Information processing system and program

Abstract

To provide a system for enabling a user to efficiently improve a web site in comparison with a system for outputting only behavior of the user estimated to contribute to conversion.SOLUTION: An information processing system is provided that has one or a plurality of processors configured to: acquire action history of a plurality of users on a website; receive a setting of conversion that is specific action of the user on the website; extract first action that is action constituting the acquired action history and second action from among the first action based on information indicating whether or not the received conversion is included in the action history; simulate a relation between change in a numerical value related to the extracted second action and change in a numerical value related to the conversion; and output simulation results.SELECTED DRAWING: Figure 5

Description

[Technical field]

[0001] The present invention relates to an information processing system and a program. [Background technology]

[0002] Today, businesses use web pages to provide advertising and other information and services, and therefore analyze historical data about users' browsing of their web pages and use the results of the analysis to help them update their web pages. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6866069 Summary of the Invention [Problem to be solved by the invention]

[0004] An object of the present invention is to provide a system that allows users to improve their website more efficiently than a system that outputs only user actions that are estimated to contribute to conversions. [Means for solving the problem]

[0005] The invention described in claim 1 is an information processing system having one or more processors, which acquires behavioral histories of multiple users on a website, accepts settings of conversions, which are specific actions by the users on the website, extracts second actions from the first actions based on a first action, which is an action that constitutes the acquired behavioral history, and information on whether the accepted conversion is included in the behavioral history, simulates the relationship between numerical changes related to the extracted second action and numerical changes related to the conversion, and outputs the results of the simulation. The invention described in claim 2 is the information processing system described in claim 1, wherein the one or more processors, when receiving a designation of the second action, output the results of the simulation for the received second action. The invention described in claim 3 is the information processing system described in claim 1, wherein the one or more processors, when receiving a specification of a numerical change amount related to the second action, output the result of the simulation for the received numerical change amount related to the second action. The invention described in claim 4 is the information processing system described in claim 1, wherein when the one or more processors receive a specification of a numerical change amount related to the conversion, they output the numerical change amount related to the second action that realizes the change amount as the result of the simulation. The invention described in claim 5 is the information processing system described in claim 1, wherein the one or more processors extract one or more characteristics of web pages contained in the website based on the results of the simulation, and output the relationship between the extracted characteristics and changes in numerical values ​​related to the conversion. The invention described in claim 6 is an information processing system described in claim 5, wherein the one or more processors output the relationship between changes in one or more characteristics of the web page and changes appearing in the conversion based on a weighting calculation according to the characteristics. The invention described in claim 7 is an information processing system described in claim 1, in which the one or more processors input a behavioral history and extract the second behavior based on a learning model that outputs the likelihood that a user assumed based on the input behavioral history will perform a behavior that is set as a conversion. The invention described in claim 8 is an information processing system described in claim 7, wherein the one or more processors update the weights set for each of the first actions through learning during the calculation process performed by the learning model, and extract the second actions based on the updated weights. The invention described in claim 9 is the information processing system described in claim 1, in which the one or more processors provide a first behavioral history as input to a learning model that receives an input of a behavioral history and outputs whether or not a user imagined based on the input behavioral history will perform an action set for a conversion, obtain a first result regarding the conversion based on the provided first behavioral history, provide a second behavioral history generated based on the first behavioral history as input to the learning model, obtain a second result regarding the conversion based on the provided second behavioral history, and simulate the relationship between the change in numerical value regarding the second behavior and the change in numerical value regarding the conversion based on the first result and the second result. The invention described in claim 10 is a program for causing a computer to realize the following functions: acquiring the behavioral history of multiple users on a website; accepting settings of conversions, which are specific actions by the users on the website; extracting a second action from the first actions, which are actions that constitute the acquired behavioral history, based on a first action that is an action constituting the acquired behavioral history and information as to whether the accepted conversion is included in the behavioral history; simulating the relationship between changes in numerical values ​​related to the extracted second action and changes in numerical values ​​related to the conversion; and outputting the results of the simulation. Effect of the Invention

[0006] According to the invention of claim 1, it is possible to provide a system that allows users to improve their websites more efficiently, compared to a system that outputs only user actions that are estimated to contribute to conversions. According to the invention described in claim 2, the effect that the designated second action has on the conversion can be output as a numerical value. According to the invention described in claim 3, the effect that the designated second action has on the conversion can be output as a numerical value. According to the invention described in claim 4, it is possible to output a change in a numerical value related to a second action that realizes a change in a numerical value designated for a conversion. According to the invention described in claim 5, it is possible to output the relationship between the change in the designated numerical value regarding the conversion and the change in the numerical value regarding the feature of the web page. According to the invention described in claim 6, it is possible to output the relationship between changes in the characteristics of a web page and changes appearing in conversions. According to the invention described in claim 7, even when a large number of behavioral histories are to be handled, it is possible to extract secondary behaviors that affect conversion. According to the invention described in claim 8, even when a large number of behavior histories are handled, the accuracy of extracting the secondary behavior that influences the conversion can be improved. According to the invention as recited in claim 9, the time required to output the simulation results can be shortened. According to the invention described in claim 10, a system can be provided that allows users to improve their websites more efficiently, compared to a system that outputs only user actions that are estimated to contribute to conversions. [Brief description of the drawings]

