An AR-based bidirectional service system, device, storage medium and program product

Abstract

The application discloses an AR-based bidirectional service system, equipment, a storage medium and a program product. The system comprises a data acquisition module, an AR display device, a touch interaction device and an intelligent sensing device are used to acquire browsing behavior data, touch operation data and spatial position data of a customer and generate behavior data archives in association; a push decision module is used to calculate an intention strength score and classify according to the customer's stay duration, interaction depth and repeated viewing times on a specific product, and generate a differentiated push strategy; and a bidirectional collaborative push module is used to push a marketing clue to a customer manager mobile terminal when the behavior data meets a preset triggering condition in a network point, push continued content to a customer mobile device according to a departure behavior type after the customer leaves, and feed back a customer response behavior to the customer manager, so that the spatiotemporal continuity and marketing accuracy of the service are improved.

Description

Technical Field

[0001] This application relates to the field of e-commerce technology, and in particular to an AR-based two-way service system, device, storage medium, and program product. Background Technology

[0002] Traditional financial branches primarily showcase financial products through paper brochures, static displays, and physical display cases, allowing customers to passively receive product information. This one-way information transmission model makes it difficult for customers to deeply understand the characteristics and value of complex financial products, and branch staff lack effective means to identify customer needs. To improve branch service efficiency, some financial institutions have begun to deploy smart interactive devices in their branches, enhancing the customer experience through digital means.

[0003] In related technologies, some financial branches have introduced intelligent display devices such as electronic displays and touch-screen query terminals. Customers can use the touchscreen to query basic product information, and some devices are also equipped with product parameter comparison functions. This type of solution represents an upgrade from paper materials to electronic displays, improving the convenience of information presentation to some extent. However, in this solution, the interaction between customers and the devices is still limited to simple click-to-search, lacking a deep sense of participation and failing to stimulate customer interest in the products. At the same time, the customer browsing data collected by the devices has not been effectively linked with the branch's business system, making it difficult for account managers to promptly understand customers' true needs and preferences, resulting in inaccurate timing of service intervention.

[0004] The aforementioned technological limitations make it difficult for branches to effectively capture and respond to customer needs in a timely manner. Customer managers suffer from low service efficiency due to a lack of behavioral data support, which affects the overall operational efficiency of the branches. Summary of the Invention

[0005] This application provides an AR-based two-way service system, device, storage medium, and program product to improve the operational efficiency of offline outlets.

[0006] Firstly, this application provides a system comprising: an AR product display module, used to display product information to customers in an augmented reality manner through AR display devices deployed in financial outlets, wherein the AR display devices support customers to interact with the three-dimensional virtual model through touch or gestures; a data acquisition module, used to collect customer behavior data within the financial outlets through interactive devices set up in the financial outlets; a push decision module, used to generate marketing leads based on the behavior data, and determine the push target, push content, and push timing according to the customer's real-time status; and a two-way collaborative push module, including a real-time push unit within the outlet and a post-departure follow-up push unit; the real-time push unit within the outlet is used to: push marketing leads to the customer manager's mobile terminal in real time when the customer is within the financial outlet and the behavior data meets preset trigger conditions; the post-departure follow-up push unit is used to: push follow-up content related to the customer's unfinished experience within the outlet to the customer's mobile device after the customer leaves the financial outlet, based on the customer's departure behavior type and the customer manager's service records; and the two-way collaborative push module is used to: feed back the customer's response behavior to the post-departure push content to the customer manager's mobile terminal.

[0007] In the above embodiments, the system improves upon the limitations of customer behavior data being disconnected from business systems and services being confined to the physical space of the branch by constructing a two-way collaborative push mechanism that covers both inside and outside the branch and integrates with AR technology. When a customer is inside the branch, the system generates marketing leads based on behavioral data collected in real time through multi-dimensional AR interactive devices and pushes them to the customer manager's mobile terminal, providing the customer manager with data support supporting the customer's true behavioral tendencies. After the customer leaves the branch, the system generates follow-up push content based on the unfinished experience and the customer manager's service records, extending the service touchpoint from the branch space to the time dimension after the customer leaves. At the same time, the system feeds back the customer's response to the departure push to the customer manager, forming a complete service cycle of "push within the branch - push after departure - customer response - customer manager follow-up". This two-way information flow mechanism establishes a continuous interactive channel between the customer manager and the customer across the branch, improving the spatiotemporal continuity of services and the timeliness of the customer manager's response to changes in customer needs.

[0008] In conjunction with some embodiments of the first aspect, in some embodiments, the push decision module includes an intention judgment unit and a push strategy generation unit; the intention judgment unit is used to: calculate a customer intention strength score based on the customer's dwell time, interaction depth, and number of repeated views on a specific product type in behavioral data, and classify customers into high-intent customers, hesitant customers, and comparison-oriented customers based on the score; the push strategy generation unit is used to: generate differentiated push content for different types of customers, for high-intent customers, the content pushed to the account manager includes suggestions for immediate intervention and key recommended solutions, for hesitant customers, the content pushed to the account manager includes suggestions for continued monitoring, and for comparison-oriented customers, the content pushed to the account manager includes a product differentiation comparison explanation.

[0009] In the above embodiments, the system calculates an intent strength score by integrating three behavioral dimensions: customer dwell time on a specific product, interaction depth, and number of repeated views. Based on the score, customers are categorized into high-intent customers, hesitant customers, and comparison-oriented customers. This process of transforming behavioral data into intent metrics aggregates scattered customer behavioral signals into a structured score representing purchasing tendencies, matching differentiated push content to each customer type. This processing chain, which transforms behavioral data into intent judgments and then into strategy guidance, provides account managers with a clearer direction for service intervention and decision-making basis, improving the targeting of service interventions and the accuracy of push strategies.

[0010] In conjunction with some embodiments of the first aspect, in some embodiments, the preset triggering conditions include: the customer's dwell time on the interactive interface of any product type exceeds a preset duration threshold; the customer's repeated viewing number for any information dimension exceeds a preset number threshold, the information dimensions including price information, risk rating, and rate of return; the customer's interaction depth reaches a preset level, the interaction depth being determined by the number of information levels the customer clicks to enter; the real-time push unit within the branch is also used to: simultaneously display the customer's real-time location information, the list of products viewed, and the dwell time for each product when pushing to the customer manager's mobile terminal.

