customer experience: By ensuring the consistency and accuracy of the conversation, the system can enhance customers' trust and satisfaction with telemarketers, thereby increasing sales opportunities and performance.
[0098] In summary, the dynamic connection rule in the telemarketer training method is an efficient and flexible training strategy that combines the output results of the artificial intelligence large model with the actual needs of telemarketers, achieving a smooth transition of the conversation content and accurate connection of the topic. The application of this rule will help improve the sales ability and customer satisfaction of telemarketers, thereby promoting the improvement of sales performance.
[0099] In a specific implementation manner, the large model is a large language model.
[0100] It should be noted that Large Language Model (LLM) is one of the hottest topics in the field of artificial intelligence in recent years. It represents the latest progress in natural language processing (NLP) and demonstrates unprecedented ability to understand and generate human language. The following is an explanation of the Large Language Model:
[0101] 1. Definition and Principle
[0102] A big language model is a deep learning model trained with a large amount of text data that can understand and generate human language. It is based on the Transformer architecture and can learn complex patterns of language and complete various language tasks. Its core goal is to accurately learn and understand human language, enabling machines to interpret language like humans, completely changing the way computers understand and generate human language.
[0103] 2. Features and advantages
[0104] Large-scale data training: LLM is trained using massive amounts of text data, typically millions or even billions of words, thereby gaining a deep understanding of language.
[0105] Strong language understanding capabilities: Ability to understand language nuances, semantic relationships, and contextual information, which is the basis for generating coherent and contextually relevant text.
[0106] Flexible application scenarios: It can be applied to a variety of language tasks, such as text generation, sentiment analysis, question-answering systems, automatic summarization, machine translation, document classification, etc., showing strong versatility.
[0107] Emergent capabilities: As the model size increases, LLM demonstrates some amazing capabilities, especially showing potential huge advantages in solving complex tasks.
[0108] Specifically, telephone agents sell products and serve customers by talking to them. The essence of agent skills is the ability to talk, and the essence of simulating customers is also the ability to talk. The training method for telephone sales personnel includes:
[0109] Step 1: Collect the conversations between agents and customers and divide them into segments according to the specific content discussed.
[0110] Step 2: Summarize the discussion content of each segment and summarize it into themes; label each segment with a theme according to the discussion content
[0111] Step 3: If a segment discusses multiple topics, the first option is to continue to divide the segment into multiple segments until each segment has only a unique topic label; the second option is to merge multiple topics into one topic.
[0112] Taking telephone auto insurance as an example, the content of a call between the agent and the customer may be opening remarks, quotation, price objection, and issuance of a policy, while the content of another call may be opening remarks, quotation, insurance type objection, and issuance of a policy. For plan one, the third call may be opening remarks, quotation, [insurance type objection + price objection], and issuance of a policy. According to the above steps 1, 2, and 3, we first summarize the five themes of opening remarks, quotation, insurance type objection, price objection, and issuance of a policy. Then the segment [insurance type objection + price objection] will be divided into two, namely insurance type objection and price objection. For plan two, the third call may be opening remarks, quotation, [objection], and issuance of a policy. According to the above steps 1, 2, and 3, we first summarize the five themes of opening remarks, quotation, insurance type objection, price objection, and issuance of a policy. Then the themes of insurance type objection and price objection will be merged into one theme objection.
[0113] Step 4. Select a suitable big model, write prompt words, and let the big model act out a customer dialogue within the scope of the topic; the output results of the big model include 1) the reply to the student 2) whether the topic is over 3) the student's speech evaluation 4) the summary of the conversation content.
[0114] Step 5: Determine the real customer information of the robot customer as needed, including but not limited to 1) personnel information, age, gender, personality, etc.; 2) vehicle information, license plate, car price, etc.; 3) customer group information (customer portrait), service sensitivity, price sensitivity, etc. Select the possible topics to be discussed according to the real customer information, and arrange the topic connection relationship according to the real customer behavior.
[0115] Step 6. For multiple calls in step 5, arrange the topic of each call according to the characteristics of each call; write prompt words for the call, use the big model to analyze the call and output the call summary and student speech evaluation; the summary and evaluation are used to select the topic that the robot will discuss in the next call.
[0116] Step 7: For the arrangement of topics described in step 5, there is no fixed order of topics, but dynamic connection rules. For example, in response to "price objections", the agent may say that additional roadside assistance value-added services are provided, or that glass insurance services are provided for free; after the price objection topic ends, the conversation content summary will include the agent's handling of the objection information. For the roadside assistance increase, the subsequent connection is to topics related to roadside assistance, such as the robot saying that it does not need roadside assistance; for the free glass insurance, the subsequent connection is to topics related to glass insurance, such as the robot asking what brand of glass it is.