[0007] [Figure 1] FIG. 2 is a diagram illustrating an example of the configuration of an information processing system assumed in the first embodiment. [Diagram 2] FIG. 2 is a diagram illustrating an example of a hardware configuration of a business support server. [Diagram 3] FIG. 2 is a diagram illustrating an example of a hardware configuration of an information terminal. [Figure 4] FIG. 1 is a diagram illustrating information related to conversion of a web page. [Diagram 5] 11 is a flowchart illustrating an example of a processing operation executed by a business support server. [Figure 6] FIG. 13 is a diagram illustrating an example of a behavior history of a browsing user. [Figure 7] FIG. 13 is a diagram illustrating an example of a behavior history after grouping. [Figure 8] FIG. 13 is a diagram illustrating an example of a screen displaying the top N actions having the highest contribution to conversion. [Figure 9]11A and 11B are diagrams illustrating an example of a screen used for inputting simulation conditions and displaying simulation results. [Figure 10] 11 is a diagram illustrating an example of a method for generating pseudo improvement data used in simulating the effect of a change in a behavior-related value on a conversion-related value. FIG. [Figure 11] 13 is a diagram for explaining another example of a method for generating pseudo improvement data used in simulating the effect of a change in a behavior-related value on a conversion-related value. FIG. [Figure 12] 13 is a diagram for explaining the continuation of another example of the method for generating pseudo improvement data used in simulating the influence of a change in a behavior-related numerical value on a conversion-related numerical value. FIG. [Figure 13] FIG. 13 is a diagram illustrating a processing operation related to prediction of improvement in conversion. [Figure 14] 11A and 11B are diagrams illustrating an example of a screen used for inputting simulation conditions and displaying simulation results. [Figure 15] 11A and 11B are diagrams illustrating an example of a screen used for inputting simulation conditions and displaying simulation results. [Figure 16] 13A and 13B are diagrams illustrating changes in the display on the screen when modal information is added to the simulation conditions. [Figure 17] FIG. 1 is a diagram illustrating the process of machine learning a predictive model. [Figure 18] FIG. 13 is a diagram for explaining a process of extracting the top N actions. [Figure 19] FIG. 13 is a diagram illustrating an example of a simulation of a conversion using a prediction model. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0008] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. <Terminology> In this specification, the term "conversion" is used to mean a specific action taken by a user on a website, such as membership registration, subscription to an e-mail newsletter, product purchase, inquiry, download, application, reservation, or transition to a specific page. Conversions include not only cases where a user takes a specific action without leaving the target site, but also cases where a user leaves the target site and then returns to the target site to take a specific action.

[0009] Conversions may also include actions that occur just before downloading, applying, making a reservation, etc., such as actions that do not result in a download but reach a download page, actions that do not result in an application or reservation but reach an application page or reservation page, and actions that reach a confirmation screen for the information entered. In the following embodiment, a specific action will be described in the sense of a transition to a specific web page.

[0010] <Embodiment 1> <System configuration> Fig. 1 is a diagram for explaining a configuration example of an information processing system 1 assumed in the embodiment 1. The information processing system 1 shown in Fig. 1 is composed of a web server 10, a server (hereinafter referred to as a "business support server") 20 that provides a service for supporting a business that improves conversion to a web page, an information terminal 30, and a network N.

[0011] Web server 10 is a computer that displays an accessed web page in a web browser in response to a request from the web browser as a client. In Fig. 1, the provided web pages are denoted as WP1, WP2, .... In this embodiment, a collection of multiple web pages is called a website. The web page that is the entrance to a website is called a home page.

[0012] 1, the web browser is installed in at least one of the information terminals 30. An information terminal 30 in which a web browser is installed is called a viewing terminal. In this embodiment, a user who operates the viewing terminal is called a viewing user. The web server 10 records a history of web page browsing by a browsing user (hereinafter, also referred to as a “behavior history”). The web server 10 is an example of a terminal connected to a network N.

[0013] The business support server 20 is a computer that provides a user with the relationship between the change in the conversion value for a web page specified by an engineer or the like and the change in the value related to the behavior of the browsing user. In the present embodiment, improved conversion is used to mean an increase in conversion. In the case of FIG. 1, the business support server 20 is operated from at least one of the information terminals 30.

[0014] In this embodiment, the information terminal 30 used to access the business support server 20 is called a business terminal. Assuming that the user who operates the business terminal is an engineer who analyzes behavior that contributes to improving conversion. In this embodiment, the engineer is called a business user. The business support server 20 is also an example of a terminal connected to the network N.

[0015] The information terminal 30 is, for example, a desktop computer, a notebook computer, a tablet computer, or a smartphone. The information terminal 30 operating as a client terminal operated by a browsing user and the information terminal 30 operating as a user terminal operated by a business user may be one or more. The information terminal 30 is also an example of a terminal connected to the network N.

[0016] The network N is, for example, the Internet, a wireless LAN (=Local area network), 4G, 5G, or other mobile communication system. Figure 1 illustrates an example in which the web server 10, business support server 20, and information terminal 30 are connected to one network N, but the network N used for viewing web pages and the network N used for accessing the business support server 20 may be different.

[0017] <Server hardware configuration> 2 is a diagram illustrating an example of the hardware configuration of the business support server 20. The hardware configuration of the web server 10 is also basically the same as that of the business support server 20. 2 includes a processor 21, a ROM (=Read Only Memory) 22 in which a BIOS (=Basic Input Output System) and the like are stored, a RAM (=Random Access Memory) 23 used as a work area for the processor 21, an auxiliary storage device 24, and a communication interface 25. Each device is connected via a bus and other signal lines 26.

[0018] The processor 21 is a device that realizes various functions through the execution of programs. The processor 21, the ROM 22, and the RAM 23 function as a computer. The auxiliary storage device 24 is configured by, for example, a hard disk device or a semiconductor storage. Programs and various data are stored in the auxiliary storage device 24. The program here is used as a general term for the OS (=Operating System) and application programs. In the case of the business support server 20, one of the application programs is a support program.

[0019] The communication interface 25 is an interface for communicating with the information terminal 30 that accesses information that supports business operations through the network N. The communication interface 25 supports various communication standards. Examples of the communication standards include Ethernet (registered trademark), Wi-Fi (registered trademark), and mobile communication systems.

[0020] <Hardware configuration of information terminal> FIG. 3 is a diagram illustrating an example of a hardware configuration of the information terminal 30. As shown in FIG. The hardware configuration of the information terminal 30 shown in FIG. 3 is approximately the same as that of the business support server 20 (see FIG. 2). That is, the information terminal 30 has a processor 31, a ROM 32 in which the BIOS and the like are stored, a RAM 33 used as a work area for the processor 31, an auxiliary storage device 34, a communication interface 35, a display 36, and an input reception device 37. Each device is connected via a bus and other signal lines 38.