[0011] In the above embodiments, the system establishes automatic identification rules for key customer behavior nodes by setting quantifiable trigger conditions. When a customer's behavioral data meets any trigger condition, the system pushes marketing leads to the account manager's mobile terminal and simultaneously displays the customer's real-time location information, the list of products viewed, and the duration of time spent on each product. This mechanism transforms service intervention timing into an automatic triggering process based on the actual intensity of customer behavior. The contextual information provided allows account managers to quickly understand the customer's focus and browsing history before intervention, improving the accuracy of service intervention timing and the efficiency of account managers' understanding of customer needs.

[0012] In conjunction with some embodiments of the first aspect, in some embodiments, the data acquisition module includes an interactive device deployment unit and a behavior data acquisition unit; the interactive device deployment unit is used to: deploy various interactive devices in financial outlets, including AR display devices, touch interactive devices, and intelligent sensing devices; the AR display devices include AR glass curtain walls for displaying AR product information at the outlet entrance or display area; the touch interactive devices are used to receive touch operation commands from customers; the intelligent sensing devices include facial recognition sensors and spatial positioning sensors; the behavior data acquisition unit is used to collect customer behavior data through the interactive devices, including: collecting customer browsing behavior data on the AR interface through the AR display devices, the browsing behavior data including products viewed by the customer. The system collects customer touch operation data through touch interaction devices, including click location coordinates, swipe trajectory, gesture type, number of touch points for multi-touch, and operation response time. It also collects customer spatial location data through intelligent sensing devices, including real-time location coordinates, movement path, dwell time in different product display areas, and number of times the customer approaches a product display area. The behavior data collection unit further associates browsing behavior data, touch operation data, and spatial location data to form behavior data, links this behavior data with customer identity information to generate a customer behavior data profile, and transmits it to the push decision module.

[0013] In the above embodiments, the system constructs a multimodal data acquisition system by deploying AR display devices, touch interaction devices, and intelligent sensing devices. It acquires customer browsing behavior data from the AR interface, refined operational data from the touch devices, and spatial location data from the sensing devices, improving upon the limitation of data acquisition to simple click-based queries. The system correlates these three types of data from different sources and with varying granularities, generating behavioral data profiles bound to customer identities and transmitting them to the push decision module. This achieves the transformation from scattered customer behavior at the branch to structured data assets. This multi-source data acquisition and correlation processing mechanism enables the system to comprehensively depict customer behavioral characteristics at the branch from browsing, operational, and spatial dimensions, providing rich and structured data input for subsequent intention judgment and push decision-making, thus improving the richness and quality of customer behavior data collection.

[0014] In conjunction with some embodiments of the first aspect, in some embodiments, the extended content includes: supplementary content for customers who have not completed their experience, including detailed comparative analysis of products that customers viewed at the outlet but did not understand in depth, and customer case studies; extended content of in-store services, including product solutions recommended to customers by account managers at the outlet, personalized messages from account managers, and contact information of account managers; and personalized content generated based on customer concerns, generating targeted product interpretations based on the information dimensions that customers focus on at the outlet.

[0015] In the above embodiments, the system establishes a service extension channel from within the store to after the customer leaves by constructing three types of continuation content: supplementary content for customers' incomplete experiences, extended content for in-store services, and personalized content based on customer concerns. This content generation strategy makes post-departure push notifications a natural continuation of the in-store experience rather than an independent marketing activity, avoiding the problem of traditional store services ending after the customer leaves and the inability to continue customer interests. By transforming in-store behavioral data into personalized post-departure content, the system improves the continuity of service and the matching degree between push content and customers' actual needs.

[0016] In conjunction with some embodiments of the first aspect, in some embodiments, the two-way collaborative push module achieves a service closed loop through the following mechanism: monitoring the customer's response behavior to the content pushed after leaving the store, including clicking on the pushed content, browsing the pushed link, and filling in the feedback form; when the customer's response behavior is detected, generating feedback information and pushing it to the corresponding account manager's mobile terminal; displaying follow-up suggestions on the account manager's mobile terminal, including the timing of suggested telephone follow-ups and suggested product features to be emphasized.

[0017] In the above embodiments, the system establishes a reverse information flow from the client to the account manager by monitoring the customer's response to the content pushed after leaving the store. This connects the originally scattered three links of in-store push, post-departure push, and customer response into a complete information loop, enabling account managers to adjust subsequent service strategies based on the actual response of customers after leaving the store. This improves the problem that account managers cannot know the push effect in the traditional one-way push mode, and enhances the completeness of the service loop and the account manager's follow-up decision support capabilities.

[0018] In conjunction with some embodiments of the first aspect, some embodiments further include: a departure behavior analysis module, used to: classify customer departure behavior into four types based on the customer's dwell time at the branch, interaction depth, and whether they communicated with the account manager: proactive departure, in-depth experience, service interruption, and decision-making hesitation; and a post-departure follow-up push unit executes differentiated push strategies based on the departure behavior type: for proactive departure customers, push branch experience review content after a first preset time after departure to guide customers to relearn about the products; for in-depth experience customers, push product comparison analysis that the customer is interested in after a second preset time after departure to provide decision support, the second preset time being longer than the first preset time; for service interruption customers, push content containing account manager service follow-up information after a third preset time after departure to prompt them to schedule a follow-up communication, the third preset time being longer than the second preset time; and for decision-making hesitation customers, push notifications are sent at multiple preset time points after departure, with the content of each push adjusted according to the customer's response.

[0019] In the above embodiments, the system analyzes three dimensions—customer dwell time at the branch, interaction depth, and whether they communicate with the account manager—to categorize departure behavior into four types: proactive departure, in-depth experience, service interruption, and decision-making hesitation. Differentiated push notification timing and content are designed for each type. This categorized push strategy transforms departure behavior from a simple "customer leaving" event into a signal reflecting the customer's needs, ensuring that push notification timing and content match the customer's actual needs. This improves the effectiveness of push notifications that are poorly implemented when a uniform push strategy treats customers with different departure reasons the same way. By establishing a correlation mechanism between departure behavior and push strategy, the accuracy of post-departure push notification timing and content relevance are enhanced.

[0020] In a second aspect, embodiments of this application provide a bidirectional service device, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors call the computer instructions to cause the bidirectional service device to implement the system as described in the first aspect and any possible implementation of the first aspect.

[0021] Thirdly, embodiments of this application provide a computer program product containing instructions that, when the computer program product is run on a bidirectional service device, cause the bidirectional service device to implement the system as described in the first aspect and any possible implementation of the first aspect.

[0022] Fourthly, embodiments of this application provide a computer-readable storage medium including instructions that, when executed on a bidirectional service device, cause the bidirectional service device to implement the system described in the first aspect and any possible implementation thereof.