[0117] refer to Figure 3 The left side shows a telephone auto insurance scenario, and what the agent and the customer discuss. According to the behavior of a specific real customer, the robot customer discusses topics, and these topics are strung together into story lines. Different robot customers have different story lines, and the story line of the same robot customer is related to the student's performance, and may issue an order or may reject it.
[0118] In the above technical solution, the robot customer's degree of anthropomorphism far exceeds that of the existing training system, which is reflected in its strong ability to understand trainees, rich and natural responses to trainees, strong interactive ability, and the ability to decide whether to place an order or reject it based on the trainee's performance. It can simulate real customers and customer groups, allowing agents to practice their skills in actual combat, reducing training costs and enhancing training effectiveness.
[0119] exist Figure 2 In the embodiment of the present application, a training system for telemarketers is provided, including:
[0120] The theme module is used to extract all the content discussed between agents and customers in real scenarios and obtain the theme;
[0121] The dialogue module is used to use the large model to arrange prompt words, conduct natural human-computer dialogue under the topic, and simulate real customers.
[0122] In the above technical solution, themes are obtained by extracting all the contents discussed between agents and customers in real scenarios; prompt words are arranged using a large model, and natural human-computer dialogue is conducted under the theme to simulate real customers; real customers and customer groups are simulated to allow agents to practice their skills in actual combat, reduce training costs, and enhance training effects.
[0123] Those skilled in the art will appreciate that the present application may be implemented as a system, method or computer program product.
[0124] Therefore, the present disclosure may be specifically implemented in the following forms, namely: it may be completely hardware, it may be completely software (including firmware, resident software, microcode, etc.), or it may be a combination of hardware and software, generally referred to herein as a "circuit", "module" or "system". In addition, in some embodiments, the present application may also be implemented in the form of a computer program product in one or more computer-readable media, and the computer-readable medium may contain computer-readable program code.
[0125] Any combination of one or more computer-readable media can be used. Computer-readable media can be computer-readable signal media or computer-readable storage media. Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: electrical connections with one or more wires, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this document, computer-readable storage media can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.
[0126] Although the embodiments of the present application have been shown and described above, it is understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art can change, modify, replace and modify the above embodiments within the scope of the present application. On this basis, a variety of replacements and improvements can be made to the present application, all of which fall within the scope of protection of the present application.
Claims
1. A training method for telemarketers, characterized in that: The following steps are involved: Distill all the conversations between agents and customers in real scenarios to get the theme; Use the big model to arrange prompt words, conduct natural human-computer dialogue under the topic, and simulate real customers.
2. The method for training telemarketers according to claim 1, characterized in that: The steps to extract all the content discussed between agents and customers in real scenarios and obtain the theme include: Segment all the discussions between agents and customers in real scenarios to obtain segmented discussion content; The discussion content of each segment is refined to obtain the theme.
3. The method for training telemarketers according to claim 2, characterized in that: If a segment discusses multiple topics, then: Divide the segmented discussion content into multiple topics; or Combine multiple said topics into one large topic.
4. The method for training telemarketers according to claim 3, characterized in that: The output results of the large model include the reply to the student, whether the topic is ended, the student's speech evaluation, and the summary of the conversation content.
5. The method for training telemarketers according to claim 4, characterized in that: Also includes: The topics are arranged according to the output results of the large model.
6. The method for training telemarketers according to claim 5, characterized in that: The steps of arranging the topics according to the output results of the large model are specifically as follows: The topic of the next call is selected according to the output result of the large model, and the summary of the previous call is brought into the topic of the next call to complete the arrangement of the topic.
7. The method for training telemarketers according to claim 6, characterized in that: The arrangement of the topics adopts dynamic cohesion rules.
8. The method for training telemarketers according to claim 7, characterized in that: The dynamic connection rules include: Selecting the topic according to the output result of the large model; Insert the summary of the conversation content into the selected topic to complete the direct and smooth connection of the topic.
9. The method for training telemarketers according to claim 8, characterized in that: The large model is a large language model.
10. A training system for telemarketers, characterized in that: include: The theme module is used to extract all the content discussed between agents and customers in real scenarios and obtain the theme; The dialogue module is used to use the large model to arrange prompt words, conduct natural human-computer dialogue under the topic, and simulate real customers.