[0021] The display 36 is, for example, a liquid crystal display or an organic EL (Electro Luminescence) display. The input reception device 37 is, for example, a mouse or a keyboard. Incidentally, when the information terminal 30 is a notebook computer, a tablet computer, or a smartphone, a capacitive touch sensor having transparency that does not interfere with the visibility of the image displayed on the display 36 is used as the input reception device 37. A device that combines this type of touch sensor with a display is called a touch panel.

[0022] <Web page conversion related information> Fig. 4 is a diagram for explaining information related to web page conversion. In Fig. 4, conversion is written as "CV". Conversion involves a combination of "modal" and "behavior," as shown in Figure 4. In Figure 4, combinations of information are represented by the symbol X. A "modal" is information that gives a characteristic to a web page, and is composed of, for example, a combination of page structure, content, and design.

[0023] Figure 4 shows examples of page structures including "links" and "banners." Links are information that connects web pages and are also called "hyperlinks." Banners are information that introduces other websites or other web pages within the same website. Many banners have links to the websites or web pages they introduce. In Figure 4, "documents" and "images" are shown as examples of content. Documents consist of text written in HTML or other languages ​​and style sheets. Images include still images and video images. Figure 4 shows examples of design such as "color" and "layout."

[0024] "Behavior" is information related to the operation of a user viewing a web page. In Fig. 4, examples of behavior are transition, stay time, referral source, time period, search engine, referral type, browser, device, and region. There are two types of transitions: page transitions and inter-page transitions. A page transition is information indicating a transition to a specific web page. A source web page is not necessarily required. An inter-page transition is information indicating a transition from another web page to a specific web page. An inter-page transition may be specified by a source web page and a destination web page, or it may be information indicating a transition between three or more web pages.

[0025] The visit time is the time spent on a particular web page, in other words, the time during which a browsing user views a particular web page. The referrer is information indicating the link source web page when a link is clicked to transition to a specific web page. The time period is information indicating the time period during which a particular web page was displayed. The search engine is information indicating the name of the search engine used to refer to a particular web page.

[0026] Referral type is information that indicates the type of method used to refer to a particular web page, such as a search engine, manually entering a URL, or clicking a bookmark. The browser is information that indicates the web browser that was used to display a particular web page. The device is information that indicates the type of information terminal 30 used by the user who viewed the web page. The region is information indicating the location of the information terminal 30 on which the web browser operates. The location here is not limited to a location measured by a positioning system, but may be an estimated location. The location is not limited to a physical location, but may be a logical location on a network.

[0027] In addition, when it is possible to access a database that manages information about the user operating the information terminal 30 used to view web pages (hereinafter referred to as "user information"), information related to conversion to a specific web page may include user information such as the user's age, gender, address, etc. The number of combinations of information just listed as examples is enormous. For this reason, it is not easy to estimate behaviors that are highly related to conversions to a specific web page based on the experience and skills of engineers, etc.

[0028] <Processing operation of the business support server> Fig. 5 is a flowchart for explaining an example of a processing operation executed by the business support server 20. The symbol S shown in the figure means "step." The processing operation shown in Fig. 5 is realized through the execution of a program by the processor 21 (see Fig. 2). In this embodiment, the processing operation shown in FIG. 5 is started by an engineer or the like accessing the business support server 20 or the like.

[0029] <Step 1> First, the processor 21 acquires the behavior history of the browsing user from the web server 10 (step 1). The behavior history of the browsing user is stored in the web server 10 (see FIG. 1) as a so-called access log, and is also copied to the auxiliary storage device 24 of the business support server 20 (see FIG. 2). FIG. 6 is a diagram illustrating an example of a behavior history of a browsing user.

[0030] In the case of Figure 6, the actions that make up the behavioral history are exemplified as the viewing date and time 24B, entry page 24C, reference type 24D, search engine 24E, exit page 24F, previous page 24G, target page 24H, next page 24I, number of clicks 24J, browser 24K, and number of views 24L. In the case of FIG. 6, each row corresponds to the behavior history of a viewing user 24A (e.g., A, B, C, ...) in each viewing. Therefore, the behavior history of the same viewing user appears in multiple rows. Incidentally, each column is the behavior that constitutes the behavior history. In FIG. 6, the viewing user corresponding to the behavior history is written to explain the correspondence relationship. The same applies to other figures.

[0031] The viewing user 24A is information for identifying a user who viewed a particular web page. For example, the IP address of the information terminal 30 (see FIG. 1) operated by the user is stored in the viewing user 24A. The viewed date and time 24B stores information on the date and time when a specific web page was viewed. The entry page 24C stores information about the web page that is accessed first in a series of page transitions by the browsing user. The exit page 24F stores information about the last web page accessed by the browsing user in a series of page transitions.

[0032] The previous page 24G stores information indicating the web page that was viewed immediately before transitioning to a specific web page. The target page 24H stores information indicating a specific web page. The next page 24I stores information indicating a web page to which a transition will be made from a specific web page. The number of clicks 24J stores information indicating the number of clicks made by the browsing user during a series of page transitions. The browser 24K stores information indicating the name of the browser used by the browsing user to browse a web page. The number of times a web page has been viewed by a viewing user is stored in the number of times of viewing 24L.

[0033] <Steps 2 and 3> Next, the processor 21 accepts settings of a web page that is a target of conversion (step 2). Specifically, the processor 21 accepts designation of a web page that is a target of conversion from an information terminal 30 operated by an engineer or the like. Next, the processor 21 executes grouping of the behaviors related to the web page that is the target of the conversion (step 3). In the grouping, the behaviors related to the web page that is the target of the conversion are integrated for each browsing user.

[0034] Fig. 7 is a diagram for explaining an example of a behavior history after grouping. In Fig. 7, parts corresponding to those in Fig. 6 are denoted by the same reference numerals. Each behavior in the behavior history shown in Fig. 7 is organized as a behavior related to the web page that is the target of conversion. In the case of Fig. 7, the behaviors constituting the behavior history are exemplified as the browsing user 24A, the number (or rate) of transitions between pages 24M, the number (or rate) of page inflows 24N, the number (or rate) of page transitions 24O, the number (or rate) of referrers 24P, and the number (or rate) of session time periods 24Q.

