RU2703968C1 - Method and system for interactive targeted video advertisement based on user sessions history - Google Patents

Abstract

FIELD: interactive television.SUBSTANCE: method of controlling interactive video advertisement includes the following actions: in the user device forming an interactive layer request and sending it to the advertisement server, a set of interactive layers is formed on the server for subsequent selection, an interactive layer is selected in the advertisement server based on the expected efficiency, the selected interactive layer data is transmitted from the advertising server to the user device, events associated with the user actions with respect to the interactive layer during the current session are recorded in the user device.EFFECT: technical result is aimed at widening the range of tools for the same purpose.8 cl, 5 dwg

Description

Technical field

The invention relates to the field of interactive television, in particular, to a method and system for implementing a targeted interactive layer displayed over a layer of video content in so-called “smart TVs” (SmartTVs), video consoles, mobile devices and personal computers. Targeting is provided by taking into account the history of user sessions.

State of the art

A video player is a computer program designed to play video content (file or streaming). Video players are present by default in most modern operating systems, including operating systems for personal computers, mobile devices, SmartTVs and set-top boxes. Some applications may use their own video player or extend the capabilities of a standard video player. Applications of online cinemas and television channels allow you to view video content (for example, movies or television programs) using video players, including using interactive elements.

Interactive elements can serve, for example, for navigation by means of a video player. In addition, interactive elements may be associated with additional content corresponding to the current video content. The last use of interactive elements is hereinafter referred to as the interactive layer, since these interactive elements complement the video content, and do not act independently of it.

Interactive layers are quite common in video advertising, especially for personal computer browsers on the Internet, but they can also be found on mobile devices, SmartTVs and, in some cases, on set-top boxes. Interactive layers can also be used to provide additional content to films or television programs in the form of information, voting forms, polls, etc.

To implement proper management and statistics collection, advertising content (including advertising videos and interactive layers) is usually requested by terminal devices from advertising servers. Advertising servers are able to make decisions about what kind of advertising material to send to each specific terminal device, based on the information about this terminal device known to the server. Terminal devices are usually able to record events that occur when a user interacts with advertising content. In particular, the key indicators commonly defined for video advertising include the following events and their parameters: display, selection (the so-called “click” in the presence of interactive elements), viewing depth (which part of the video the user managed to see before stopping viewing video ads), engagement (total number of user interactions with the interactive layer).

Existing video advertising standards describe formats of advertising content for display on various devices, as well as methods for transmitting advertising content and mechanisms for recording events in terminal devices.

The VPAID (Digital Video Player-Ad Interface Definition) and MRAID (Mobile Rich Media Ad Interface Definitions) standards describe the interfaces through which interactive video advertising can be integrated into a video player. These standards relate to the algorithms of the video player, which are available for use in the source code of programmable advertising content, as well as to methods of interaction with the user. Both standards require the use of HTML5 and JavaScript technologies for correct operation. On some types of devices (in particular, on set-top boxes), these technologies may be unavailable or insufficiently productive.

The VAST (Digital Video Ad Serving Template) standard describes the format for transferring advertising materials from the advertising server to the video player of the terminal device, as well as the expected behavior of the video player. Among other things, the standard describes a mechanism for tracking events related to user interactions with advertising content. It is worth noting that the VAST standard does not provide for the grouping of events by user session and describes only anonymous registration of events without any binding to users, sessions, or devices.

The VAST standard mentions how events are recorded using so-called “audit pixels” - links that are attached to 1-pixel images and address the user to server scripts or other resources. This method is very common in an advertising environment. However, this method provides the ability to obtain statistics only about the total number of events and does not allow you to analyze these events for correlation for a specific user. For example, the audit-pixel method, together with A / B testing and impression targeting (most often by socio-demographic characteristics), allows you to determine the effectiveness of advertising in user groups, but does not allow you to determine the effectiveness of advertising for each individual user. Consequently, the existing method provides not quite accurate advertising targeting, focusing on wide groups instead of individual users.