[0023] Understandably, the bidirectional service device provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to implement the system provided in the embodiments of this application when executed. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding system, and will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0025] 1. This application constructs a two-way collaborative push mechanism covering both inside and outside the outlet. When the customer is in the outlet, marketing leads generated based on behavioral data are pushed to the customer manager's mobile terminal. After the customer leaves the store, follow-up content is generated based on the unfinished experience and pushed to the customer's mobile device. The customer's response to the post-departure push is fed back to the customer manager to form a complete service cycle, which improves the time and space continuity of the service and the timeliness of the customer manager's response to changes in customer needs.

[0026] 2. This application constructs a multimodal data collection system by deploying AR display devices, touch interaction devices, and intelligent sensing devices. It collects customer behavior data from browsing, operation, and spatial dimensions, and performs correlation processing to generate behavior data archives. This improves the richness and quality of customer behavior data collection dimensions, and provides structured data support for push decision-making.

[0027] 3. This application calculates the intensity of intent score by integrating customer dwell time, interaction depth, and number of repeated views on specific products, and classifies customers accordingly. It generates differentiated push strategies and service intervention suggestions for different types of customers, thereby improving the targeting of service intervention and the accuracy of account managers' judgment of customer needs. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of a module of the AR-based bidirectional service system in an embodiment of this application;

[0029] Figure 2 This is a schematic diagram of the physical device structure of a two-way service device in the embodiments of this application. Detailed Implementation

[0030] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification of this application, the singular expressions “a,” “an,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to any or all possible combinations including one or more of the listed items.

[0031] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0032] The system provided in this implementation is described below. Please refer to [link / reference]. Figure 1 This is a schematic diagram of a module of an AR-based bidirectional service system in an embodiment of this application. The system includes:

[0033] The AR product display module is used to display product information to customers in an augmented reality manner through AR display devices deployed in financial outlets. The AR display devices support customers to interact with the three-dimensional virtual model through touch or gestures.

[0034] AR display devices refer to display devices that can overlay virtual digital content onto real-world scenes, including AR glass curtain walls, 3D holographic display cases, etc.

[0035] When a customer enters a financial branch and approaches the AR display device, the AR product display module automatically initiates the display process. Specifically, the system identifies the customer using a facial recognition sensor and retrieves information such as the customer's risk preferences and asset status; then, based on the customer profile, it selects suitable product types and calls the AR content generation unit to generate AR display content for the selected products; finally, it presents the content to the customer through the AR display device.

[0036] Preferably, the AR product display module includes an AR content generation unit and an AR interaction unit;

[0037] The AR content generation unit is used for: for physical products, generating a 3D virtual model containing process details and material textures based on the product's 3D scanning data, and overlaying and displaying the product's real-time market price and historical price trends on the 3D virtual model; for virtual products, generating a dynamic return curve based on the product's rate of return data and risk coefficient, and rendering the expected return display under different market scenarios in real time according to the investment amount and investment period parameters input by the customer.

[0038] The AR interaction unit is used to: update the display status of the virtual model in real time based on the customer's scaling and rotation operations on the 3D virtual model; calculate and update the profit prediction results in real time based on the customer's parameter adjustment operations on the dynamic profit curve; record the customer's interaction operation data and transmit it to the data acquisition module.

[0039] The data acquisition module is used to collect customer behavior data within financial outlets through interactive devices installed in financial outlets;

[0040] Interactive devices refer to intelligent hardware devices deployed within financial outlets that can exchange information with customers, including AR display devices, touch interactive devices, and intelligent sensing devices, which are used to automatically record customer operation behavior when customers interact with the devices; behavioral data refers to behavioral characteristic information that can reflect customers' product attention tendencies and demand intentions within the outlets, including multi-dimensional data such as browsing behavior, touch operation, and changes in spatial location.

[0041] When a customer enters a financial branch and begins interacting with any interactive device, the data acquisition module automatically initiates the data collection process. Specifically, the system captures customer behavior in real time using multiple types of sensors and recording devices: the motion capture camera built into the AR display device continuously records the area where the customer's gaze lingers and the frequency of switching on the AR interface at a frame rate of 30fps; the touch interaction device records the position coordinates and operation duration of each touch with millisecond-level precision through a capacitive sensing layer; and the spatial positioning sensor of the intelligent sensing device updates the customer's real-time coordinates within the branch at a frequency of seconds.

[0042] Preferably, the data acquisition module includes an interactive device deployment unit and a behavioral data acquisition unit;

[0043] The interactive device deployment unit is used to: deploy various interactive devices in financial outlets, including AR display devices, touch interactive devices, and intelligent sensing devices. The AR display devices include AR glass curtain walls, used to display AR product information at the entrance of the outlet or in the display area; the touch interactive devices are used to receive touch operation commands from customers; and the intelligent sensing devices include facial recognition sensors and spatial positioning sensors.

[0044] Among them, AR glass curtain walls refer to large-size display devices that use a double-layer transparent tempered glass structure and embed micro sensors and projection units, which can display virtual product information while maintaining transparency; touch interaction devices refer to smart terminals equipped with multi-point capacitive touch screens that support gesture recognition and multi-point simultaneous touch; spatial positioning sensors refer to devices that achieve indoor positioning based on Bluetooth beacon or Wi-Fi triangulation technology, with positioning accuracy reaching the meter level.

[0045] Equipment deployment is carried out during the branch's renovation or intelligent transformation phase. Specifically, the AR glass curtain wall is installed at the branch entrance, directly facing the customer's entrance, with a height set between 1.8 and 2.2 meters to cover the natural line of sight of adult customers. Facial recognition sensors are embedded in the grooves of the glass edge and the recognition distance is adjusted to 0.5-3 meters to ensure that customers can be recognized from the moment they enter until they approach the glass. Touch-screen interactive tables are placed in the transition area between the consultation and guidance area and the experience area, with the table height set at 45 centimeters according to ergonomic standards for easy operation by customers for extended periods. Bluetooth beacons for intelligent sensing devices are gridded on the branch's ceiling at a density of one beacon every 10-15 square meters. The core idea of ​​this deployment strategy is to build a three-dimensional data collection network of "entrance recognition - movement tracking - area interaction," enabling the system to perceive the entire process of customer behavior from entering the branch to leaving. The collaborative deployment of multiple devices also achieves data redundancy. When a sensor loses data due to obstruction or malfunction, other devices can provide supplementary data, ensuring the integrity of behavioral data.

[0046] The behavior data collection unit is used to collect customer behavior data through interactive devices, including:

[0047] The data on customer browsing behavior on the AR interface is collected through AR display devices. The browsing behavior data includes the types of products that customers view, the duration of stay at different information levels, and the frequency of switching product information.