[0035] The number (or rate) of page transitions 24M is information on the number of transitions from web page A to web page B for each browsing user or the transition rate obtained by dividing the number of transitions by the number of sessions. The page inflow number (or rate) 24N is information on the number of inflows per browsing user to web page A or the transition rate obtained by dividing the number of inflows by the number of sessions. The page transition count (or rate) 24O is information on the number of transitions to web page A for each browsing user or the transition rate obtained by dividing the number of transitions by the number of sessions.

[0036] The number of referrers (or rate) 24P is information on the number of inflows per browsing user from the referrer web page A or the inflow rate obtained by dividing the inflow rate by the number of sessions. The number of session time periods (or rate) 24Q is information on the number of time periods of sessions for each browsing user or the time period rate obtained by dividing the number of sessions.

[0037] <Step 4> Next, the processor 21 extracts the top N behaviors with the highest contributions to conversion (step 4). The contribution is a numerical value indicating the degree of possibility that a particular behavior contributes to conversion. In the case of this embodiment, the processor 21 arranges each of the actions (that is, each column) shown in FIG. 7 in order of the magnitude of the contribution degree, and extracts the top N actions.

[0038] Here, the number of the top N may be a fixed value, or may be specified by an engineer, etc. If the number can be specified by an engineer, etc., a field for specifying the number to be extracted is displayed on the display 36 of the information terminal 30 (see FIG. 3). The top N actions are examples of second actions extracted from the first actions.

[0039] In this embodiment, the contribution degree value is updated according to the following rule each time a new behavior history is acquired: if a specific web page is reached after one or more screen transitions, for example, "1" is added to the contribution degree value of the related first behavior, and if the specific web page is not reached, for example, "1" is subtracted from the contribution degree of the related first behavior. As a result, the contribution of certain actions that are likely to contribute to conversions will be greater, and the contribution of actions that are unlikely to contribute to conversions will be smaller. For example, the contribution of action Y, which appears 10 times in the history of transitions to web page A, will be greater than that of action X, which appears only once. In this way, the top N actions are extracted based on the actions (first actions) that constitute the action history and information on whether or not a conversion is included in the action history.

[0040] <Steps 5, 6, and 7> Next, the processor 21 displays the extracted top N actions (step 5). Specifically, the top N actions are displayed on the display 36 of the information terminal 30 operated by the engineer or the like. This display enables the engineer or the like to confirm actions that have a high degree of contribution to the viewing of the web page that is the target of conversion. In this embodiment, the magnitude of the contribution of each action is also displayed. Next, the processor 21 simulates the relationship between the change in the numerical value related to the specific behavior and the change in the numerical value related to the conversion (step 6).

[0041] In this embodiment, a change in the conversion value when the value related to a specific behavior changes is simulated. For example, an expected change in the conversion value when the value related to a specific behavior increases by 10% is calculated. When the simulation is completed, the processor 21 displays the simulation results (step 7). Specifically, the change in the numerical value related to the specific action and the change in the numerical value appearing in the conversion are displayed on the display 36 of the information terminal 30 operated by an engineer or the like.

[0042] <Example of processing operation> A specific example of the processing operation executed by the business support server 20 will be described below. <Top N behavior display screen> Fig. 8 is a diagram illustrating an example of a screen 300 that displays the top N actions having the highest contribution to conversion. The screen 300 shown in Fig. 8 is displayed on the display 36 (see Fig. 3) of the information terminal 30 (see Fig. 1) operated by an engineer or the like.

[0043] 8 is composed of a conversion input field 301, a display number input field 302, a field 303 for specifying the behavior to be simulated, an input field 304 for the improvement value required for the specified behavior, a display field 305 for the top N behaviors, and a total number of viewing users field 306. However, the contents and numbers of the displayed items are merely examples. In the case of FIG. 8, "Access page XX" is displayed in the conversion input field 301. Page XX refers to web page XX. For example, it is "Product List." Incidentally, a list of one or more web pages that make up the website is prepared in advance in the input field 301. The list of one or more web pages is displayed when a pull-down menu is opened.

[0044] The display number input field 302 is used to specify the number of actions to be displayed as actions that contribute highly to conversion. The numerical value here corresponds to N in step 4 (see FIG. 5) or step 5 (see FIG. 5). In the case of FIG. 8, "12" is input. "12" in FIG. 8 is the initial value. Therefore, an engineer or the like can change the display number. In the case of FIG. 8, the numerical value in the input field 302 can be selected by opening a pull-down menu.

[0045] The behavior specification field 303 is used to specify one or more behaviors to be used in the simulation. In the case of FIG. 8, a total of seven check boxes are displayed, corresponding to "All," "Transition," "Referrer," "Inflow," "Time Period," "Inter-page Transition," and "Exit." As these are check boxes, it is possible to select not just one but multiple actions. When "All" is selected, the other six actions are selected. However, at the time of FIG. 8, all of the check boxes are blank, since the top N actions have just been displayed.

[0046] The improvement value input field 304 is used to input the degree of improvement desired for the numerical value related to the selected behavior. At the time of Fig. 8, since the top N behaviors have just been displayed, the numerical value representing the degree of improvement is not displayed. The top N behaviors (12 in FIG. 8) extracted in step 4 (see FIG. 5) are displayed in the top N behaviors display column 305. In addition, in the column to the right of the top N behaviors, the magnitude of the contribution is displayed as the length of a bar graph.

[0047] In the case of FIG. 8, it can be seen that the contribution of "number of transitions_page A" is significantly larger than the other 11. The total number of viewing users column 306 indicates that a total of 200,000 viewing users were used in the analysis, and the target period was five months from "20220801" to "20221231."

[0048] <Simulation result display screen> Fig. 9 is a diagram for explaining an example of a screen 310 used for inputting simulation conditions and displaying simulation results. In Fig. 9, parts corresponding to those in Fig. 8 are denoted by the same reference numerals. A screen 310 shown in FIG. 9 is also displayed on the display 36 (see FIG. 3) of the information terminal 30 (see FIG. 1) operated by an engineer or the like. 9, "10" is input in the display number input field 302. Therefore, 10 actions reflecting the simulation results are displayed in the top N action display field 305.

[0049] In addition, in the case of FIG. 9, “Transition”, “Flow-in”, and “Inter-page transition” are selected in the action specification field 303, and “10%” is entered in the improvement value input field 304 as a numerical value indicating the degree of improvement. As a result, in the display column 305 for the top N behaviors, under the title "Predicted improvement in CV when the numerical value of each behavior is improved by 10%," the predicted improvement in conversions for each behavior is displayed in a bar graph. From the display in FIG. 9, it can be seen that if the "number of inter-page transitions" from web page C to web page A increases by 10%, the conversion to web page XX increases by 35 points or more.

[0050] By displaying display field 305 of this screen 310, engineers and the like can easily check not only the actions of viewing users that are estimated to contribute to conversion, but also the differences in the degree to which each action contributes to improving conversion. As a result, engineers and others will be able to more efficiently carry out work related to actions that contribute to improving conversions or to revising web page XX, compared to a system that outputs only user actions that are estimated to contribute to conversions.

[0051] <Generating pseudo improvement data required for simulation> FIG. 10 is a diagram for explaining an example of a method for generating pseudo improvement data used in simulating the influence of a change in a behavior-related value on a conversion-related value. FIG. 10 shows processing stage ST1 and processing stage ST2. The processing stage ST1 represents the relationship between the number of transitions for each web page and the browsing users before the simulation is executed. Note that "Number of page A transitions" represents the aggregated value of the number of transitions to web page A, and the other columns "Number of page B transitions," "Number of page C transitions," "Number of page D transitions," and "Number of page XX transitions" also represent the aggregated values ​​of the number of transitions for each web page.

[0052] Here, the "number of transitions to page A" is an example of a second action extracted from a first action. 10, the number of transitions to web page A by viewing user A is 1, the number of transitions to web page A by viewing user B is 2, the number of transitions to web page A by viewing user C is 2, the number of transitions to web page A by viewing user D is 1, and the number of transitions to web page A by viewing user E is 4. In this case, the average number of transitions to web page A is calculated to be "2."

[0053] The processing stage ST2 is a diagram for explaining an example of generating pseudo improvement data in the case where a 20% increase in the "number of transitions" is required. A 20% increase in the average value of 2 gives a value of 2.4. To increase the average value, data on browsing users whose number of transitions is less than the average value is deleted, and data on browsing users whose number of transitions is greater than the average is added. In the case of the processing stage ST2, the data of browsing user C is deleted and overwritten with the data of browsing user E instead.

[0054] As a result, the average number of transitions to web page A calculated using the pseudo-improved data is calculated to be "2.4." In other words, pseudo-improved data that satisfies a 20% increase in the "number of transitions" is created by replacing the browsing user data once. However, even if data of a viewing user having a number of transitions smaller than the average is replaced with data of a viewing user having a number of transitions larger than the average in processing stage ST1, the average value of the pseudo-improved data after replacement does not necessarily satisfy the specified 20% increase.

[0055] Fig. 11 is a diagram for explaining another example of a method for generating pseudo improvement data used in simulating the effect of a change in a behavior-related value on a conversion-related value. In Fig. 11, the parts corresponding to those in Fig. 10 are denoted by the same reference numerals. FIG. 11 also shows processing stage ST1 and processing stage ST2. The difference is that the data of browsing user C is replaced with the data of browsing user D in the processing stage ST2.

[0056] 11, for convenience of explanation, the number of transitions to web page A by browsing user D is changed to "3." The other transition numbers are the same as those in FIG. In the case of FIG. 11, the data of browsing user D, whose transition count is "3", is used as a browsing user whose transition count is greater than "2.4". Therefore, the average number of transitions to web page A in the generated pseudo-improvement data is calculated to be "2.2." However, the average value of the generated pseudo-improvement data does not meet the specified 20% increase in the "number of transitions". Therefore, the deletion of the viewing user's data and the duplication of the data of other viewing users are repeated until the 20% increase in the number of transitions is met.

[0057] Fig. 12 is a diagram for explaining the continuation of another example of the method for generating pseudo improvement data used in simulating the effect of changes in numerical values ​​related to behavior on numerical values ​​related to conversion. In Fig. 12, the parts corresponding to those in Fig. 11 are denoted by the same reference numerals. FIG. 12 shows processing stage ST2 and processing stage ST3. In the processing stage ST3 in FIG. 12, the data of browsing user A is replaced with the data of browsing user D. Therefore, the average number of transitions to web page A in the generated pseudo-improvement data is calculated to be "3." This number corresponds to a 50% increase in the "number of transitions," which satisfies the 20% increase. This completes the creation of pseudo improvement data required for calculating the amount of change in the numerical value related to conversion.

[0058] <Conversion improvement prediction> FIG. 13 is a diagram for explaining a processing operation related to prediction of improvement in conversion. As mentioned above, the "number of transitions to page A" is an action that contributes highly to conversions of web page XX, but it is merely one of the actions with a high contribution rate, and the number of conversions to web page XX is also related to actions other than the "number of transitions to page A." In FIG. 13, the numerical value of the second action (i.e., the number of transitions to page A) used to generate the pseudo-improvement data is denoted as the original data.

[0059] In the case of Figure 13, there are two browsing users whose behavior history includes conversions to web page XX: browsing users D and E. Therefore, the actual measured value of the number of conversions in the original data is "2," and the actual measured value of the number of conversions in the pseudo-improved data is "3." The processor 21 (see FIG. 2) calculates the conversion improvement rate and the number of conversion improvements by comparing the obtained numbers of conversions, and displays them on the display 36 of the information terminal 30 operated by an engineer or the like. In the case of Figure 13, the number of conversion improvements is "1." The conversion improvement rate is calculated as 50% by dividing the number of improvements "1" by the number of conversions in the original data, "2."

[0060] <Embodiment 2> This embodiment is also premised on the information processing system 1 described in the first embodiment. In the above-mentioned first embodiment, the case where the amount of change in the value related to the top N displayed actions is specified and the effect of the change in the action on the value related to the conversion is simulated is described, but in the present embodiment, the relationship between the premise and the result is reversed. That is, the case where the amount of change in the value related to the conversion is specified and the amount of change in the top N actions that makes that change possible is simulated is described.