[0048] The system collects customer touch operation data through touch interaction devices. The touch operation data includes click location coordinates, swipe trajectory, gesture type, number of touch points in multi-touch, and operation response time.

[0049] Customer spatial location data is collected through intelligent sensing devices. The spatial location data includes the customer's real-time location coordinates, movement path, dwell time in different product display areas, and number of times they approach the product display area within the outlet.

[0050] Among them, information hierarchy refers to the depth of product information presentation in the AR interface, with a progressively deeper information structure from product list (level 1), product details (level 2) to revenue simulation (level 3); swipe trajectory refers to the path of finger movement on the screen surface during touch operation, represented by the sequence of starting point coordinates, ending point coordinates, and intermediate sampling points; movement path refers to the customer's movement trajectory within the network space, formed by connecting position coordinate points arranged in a time sequence.

[0051] When a customer begins interacting with any interactive device, the corresponding device's data acquisition program is automatically activated and continuously records until the interaction ends. Specifically, the AR display device analyzes changes in the customer's pupil position captured by the camera using an eye-tracking algorithm, calculates the customer's gaze duration in different product display areas of the AR interface, and records the information level depth corresponding to the click behavior when the customer clicks to enter a product's detailed information page; the touch interaction device records the real-time coordinates of touch points at a sampling frequency of 100Hz, and when two touch points are detected to appear simultaneously within a specific range, it is recognized as a zoom gesture and the zoom ratio is recorded; the intelligent sensing device updates the customer's location at a frequency of seconds by receiving Wi-Fi probe signals from the customer's mobile phone or Bluetooth beacon responses from mobile phones with the bank's app installed, and accumulates the dwell time when the location coordinates remain within the same product display area for multiple consecutive cycles.

[0052] The behavior data collection unit is also used to: associate browsing behavior data, touch operation data, and spatial location data to form behavior data, associate behavior data with customer identity information, generate customer behavior data profiles, and transmit them to the push decision module.

[0053] Among them, behavioral data profiles refer to structured data records organized by customer dimension, which include customer identity information, complete behavioral data sequences and corresponding timestamps, and are stored in JSON or relational database format.

[0054] After the system completes data collection for a single interaction cycle (e.g., an interaction cycle defined as more than 30 seconds from the start of interaction to the customer leaving the device or ceasing interaction), the data association processing flow is immediately initiated. Specifically, the system first uses the customer's identity ID obtained from the facial recognition sensor as the primary key to extract all device data generated by that customer within the same time period from the data cache; then, it performs time-series alignment based on timestamps, horizontally associating browsing behaviors, touch operations, and location changes occurring within the same second; next, it verifies data consistency by matching spatial location with device location—for example, if the customer's spatial location shows that they are near the touch interaction table, then the touch data for that time period should belong to that customer. After data association is completed, the system retrieves the customer's historical transaction records, risk preferences, and other static information from the CRM system and merges them with the current behavioral data to form a complete behavioral data profile. Finally, the system transmits the behavioral data profile to the push decision module in real time through an encrypted channel, with transmission latency controlled within 1 second, ensuring that subsequent decisions are made based on the latest data.

[0055] The push decision module is used to generate marketing leads based on behavioral data and determine the push recipients, push content, and push timing based on the customer's real-time status.

[0056] Marketing leads refer to structured information extracted from customer behavior data that indicates a customer's potential purchase intention, including the types of products the customer is interested in, the intensity of their intention, and suggested service intervention methods.

[0057] Once the push decision module receives the behavioral data archive transmitted by the data collection module, it immediately initiates the decision calculation process. Specifically, the system first inputs the behavioral data into the intention judgment unit, and generates a customer intention strength score through weighted calculation: the customer's dwell time on a specific product, interaction depth, and number of repeated views are assigned weights (e.g., 0.4, 0.4, 0.2), and after normalization, a comprehensive score is calculated. Customers with a score greater than 0.7 are marked as high-intent customers. Subsequently, the push strategy generation unit assembles push content based on the customer classification results and real-time status: for high-intent customers who are operating a financial product return simulation on the touch table, the system generates a marketing lead containing "The customer is deeply understanding XX financial product, and we suggest immediate intervention and consultation"; for hesitant customers who are lingering in the precious metals display area but have not engaged in in-depth interaction, a lead is generated "The customer has initial interest in precious metals products, and we suggest following up later."

[0058] This real-time-based dynamic decision-making mechanism aims to avoid disturbing customers at inappropriate times, such as not sending push notifications when a customer has just entered the branch and has not yet shown a clear intention. Instead, it waits until behavioral data has accumulated enough to determine the customer's needs before triggering a push. The system also determines the timing of push notifications based on the account manager's work status (determined by the online status of their mobile device). If the account manager is serving other customers, the system will delay the push or forward it to other available account managers to ensure that leads receive a timely response.

[0059] Preferably, the push decision module includes an intention judgment unit and a push strategy generation unit;

[0060] The intent judgment unit is used to: calculate the customer intent strength score based on the customer's dwell time, interaction depth, and number of repeated views on a specific product type in the behavioral data, and classify customers into high-intent customers, hesitant customers, and comparison customers based on the score;

[0061] Among them, specific product types refer to the major categories of products offered by financial outlets, such as precious metals, wealth management products, and credit cards; interaction depth refers to the number of information levels that customers click to enter, with a higher value indicating a deeper exploration of product information; and intention strength score refers to quantifying customer behavior characteristics into a value in the range of 0-1, used to characterize the strength of customer purchase intention.

[0062] After the system acquires complete customer behavior data profiles, it automatically executes an intent judgment algorithm. Specifically, the system first normalizes three behavioral dimensions: mapping dwell time to a 0-1 range (setting 5 minutes as the maximum score), mapping interaction depth to a 0-1 range (setting clicking to enter a third-level page as the maximum score), and mapping repeat viewing count to a 0-1 range (setting 5 times as the maximum score). Then, it calculates a weighted sum using preset weights of 0.4, 0.4, and 0.2 to obtain an intent strength score. Next, it classifies customers according to the score thresholds: customers with a score greater than or equal to 0.7 are classified as high-intent customers, indicating that their behavior clearly points to a purchase intention; customers with a score between 0.4 and 0.7 and a low number of repeat viewings are classified as hesitant customers, indicating that they are in the information gathering stage; and customers with a score between 0.4 and 0.7 but a high number of repeat viewings are classified as comparison customers, indicating that they are comparing and weighing multiple products.