[0061] Fig. 14 is a diagram for explaining an example of a screen 320 used for inputting simulation conditions and displaying simulation results. In Fig. 14, the same reference numerals are used to denote parts corresponding to those in Fig. 9. A screen 320 shown in FIG. 14 is also displayed on the display 36 (see FIG. 3) of the information terminal 30 (see FIG. 1) operated by an engineer or the like. The difference with screen 310 shown in FIG. 9 is that the title of the improvement value input field 304 has been changed to "Conversion Improvement Rate," and the actions displayed in display field 305 are "Top N actions that achieve conversion improvement rate."

[0062] In the case of FIG. 14, "Accessed page XX" is displayed in the conversion input field 301, "5" is displayed in the number of views input field 302, and "10%" is entered in the improvement value input field 304. For this reason, display section 305 is given the title "Improvement rate of each action required to improve conversion by 10%" and displays five actions.

[0063] In the display field 305, actions with smaller changes in the values ​​are displayed higher. In the example of Fig. 14, "Number of page transitions_Page C -> Page A" is displayed at the top. The length of the bar graph for "Number of page transitions_Page C -> Page A" is 20%. This means that if you can increase the number of visitors from web page C to web page A by 20%, you can improve conversions by 10%. To achieve a similar improvement, a 50% increase in transitions from web page D to web page A would be required, a 60% increase in transitions from web page E to web page A, a 70% increase in the number of visitors to web page A, and a 120% increase in the number of transitions to web page A would be required.

[0064] Incidentally, the extraction of the improvement rate required for the top N actions (i.e., the second actions) to satisfy the specified conversion improvement rate is realized by reverse lookup from the relationship calculated by the method described in the first embodiment. The method will be described below. First, various improvement rates are assigned to each of the top N actions to calculate the conversion improvement rate, etc. Next, samples that satisfy the numerical value in the input field 304 are extracted from the calculated conversion improvement rates, and the top N actions used in the calculation of the extracted samples and the corresponding improvement rates are extracted. Through the above processing, the display of screen 320 shown in FIG. 14 is realized.

[0065] <Embodiment 3> This embodiment is also premised on the information processing system 1 described in the first embodiment. In this embodiment, a case will be described where it is possible to select weighting based on a modal that represents the characteristics of a web page in addition to the improvement rates of the top N actions. Fig. 15 is a diagram for explaining an example of a screen 330 used for inputting simulation conditions and displaying simulation results. In Fig. 15, parts corresponding to those in Fig. 9 are denoted by the same reference numerals.

[0066] Screen 330 shown in Fig. 15 differs from screen 310 shown in Fig. 9 in that a selection field 331 for specifying whether or not to apply modal weighting is added. The other screen configurations of screen 330 are the same as those in Fig. 9. In the case of Fig. 15, the selection field 331 is titled "Modal Weighting." Note that in the case of Fig. 15, no check mark is placed in the selection field 331. In other words, modal weighting has not been selected. 9. Therefore, the display in the display field 305 in FIG. 15 is the same as the display in the display field 305 in FIG.

[0067] Fig. 16 is a diagram for explaining the change in display on screen 330 when modal information is added to the simulation conditions. In Fig. 16, parts corresponding to those in Fig. 15 are denoted by the same reference numerals. As described above, in the case of Fig. 16, a check mark is placed in the selection column 331. In other words, modal weighting is selected.

[0068] As explained with reference to Fig. 4, modal is information on the page structure, contents, and design of a webpage, for example. Therefore, when modal weighting is enabled, it means evaluating the influence of the page structure, contents, and design of a webpage and the behavior of a browsing user on conversion. The influence of a modal on a conversion is evaluated, for example, as follows: As a premise for the evaluation, for example, a weighting coefficient for each modal is set in advance by the user for each action.

[0069] For example, the weighting coefficient of "page structure" for "inter-page transition" is set to 1.0, the weighting coefficient of "content" is set to 0.7, and the weighting coefficient of "design" is set to 0.3. In addition, the weighting coefficient of "page structure" for "traffic" is set to 0.8, the weighting coefficient of "content" is set to 0.8, and the weighting coefficient of "design" is set to 0.3. With this setting, a weighting coefficient of 1.0 is used to represent the influence of "page structure" on inter-page transitions (e.g., from page A to page B), and a weighting coefficient of 0.8 is used to represent the influence of "content" on the "incoming page." These weighting factors may be stored as a matrix by the system and may be changeable by the user.

[0070] In the case of the display field 305 shown in FIG. 16, the top ranking of the actions that contribute most to the improvement of conversion remains unchanged, but the length of the bar graphs representing the improvement of conversion changes. For example, if the second-highest conversion improvement rate, “Number of page transitions from page D to page A,” is improved by 10%, the conversion improvement rate drops from over 35% to below 35% when “structure,” one of the modals, is also included. In addition, for conversion improvement rates ranked 4th and below, the improvement numbers are all lower than before weighting by modality. By observing changes in the length of the bar graph and the order of actions, engineers can easily understand the relationship between actions and modals that affect conversion improvements.

[0071] <Fourth embodiment> This embodiment is also premised on the information processing system 1 described in the first embodiment. In this embodiment, a specific example of the processing operation in step 4 (see FIG. 5) will be described. In this embodiment, we will explain the case where the top N actions (secondary actions) are extracted using a predictive model that uses machine learning to determine the relationship between the actions (first actions) that make up the behavioral history and actions that are likely to lead to conversion (i.e., the degree of contribution).

[0072] FIG. 17 is a diagram for explaining the process of machine learning the prediction model 200. In the case of this embodiment, the prediction model 200 employs a learning model based on, for example, a Gradient Boosting Decision Tree (GBDT) or other decision trees. In the machine learning of the prediction model 200 shown in FIG. 17, for example, a part of the behavior history acquired in step 1 (see FIG. 5) is used as training data.