[0063] This classification logic is based on the psychological process of financial product purchase decisions: high-intent customers have already clarified their needs and collected information, requiring account managers to provide final professional advice; hesitant customers are still in the demand exploration stage, and premature intervention may cause resentment; comparison-oriented customers need clear explanations of product differentiation to support their decisions. For example, when a customer spends 4 minutes on a 3-year wealth management product page, clicks to view two secondary pages (return simulation and risk description), and repeatedly returns to the product details page 3 times, the system calculates their intent strength score as 0.75, classifies them as a high-intent customer, and marks the product they are interested in as a "3-year stable wealth management product."

[0064] The push strategy generation unit is used to generate differentiated push content for different types of customers. For high-intent customers, the content pushed to the account manager includes suggestions for immediate intervention and key recommended solutions. For hesitant customers, the content pushed to the account manager includes suggestions for continued monitoring. For comparison-oriented customers, the content pushed to the account manager includes a product differentiation comparison explanation.

[0065] Once the intent assessment unit completes customer classification, the push strategy generation unit executes differentiated content generation logic based on the classification results. Specifically, for high-intent customers, the system extracts the product with the longest customer dwell time from behavioral data as the primary recommended product. Combining this with customer risk preference data from the CRM system, it matches 2-3 products with similar characteristics from the product library as alternatives, assembling them into push content: "Customer [Zhang San] is deeply understanding the [Zodiac Gold Bar] in the precious metals display area, having examined the craftsmanship details for over 3 minutes. We suggest you immediately proceed with your purchase. We can highlight the [Zodiac Gold Bar] and [Investment Gold Bar] sets. The customer's risk preference is conservative, so we can emphasize its value preservation and appreciation functions." For hesitant customers, the system generates more moderate push content: "Customer [Li Si] has spent 1.5 minutes browsing multiple products on the financial product touch table, but has not yet shown a clear preference. We suggest continuing to monitor and intervening when they engage more deeply." For customers who prefer comparison, the system automatically generates a product comparison description: "Customer [Wang Wu] has viewed [XX Wealth Management] and [YY Wealth Management] a total of 5 times. The system has generated a product comparison table (see attachment). We suggest you proactively provide a comparison explanation, highlighting the advantages of [XX Wealth Management] in terms of liquidity and return stability." The core design idea of ​​this differentiated push strategy is to match the customer's decision-making stage with their service needs, avoiding the waste of account manager resources or inappropriate service timing caused by uniform pushes.

[0066] The two-way collaborative push module includes a real-time push unit within the outlet and a continued push unit after the customer leaves the outlet;

[0067] The in-branch real-time push unit is used to push marketing leads to the account manager's mobile terminal in real time when the customer is in the financial branch and the behavioral data meets the preset trigger conditions.

[0068] Among them, the preset trigger conditions refer to the behavioral characteristic thresholds that the system predefines to determine whether to start push notifications, including quantifiable indicators such as dwell time threshold, number of repeated views threshold, and interaction depth level threshold.

[0069] When the system detects that a customer's behavioral data meets any preset trigger condition, it immediately initiates the push notification process. Specifically, the system uses a rule engine to compare customer behavioral data with the trigger conditions in real time: whenever a customer's cumulative dwell time on a product reaches a preset threshold (e.g., 2 minutes), the system marks the event as a trigger point; or when a customer repeatedly views a certain information dimension a certain number of times (e.g., 3 times), the system identifies it as a high level of interest in that information; or when a customer clicks to enter a third-level information page, the system determines that the interaction depth has reached a level indicating deep interest. Once the trigger condition is met, the system immediately queries the binding relationship between the customer and the account manager (obtained through the customer affiliation field in the CRM system) and sends the marketing lead to the corresponding account manager's mobile terminal in the form of a message notification via the push service, with the push delay controlled within 2 seconds.

[0070] This real-time trigger push mechanism aims to capture the critical moments in the formation of customer intentions, delivering information to account managers when customer needs are strongest and professional guidance is most required. For example, when a customer adjusts the investment amount parameters of a financial product three times consecutively on the touch screen, the system determines that the customer has reached the critical point of investment decision-making. At this time, a push notification to the account manager enables them to seize the best service opportunity and prevent the customer from abandoning the purchase due to unresolved doubts.

[0071] Preferably, the preset trigger conditions include: the customer's dwell time on the interactive interface of any product type exceeds a preset time threshold; the customer's repeated viewing number for any information dimension exceeds a preset number threshold, and the information dimensions include price information, risk rating, and rate of return; the customer's interaction depth reaches a preset level, and the interaction depth is determined by the number of information levels that the customer clicks to enter.

[0072] Among them, the preset duration threshold refers to the benchmark for judging the dwell time set according to the information complexity of different product types, such as 2 minutes for precious metal products and 3 minutes for wealth management products; the preset number of times threshold refers to how many times a customer repeatedly views the same information dimension to be considered highly concerned, usually set to 3 times; the preset level refers to the depth position in the information architecture, and pages at level three and above usually contain the core terms of the product, profit simulation and other key decision information.

[0073] Specifically, the system maintains an independent timer for each product type. The timer starts when a customer begins to view a product. If the customer's cumulative time spent on the product page (excluding idle time without any operation) exceeds the corresponding threshold, the system determines that trigger condition 1 is met. At the same time, the system maintains an access counter for each information dimension. When a customer clicks to view price information, returns, and then clicks to view price information again, the counter increments. Once a preset number of times is reached, trigger condition 2 is met. The system also tracks the customer's page navigation path in real time. When it detects that a customer enters the product details (second level) from the product list (level 1) and then enters the revenue simulator (level 3), it determines that the interaction depth has reached level 3, and trigger condition 3 is met.

[0074] The real-time push unit within the branch is also used to simultaneously display the customer's real-time location information, the list of products viewed, and the duration of time spent on each product when pushing to the customer manager's mobile terminal.

[0075] While the system pushes marketing leads to account managers' mobile devices, it automatically attaches contextual information about the customer to aid decision-making. Specifically, the push message uses a structured format to organize information: the top displays basic customer information (name, customer number) and an intent strength score; the middle section shows the customer's real-time location, with the system displaying a branch map and marking the customer's current location (e.g., "Precious Metals Display Area - near Booth 3"), while also marking the account manager's own location and calculating the straight-line distance between them; the bottom lists the products the customer has viewed, with the duration of each view indicated, such as "Zodiac Gold Bar (3 minutes 15 seconds) → Investment Gold Bar (1 minute 30 seconds) → Gold Jewelry (45 seconds)". This information integration design aims to provide account managers with comprehensive decision support, enabling them to quickly understand the customer's focus and behavioral patterns before intervening in service. For example, account managers can use location information to determine whether they can quickly reach the customer's location. If the distance is far, they can ask other nearby account managers for assistance. Through the product list, they can understand the customer's comparison preferences. If the customer has viewed multiple similar products and spent similar amounts of time on each, it indicates that the customer is in a stage of indecision regarding product selection. When the account manager intervenes, they should focus on providing comparative analysis.

[0076] The post-departure follow-up push unit is used to push follow-up content related to the customer's unfinished experience at the branch to the customer's mobile device after the customer leaves the financial outlet, based on the customer's departure behavior type and the account manager's service record.

[0077] Among them, "continuing content" refers to personalized information generated based on customer behavior data and service records within the outlet, which aims to continue the customer's product understanding or decision-making process that was not completed within the outlet.

[0078] When the system detects a customer leaving the branch via intelligent sensing devices (the customer's location signal disappears for more than 5 minutes), the push notification process automatically starts. Specifically, the system first calls the departure behavior analysis module to determine the customer type, and then matches a differentiated push strategy based on the type: For customers with a deep experience (staying at the branch for more than 15 minutes and reaching level three of interaction depth but not completing a transaction), the system pushes content one hour after the customer leaves, including detailed comparative analysis of the products the customer is interested in and real customer cases, such as "The [XX financial product] that you learned about in depth at the branch has a 3-year term, with an average annualized return of 4.2%. Click to view detailed cases." For customers with interrupted service (leaving after service is interrupted during communication with the account manager due to phone calls or other reasons), the system pushes content including a personalized message from the account manager 30 minutes after leaving, such as "Manager Zhang: I apologize for not being able to continue serving you due to temporary matters. Regarding your interest in precious metal investment, I have prepared a customized plan for you (attached). I look forward to your next visit. My contact information is: 138XXXX." The core design idea of ​​this extended push mechanism is to extend the service from the physical space of the store to the time dimension after the customer leaves the store, so as to avoid the customer's purchase intention cooling down due to leaving the store.

[0079] Preferred extended content includes: supplementary content for customers who did not complete their experience, including detailed comparative analysis of products that customers viewed at the store but did not understand in depth, and customer case studies; extended content of in-store services, including product solutions recommended by account managers to customers at the store, personalized messages from account managers, and contact information of account managers; and personalized content generated based on customer concerns, which generates targeted product interpretations based on the information dimensions that customers focus on at the store.

[0080] Among them, detailed comparative analysis refers to creating a comparison table or chart of multiple products that the customer has viewed according to key dimensions (yield, risk level, investment period, etc.) to intuitively show the differences between the products; customer case studies refer to investment cases of real customers with similar risk preferences and investment amounts to the current customer, including information such as investment products, holding periods, and actual returns; personalized messages refer to the response content written by the account manager for the specific questions or concerns raised by the customer in the branch.

[0081] After determining the timing of the push notification, the system automatically generates follow-up content based on the customer's behavior data and service records at the branch. Specifically, for supplementary content for incomplete experiences, the system filters products viewed by the customer that have been viewed for more than 1 minute but whose interaction depth has not reached level three. It then uses detailed parameters from the product database to generate a comparative analysis table. Simultaneously, it matches the 3-5 most similar cases from the customer case library (similarity is calculated based on dimensions such as customer age, asset size, and risk preference) and assembles them into push content. For extended content related to in-store services, the system reads the service records entered by the account manager via mobile terminal during the service process, extracts the recommended product solutions and communication points, and combines them with the personalized signature and contact information pre-set by the account manager to generate extended content including "Dear customer, thank you for visiting our branch. Based on your needs, I recommend...". For personalized content, the system analyzes the information dimensions that customers view most frequently (such as yield) and uses the natural language generation module to write in-depth interpretation articles for that dimension.

[0082] This combination of multiple types of ongoing content comprehensively meets the information needs that customers may have after leaving the store, providing decision support (comparative analysis, case studies), reflecting humanistic care (customer manager messages), and supplementing knowledge gaps (personalized interpretation).

[0083] The two-way collaborative push module is used to: feed back the customer's response to the content pushed after leaving the store to the account manager's mobile terminal.

[0084] Among them, response behavior refers to the interactive actions that customers take after receiving the off-site push notification, which can be monitored by the system. These actions include clicking on the push notification, browsing the push link, the duration of stay on the linked page, filling out the feedback form, and clicking on the customer manager's contact information.

[0085] Preferably, the two-way collaborative push module achieves a service loop through the following mechanism: monitoring the customer's response behavior to the content pushed after leaving the store, including clicking on the content, browsing the link, and filling out the feedback form; when a customer's response behavior is detected, generating feedback information and pushing it to the corresponding account manager's mobile terminal; displaying follow-up suggestions on the account manager's mobile terminal, including the timing of the suggested phone follow-up and the product features to be emphasized.

[0086] The system initiates response behavior monitoring immediately after pushing outgoing content. When customer interaction is detected, a closed-loop processing flow is triggered immediately. Specifically, the system establishes an independent monitoring task for each push notification, acquiring customer interaction data in real time through tracking code in the push link. When a click event is detected, the system records the click time and calculates the interval from the push notification time (e.g., a customer clicks 5 minutes after the push notification, indicating a timely response). When browsing behavior is detected, the system analyzes page dwell time and browsing depth; if the dwell time exceeds 2 minutes and the user scrolls to the bottom of the page, it is considered deep browsing. When a form submission is detected, the system extracts structured fields such as appointment time and inquiry questions from the form. The system then runs a follow-up suggestion generation algorithm: if the customer clicks and deeply browses within 30 minutes of the push notification, the suggestion is generated: "The customer responded positively; we suggest you follow up by phone before 5 PM today, emphasizing the product's liquidity advantage to address customer concerns." If the customer only clicks but does not deeply browse, the suggestion is generated: "The customer's interest is limited; we suggest waiting 1-2 days to push supplementary content again or sending a greeting via SMS." The core of this closed-loop mechanism is to connect discrete push actions into a complete service chain, so that each push can receive feedback on its effectiveness and guide the next step.

[0087] In other embodiments of this application, the system further includes:

[0088] The departure behavior analysis module is used to classify customers' departure behavior into four categories based on their time spent at the branch, depth of interaction, and whether they communicate with the account manager: proactive departure, in-depth experience, service interruption, and decision hesitation.