[0073] The prediction model 200 shown in FIG. 17 is a model that has learned the relationship of inputting an arbitrary behavior history (i.e., a collection of behaviors (first behaviors)) and outputting the possibility of conversion linked to the behavior (first behavior). Note that the possibility here corresponds to the aforementioned degree of contribution. Therefore, the higher the possibility, the higher the probability of conversion to the specified web page. The output (prediction data) of the prediction model 200 is provided to the evaluation unit 210 and compared with the correct answer data. In this embodiment, this comparison is called evaluation.

[0074] When the difference between the two pieces of data is obtained, the evaluation unit 210 updates the weight set for the first action in the calculation process of the prediction model 200 so as to reduce the difference. As the correct answer data, for example, a part of the behavior history acquired in step 1 (see FIG. 5) is used. By repeating prediction by the prediction model 200 and evaluation by the evaluation unit 210, a prediction model 200 with high prediction accuracy is generated.

[0075] Fig. 18 is a diagram for explaining the process of extracting the top N actions (second actions). In Fig. 18, parts corresponding to those in Fig. 17 are denoted by the same reference numerals. The prediction model 200 here is a learning model generated by machine learning, as shown in FIG. When a behavioral history is input into the prediction model 200, behaviors linked to the likelihood of leading to a conversion are output.

[0076] The extraction unit 220 sorts the input actions in order of magnitude of the numerical value indicating the possibility of leading to a conversion, and outputs the top N actions. By generating the predictive model 200 through machine learning, it becomes possible to accurately predict behaviors (second behaviors) that are likely to lead to conversion, even when the number of behavioral histories being handled or the number of behaviors that make up the behavioral history are large.

[0077] <Embodiment 5> This embodiment is also premised on the information processing system 1 described in the first embodiment. In this embodiment, a specific example of the processing operation in step 6 (see FIG. 5) will be described. Fig. 19 is a diagram for explaining an example of a simulation of conversion using the prediction model 230. In Fig. 19, parts corresponding to those in Fig. 13 are denoted by the same reference numerals. A learning model based on, for example, a Gradient Boosting Decision Tree (GBDT) or other decision trees is also adopted for the prediction model 230 shown in FIG.

[0078] The prediction model 230 here is a learning model that inputs the behavior history specified in the specification field 303 (see FIG. 8) from among the behavior history, and outputs a predicted value of the number leading to conversion. The "predicted value of the number leading to conversion" here is an example of a result related to conversion. The machine learning of predictive model 230 proceeds in a similar manner to predictive model 200 (see FIG. 17). In this embodiment, the "original data" is an example of a first behavioral history, and the "pseudo-improved data" is an example of a second behavioral history. Therefore, the result output by the prediction model 230 for the "original data" is called the first result, and the result output by the prediction model 230 for the "pseudo-improved data" is called the second result.

[0079] In the case of Fig. 19 as well, the number of conversions in the original data is predicted to be "2", and the number of conversions in the pseudo-improved data is predicted to be "3". For this reason, in Fig. 19, the output of the prediction model 230 is represented as "predicted conversion number 2" and "predicted conversion number 3". When the predicted conversion number is obtained, the processor 21 (see FIG. 2) calculates the conversion improvement rate and the conversion improvement number, and displays them on the display 36 of the information terminal 30 operated by an engineer or the like.

[0080] In the case of Figure 19, the improvement in conversion is also "1". Moreover, the improvement rate of conversion is calculated as 50% by dividing the improvement in conversion "1" by the conversion number of the original data "2". In the case of this embodiment, the time required to calculate the number of conversions is shorter than in embodiment 1. As a result, it is possible to reduce the time required to output the simulation results.

[0081] <Other embodiments> (1) Although the embodiment of the present invention has been described above, the technical scope of the present invention is not limited to the scope of the above-mentioned embodiment. It is clear from the claims that the technical scope of the present invention also includes various modifications and improvements to the above-mentioned embodiment.

[0082] (2) In the above-mentioned screen 330 (see FIG. 15) in the third embodiment, the case where the presence or absence of weighting by modality can be selected when calculating the impact on improving conversion has been described. However, it may also be possible to set the presence or absence of weighting for each item constituting a modality, such as page structure, content, and design, or it may be possible to set the presence or absence of weighting for each lower-level structure, such as link, banner, document, image, color, and layout. Furthermore, even when weighting is selected, the magnitude of the weighting may be made configurable.

[0083] (3) In the above-described fourth and fifth embodiments, a gradient boosting tree is used for the prediction model. However, a neural network composed of an input layer, an intermediate layer, and an output layer may be used.

[0084] (4) The processor in the above-described embodiments refers to a processor in a broad sense, and includes general-purpose processors (e.g., CPU (=Central Processing Unit)) as well as dedicated processors (e.g., GPU (=Graphical Processing Unit), ASIC (=Application Specific Integrated Circuit), FPGA (=Field Programmable Gate Array), programmable logic devices, etc.). In addition, the operations of the processor in each of the above-mentioned embodiments may be executed by one processor alone, or may be executed by multiple processors located at physically separate locations in cooperation with each other. Furthermore, the order of execution of each operation in the processor is not limited to the order described in each of the above-mentioned embodiments, and may be changed individually.

[0085] (5) In the above-described embodiment, all of the processes for providing a service that supports a business to improve conversion to a web page are executed in the business support server 20 (see FIG. 1), but all of the corresponding processes may be executed in the information terminal 30 (see FIG. 1). Also, the processes may be executed by cooperation between the business support server 20 and the information terminal 30. In other words, the above-described processing operations may be realized by cooperation between a plurality of processors.