[0089] When the system detects a customer leaving the store, the departure behavior analysis module automatically initiates a classification process. Specifically, the system extracts complete behavioral data of the customer at the store, including entry time, departure time, number of products visited, interaction depth of each product, whether a customer manager push notification was triggered, and whether the customer manager has a service record. Then, it uses a decision tree algorithm to classify the customer: if the stay time is less than 5 minutes and the interaction depth does not reach level two and there is no customer manager service record, it is judged as an active departure; if the stay time is more than 15 minutes and the interaction depth reaches level three but there is no transaction record for that day in the CRM system, it is judged as a deep experience departure; if the customer manager service record marks "service interruption" and the customer leaves the store within 5 minutes after the marked time, it is judged as a service interruption departure; if the stay time is more than 10 minutes and the number of products visited is more than 3 and the difference in stay time among products is less than 30%, it is judged as a decision-making indecisive departure.

[0090] This classification logic based on multi-dimensional data can accurately identify the true needs of customers when they leave the store. For example, if customer Zhang San stays at the branch for 20 minutes, reviews three financial products in depth, and communicates with the account manager for 10 minutes, but leaves due to an urgent business call, the system will identify Zhang San as a service interruption customer based on the "customer interrupted communication due to business" record marked by the account manager on the mobile terminal.

[0091] The post-departure follow-up push unit executes a differentiated push strategy based on the type of departure behavior:

[0092] For customers who voluntarily leave the store, push out a review of the store experience after the first preset time after leaving the store to guide customers to learn about the products again;

[0093] For customers who are keen on in-depth experiences, a comparative analysis of the products they are interested in will be pushed to them after the second preset time after they leave the store, providing decision support. The second preset time is longer than the first preset time.

[0094] For customers whose service is interrupted, a message containing information on the continuation of service from the account manager will be sent after the third preset time since the customer leaves the store, prompting them to schedule a follow-up communication. The third preset time is longer than the second preset time.

[0095] For customers who are hesitant in making decisions, push notifications are sent at multiple preset time points after they leave the store, and the content of each push is adjusted according to the customer's response.

[0096] The first preset time is usually set 2-3 hours after the customer leaves the store. At this time, the customer has left the store but still remembers the experience. Pushing a review content can arouse the customer's interest. The second preset time is usually set 24 hours after the customer leaves the store, giving customers who want a deep experience enough time to make a decision. The third preset time is usually set 48 hours after the customer leaves the store, to prevent customers who are experiencing service interruption from switching to other channels due to long waiting times. Multiple preset time points refer to the push time sequence set at certain time intervals (such as 4 hours, 24 hours, and 72 hours after the customer leaves the store). Whether to push subsequent pushes depends on the customer's response to the previous push.

[0097] The system automatically matches the corresponding push strategy and sets up scheduled tasks based on the classification results of departure behavior. Specifically, for customers who voluntarily leave the branch, the system will push content 3 hours after they leave, including screenshots of products they viewed at the branch, introductions to the branch's AR experience highlights, and product discount information, with a gentle guiding message such as "Thank you for visiting XX Bank branch. The [XX product] you viewed is currently offering a limited-time discount..." For customers who want to experience the product in depth, the system will push detailed product comparison analysis, real customer cases, and online consultation access 24 hours after they leave, with a professional and supportive message such as "Based on your in-depth experience at the branch, we have prepared a personalized product comparison plan for you..." For customers whose service is interrupted, the system will push a personalized message from the account manager and an appointment link 1 hour after they leave, with a message such as "XX Manager: I'm sorry I couldn't continue to serve you. I have reserved the product plan for you. Please click to schedule an appointment to continue communication..." For customers who are hesitant in their decisions, the system is designed with three progressive pushes: the first (4 hours later) pushes a summary of product highlights; the second (24 hours later) pushes content based on whether the customer clicks on the first push; and the third (72 hours later) pushes limited-time discount information or a prompt for the account manager to contact them. This differentiated push strategy matches the psychological state and information needs of different customers, pushing the right content at the right time.

[0098] In one specific embodiment of this application, the association between behavioral data and customer identity information and the customer's mobile device can be achieved in the following ways: The customer uses their mobile banking APP to scan a dynamic QR code on an AR display device or touch interaction device to complete authorized login. The system then associates all behavioral data collected during this session with the customer's account ID and the device identifier bound to the APP; the branch provides visitor Wi-Fi, and customers need to receive a verification code via their mobile phone number for authentication when connecting. The system uses this mobile phone number as the customer's temporary identity identifier, and with the customer's consent, uses it to receive continued service content after leaving the store; for customers who do not actively log in, the account manager intervenes to provide service after the system pushes marketing leads. During communication, the account manager can enter the customer's contact information (such as mobile phone number, WeChat ID) into the system through their mobile terminal and manually bind it with the anonymous behavioral data file generated by the system. This application does not limit this aspect.

[0099] The bidirectional service device in the embodiments of this invention is described below from the perspective of hardware processing. Please refer to [link / reference]. Figure 2 This is a schematic diagram of the physical device structure of a two-way service device in an embodiment of this application.

[0100] It should be noted that, Figure 2 The structure of the bidirectional service device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0101] like Figure 2 As shown, the bidirectional service device includes a Central Processing Unit (CPU) 201, which can perform various appropriate actions and processes based on programs stored in Read-Only Memory (ROM) 202 or programs loaded from storage section 208 into Random Access Memory (RAM) 203, such as implementing the system described in the above embodiments. Various programs and data required for system operation are also stored in RAM 203. The CPU 201, ROM 202, and RAM 203 are interconnected via bus 204. Input / Output (I / O) interface 205 is also connected to bus 204.

[0102] The following components are connected to I / O interface 205: input section 206 including audio input devices, push-button switches, etc.; output section 207 including liquid crystal display (LCD) and audio output devices, indicator lights, etc.; storage section 208 including hard disks, etc.; and communication section 209 including network interface cards such as LAN (Local Area Network) cards, modems, etc. Communication section 209 performs communication processing via a network such as the Internet. Drive 210 is also connected to I / O interface 205 as needed. Removable media 211, such as disks, optical disks, magneto-optical disks, semiconductor memories, etc., are installed on drive 210 as needed so that computer programs read from them can be installed into storage section 208 as needed.

[0103] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing a computer program for performing the system shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 209, and / or installed from removable medium 211. When the computer program is executed by central processing unit (CPU) 201, it performs the various functions defined in the present invention.

[0104] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0105] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. Each block in a flowchart or block diagram may represent a module, program segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those shown in the drawings.

[0106] Specifically, the bidirectional service device in this embodiment includes a processor and a memory. The memory stores a computer program, and when the computer program is executed by the processor, it implements the steps of the AR-based bidirectional service system provided in the above embodiment.

[0107] In another aspect, the present invention also provides a computer-readable storage medium, which may be included in the bidirectional service device described in the above embodiments; or it may exist independently and not assembled into the bidirectional service device. The storage medium carries one or more computer programs that, when executed by a processor of the bidirectional service device, cause the bidirectional service device to implement the steps of the AR-based bidirectional service system provided in the above embodiments.

[0108] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0109] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as meaning "if...", "after...", "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is interpreted as meaning "if determining...", "in response to determining...", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

[0110] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. An AR-based two-way service system, characterized in that, include: An AR product display module is used to display product information to customers in an augmented reality manner through AR display devices deployed in financial outlets. The AR display devices support customers to interact with the three-dimensional virtual model through touch or gestures. The data acquisition module is used to collect customer behavior data within the financial outlet through the AR display device and other interactive devices set up in the financial outlet; The push decision module is used to generate marketing leads based on the behavioral data, and to determine the push recipients, push content, and push timing according to the customer's real-time status. The two-way collaborative push module includes a real-time push unit within the outlet and a continued push unit after the customer leaves the outlet; The real-time push unit within the branch is used to: push the marketing leads to the customer manager's mobile terminal in real time when the customer is located within the financial branch and the behavioral data meets the preset trigger conditions; The post-departure follow-up push unit is used to: after a customer leaves the financial outlet, push follow-up content related to the customer's unfinished experience at the outlet to the customer's mobile device based on the customer's departure behavior type and the customer manager's service record; The two-way collaborative push module is used to: feed back the customer's response to the content pushed after leaving the store to the customer manager's mobile terminal.

2. The system according to claim 1, characterized in that, The push decision module includes an intention judgment unit and a push strategy generation unit; The intention judgment unit is used to: calculate the customer intention intensity score based on the customer's dwell time, interaction depth, and number of repeated views on a specific product type in the behavioral data, and classify the customer into high-intent customers, hesitant customers, and comparison customers based on the score; The push strategy generation unit is used to generate differentiated push content for different types of customers. For high-intent customers, the content pushed to the account manager includes suggestions for immediate intervention and key recommended solutions. For hesitant customers, the content pushed to the account manager includes suggestions for continued monitoring. For comparison-oriented customers, the content pushed to the account manager includes a product differentiation comparison explanation.

3. The system according to claim 1, characterized in that, The preset triggering conditions include: The customer spends more than a preset time threshold on the interactive interface of any product type. If a customer views any information dimension more than a preset threshold number of times, the information dimension includes price information, risk rating, and rate of return. The customer's interaction depth reaches a preset level, and the interaction depth is determined by the number of information levels that the customer clicks to enter. The real-time push unit within the branch is also used to: simultaneously display the customer's real-time location information, the list of products viewed, and the dwell time of each product when pushing to the customer manager's mobile terminal.

4. The system according to claim 1, characterized in that, The data acquisition module includes an interactive device deployment unit and a behavioral data acquisition unit; The interactive device deployment unit is used to: deploy various interactive devices at the financial outlets, including AR display devices, touch interactive devices, and intelligent sensing devices; the AR display devices include AR glass curtain walls, used to display AR product information at the outlet entrance or display area; the touch interactive devices are used to receive touch operation commands from customers; The intelligent sensing device includes a facial recognition sensor and a spatial positioning sensor; The behavior data acquisition unit is used to collect customer behavior data through the interactive device, including: The AR display device collects customer browsing behavior data on the AR interface, including the types of products the customer views, the duration of stay at different information levels, and the frequency of switching product information. The touch interaction device collects the customer's touch operation data, which includes the coordinates of the click position, the swipe trajectory, the gesture type, the number of touch points for multi-touch, and the operation response time. The intelligent sensing device collects the customer's spatial location data, which includes the customer's real-time location coordinates, movement path, duration of stay in different product display areas, and number of times the customer approaches the product display area within the outlet. The behavior data collection unit is also used to: associate the browsing behavior data, touch operation data, and spatial location data to form behavior data, associate the behavior data with customer identity information to generate a customer behavior data profile, and transmit it to the push decision module.

5. The system according to claim 1, characterized in that, The continued content includes: Supplementary content for customers who did not complete the experience, including detailed comparative analysis of products that customers viewed at the store but did not understand in depth, and customer case studies; Extended content of in-store services includes product solutions recommended to customers by the account manager within the branch, personalized messages from the account manager, and the account manager's contact information; Personalized content is generated based on customer concerns, and targeted product interpretations are produced according to the information dimensions that customers focus on within the store.

6. The system according to claim 1, characterized in that, The bidirectional collaborative push module achieves a service closed loop through the following mechanism: Monitor customer response behavior to the content pushed after leaving the store, including clicking on the content, browsing the link, and filling out a feedback form; When the customer's response behavior is detected, feedback information is generated and pushed to the corresponding account manager's mobile terminal; Follow-up suggestions are displayed on the account manager's mobile terminal, including suggested timing for telephone follow-ups and suggested product features to be emphasized.

7. The system according to claim 6, characterized in that, Also includes: The departure behavior analysis module is used to classify customers' departure behavior into four categories based on their time spent at the branch, depth of interaction, and whether they communicate with the account manager: proactive departure, in-depth experience, service interruption, and decision hesitation. The post-departure follow-up push unit executes a differentiated push strategy based on the departure behavior type: For customers who voluntarily leave the store, push out a review of the store experience after the first preset time after leaving the store to guide customers to learn about the products again; For customers who are keen on in-depth experiences, a comparative analysis of the products they are interested in will be pushed to them after the second preset time after they leave the store, providing decision support. The second preset time is longer than the first preset time. For customers whose service is interrupted, a message containing information on the continuation of service from the account manager will be sent after the third preset time since the customer leaves the store, prompting them to schedule a follow-up communication. The third preset time is longer than the second preset time. For customers who are hesitant in making decisions, push notifications are sent at multiple preset time points after they leave the store, and the content of each push is adjusted according to the customer's response.

8. A two-way service device, characterized in that, The bidirectional service device includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors invoke the computer instructions to cause the bidirectional service device to perform the steps of the system as described in any one of claims 1-7 when executed.

9. A computer-readable storage medium comprising instructions, characterized in that, When the instructions are executed on the bidirectional service device, the bidirectional service device performs the steps of the system as described in any one of claims 1-7.

10. A computer program product, characterized in that, When the computer program product is run on a bidirectional service device, the bidirectional service device executes the steps of the system as described in any one of claims 1-7.

Images