[0086] <Additional Notes> (((1))) An information processing system having one or more processors, which acquires behavioral histories of multiple users on a website, accepts settings for conversions, which are specific actions by the users on the website, extracts second actions from the first actions based on a first action that constitutes the acquired behavioral history and information on whether the accepted conversion is included in the behavioral history, simulates the relationship between changes in numerical values ​​related to the extracted second action and changes in numerical values ​​related to the conversion, and outputs the results of the simulation. (((2))) The information processing system according to (((1))), wherein when the one or more processors receive a designation of the second action, they output a result of the simulation for the received second action. (((3))) The information processing system described in (((1))) or (((2))), wherein when the one or more processors receive a specification of a change in a numerical value related to the second action, they output the results of the simulation regarding the received change in the numerical value related to the second action. (((4))) The information processing system of any one of (((1))) to (((3))), wherein when the one or more processors receive a specification of a change in a numerical value related to the conversion, the one or more processors output, as a result of the simulation, a change in a numerical value related to the second action that realizes the change. (((5))) The information processing system of any one of (((1))) to (((4))), wherein the one or more processors extract one or more characteristics of a web page included in the website based on a result of the simulation, and output a relationship between the extracted characteristics and a change in a numerical value related to the conversion. (((6))) The information processing system of (((5))), wherein the one or more processors output a relationship between a change in one or more characteristics of the web page and a change in the conversion based on a weighting calculation according to the characteristics. (((7))) The one or more processors extract the second behavior based on a learning model that inputs a behavioral history and outputs the likelihood that a user assumed based on the input behavioral history will perform a behavior that is set as a conversion. An information processing system according to any one of (((1))) to (((6))). (((8))) The one or more processors update weights set for each of the first actions through learning during the calculation process performed by the learning model, and extract the second actions based on the updated weights. (((9))) The one or more processors input a behavioral history to a learning model that outputs whether or not a user hypothesized based on the input behavioral history will perform a behavior set for a conversion, provide a first behavioral history as input, and obtain a first result regarding the conversion based on the provided first behavioral history, provide a second behavioral history generated based on the first behavioral history as input to the learning model, and obtain a second result regarding the conversion based on the provided second behavioral history, and simulate a relationship between a change in numerical value regarding the second behavior and a change in numerical value regarding the conversion based on the first result and the second result. (((10))) A program for enabling a computer to perform the following functions: acquiring the behavioral history of multiple users on a website; accepting settings for conversions, which are specific actions by the users on the website; extracting a second action from the first actions, which are actions that constitute the acquired behavioral history, based on information on whether the accepted conversion is included in the behavioral history; simulating the relationship between changes in numerical values ​​related to the extracted second action and changes in numerical values ​​related to the conversion; and outputting the results of the simulation.

[0087] The information processing system according to (((1))) can provide a system that allows users to improve their websites more efficiently, compared to a system that outputs only user actions that are estimated to contribute to conversions. According to the information processing system of (((2))), the influence that the designated secondary action has on the conversion can be output as a numerical value. According to the information processing system of (((3))), the effect that the designated secondary action has on conversion can be output as a numerical value. According to the information processing system of (((4))), it is possible to output a change in a numerical value related to a second action that realizes a change in a numerical value designated for a conversion. According to the information processing system of (((5))), it is possible to output the relationship between the change in a specified numerical value regarding conversion and the change in a numerical value regarding the characteristics of a web page. According to the information processing system of (((6))), it is possible to output the relationship between changes in the characteristics of a web page and changes that appear in conversions. According to the information processing system of (((7))), even when a large number of behavioral histories are handled, it is possible to extract secondary behaviors that affect conversion. According to the information processing system of (((8))), even when a large number of behavior histories are handled, the accuracy of extracting secondary behaviors that affect conversion can be improved. According to the image processing system of (((9))), the time required to output the simulation results can be reduced. The program according to (((10))) can provide a system that allows users to improve their website more efficiently, compared to a system that outputs only user actions that are estimated to contribute to conversions. [Explanation of symbols]

[0088] 1...information processing system, 10...web server, 20...business support server, 21, 31...processor, 22, 32...ROM, 23, 33...RAM, 24, 34...auxiliary storage device, 25, 35...communication interface, 26, 38...signal line, 30...information terminal, 36...display, 37...input reception device, 210...evaluation unit, 220...extraction unit

Claims

1. one or more processors; the one or more processors: Acquire the behavioral history of multiple users on a website, Accepting a setting of a conversion, which is a specific action by the user on the website; extracting a second action from the first action based on a first action constituting the acquired action history and information on whether the accepted conversion is included in the action history; A relationship between a change in the extracted numerical value related to the second behavior and a change in the numerical value related to the conversion is simulated; outputting the results of the simulation; Information processing system.

2. the one or more processors: when the designation of the second action is accepted, a result of the simulation for the accepted second action is output. The information processing system according to claim 1 .

3. the one or more processors: when a designation of a change amount of the numerical value related to the second action is accepted, a result of the simulation regarding the accepted change amount of the numerical value related to the second action is output. The information processing system according to claim 1 .

4. the one or more processors: when a designation of a change amount of the numerical value related to the conversion is accepted, a change amount of the numerical value related to the second action that realizes the change amount is output as a result of the simulation; The information processing system according to claim 1 .

5. the one or more processors: Based on the results of the simulation, extracting one or more features of a web page included in the website, and outputting a relationship between the extracted features and a change in the numerical value related to the conversion; The information processing system according to claim 1 .

6. the one or more processors: outputting a relationship between a change in one or more features of the web page and a change in the conversion based on a weighting calculation according to the features; 6. The information processing system according to claim 5.

7. the one or more processors: extracting the second behavior based on a learning model that receives an action history as an input and outputs a possibility that a user assumed based on the input action history will perform a behavior set as a conversion; The information processing system according to claim 1 .

8. the one or more processors: updating weights set for the respective first actions during a calculation process performed by the learning model through learning, and extracting the second actions based on the updated weights; The information processing system according to claim 7.

9. The one or more processors: a learning model that receives an action history as an input and outputs whether or not a hypothesized user performs an action set for a conversion based on the input action history, the learning model receives a first action history as an input, and obtains a first result regarding the conversion based on the first action history; providing a second behavioral history generated based on the first behavioral history as an input to the learning model, and obtaining a second result regarding the conversion based on the provided second behavioral history; simulating a relationship between a change in a value related to the second action and a change in a value related to the conversion based on the first result and the second result; The information processing system according to claim 1 .

10. On the computer, A function for acquiring the behavioral history of multiple users on a website; A function of accepting settings of a conversion, which is a specific action by the user on the website; A function of extracting a second action from the first action, which is an action constituting the acquired action history, based on information as to whether the accepted conversion is included in the action history; A function of simulating a relationship between a change in a numerical value related to the extracted second behavior and a change in a numerical value related to the conversion; A function for outputting the results of the simulation; A program to achieve this.

Images