Interactive layers (also called creatives) most often consist of a plate (banner) with a call to action (usually a transition to additional content) and a microsite consisting of information pages with additional information, internal navigation and input elements (for example, for input phone number, email address or confirmation code from SMS message). The terms plate and banner are commonly used interchangeably; the banner is often called a small banner (most often elongated in rectangular shape), while the banner may look like a button, a special picture of an unusual shape, a full-screen image, etc. The format of the plate (banner) for various creatives can vary greatly in shape, location and content. Nevertheless, there are several main types of dice (banners):

- by the nature of the call to action: an information plate (for example, "Learn more"), a commercial offer (for example, "Get a discount", "Order");

- in the form of a plate: banner-button, L-banner (banner in the form of the letter L, leaving part of the screen for video content and using the remaining space for additional information), stop-banner (banner, demonstrated for a limited time with stopping video content), pack (full-screen banner, usually shown for a limited time at the end of the video).

The first interaction of the user with the interactive layer is a click on the plate / banner (on some devices - pressing the “OK” button on the remote control or a similar action). For advertisers, it is often important to increase the likelihood of the first interaction. Therefore, it is important to choose the appropriate type of die (form and call to action). It should be borne in mind that different users may respond differently to dice. Some users may be more susceptible to a button banner, while others may be more susceptible to a stop banner. Existing video advertising standards and existing methods of recording events do not allow collecting enough information to determine user groups by their preferred banner types.

The present invention implements a method for recording events, taking into account the grouping of these events into sessions and users, as well as a method for analyzing these events to target interactive layers corresponding to user preferences.

Disclosure of invention

The registration of events taking into account the grouping of these events by sessions and users, as well as the analysis of these events with the subsequent targeting of interactive layers that correspond to the preferences of users identified on the basis of the analysis, are implemented in the way of managing interactive video advertising, which includes the following actions:

- in the user’s device, an interactive layer request is generated and sent to the advertising server;

- in the advertising server form a set of interactive layers for subsequent selection;

- in the ad server, based on the historical data of past sessions, determine the expected effectiveness of each interactive layer from the set of interactive layers;

- an interactive layer is selected in the ad server based on the expected performance;

- transmit data of the selected interactive layer from the advertising server to the user's device;

- in the user device, the transmitted data of the interactive layer is received;

- in the user’s device, an interactive layer is shown to the user;

- events related to user actions regarding the displayed interactive layer during the current session are recorded in the user's device;

- transmit event registration data for the current session from the user's device to the ad server;

- in the ad server receive registration data of events of the current session;

- in the advertising server, based on the registration data of the events of the current session, at least one indicator of the effectiveness of the interactive layer of the current session is determined;

- in the ad server, the expected performance of each interactive layer from the set of interactive layers is updated taking into account at least one indicator of the effectiveness of the interactive layer of the current session.

In this case, the user session identifier obtained from the advertising server, generated in the user's device or generated in the intermediate advertising server, can be used to register events of the current session.

The method according to the invention is performed in a system for managing interactive video advertising, comprising:

- a user device configured to request an interactive layer and send it to an advertising server, receive data of an interactive layer from an advertising server, demonstrate an interactive layer to a user, register events related to user actions in relation to this interactive layer, and transmit data of registered events to ad server;

- an advertising server, configured to determine a set of interactive layers for subsequent selection, determine the expected effectiveness of the interactive layers from the set of interactive layers, select an interactive layer based on the expected efficiency, transfer data of the selected interactive layer to the user device, receive data from registered events from the user device, their storage and analysis, determination of at least one indicator of the effectiveness of the interactive layer of the current session and updated the expected effectiveness of each interactive layer from the set of interactive layers, taking into account at least one indicator of the effectiveness of the interactive layer of the current session.

 The system may further comprise an intermediate advertising server in which a user session identifier may be generated, used to record events of the current session. In alternative embodiments, the user session identifier may be generated on the user's device or may come from an ad server.

The invention allows for more effective targeting of video ads based on data obtained from the history of user sessions than other methods of selecting the type of video ads for demonstration to the user.

Brief Description of the Drawings

In FIG. 1 presents a generalized structural and functional diagram of a system for managing interactive video advertising according to the invention.

In FIG. 2 shows a diagram of the interaction of elements of a system for controlling interactive video advertising according to the invention.

In FIG. 3 illustrates an algorithm for an interactive video advertising control method according to the invention.

In FIG. 4A and FIG. 4B presents historical data of the shows in an example of a practical implementation of the invention.

The implementation of the invention

Methods for registering and grouping events

Interactive layers are targeted based on past user experience, namely, based on recorded data about past impressions on the user's terminal devices. The following discusses the key events necessary for further analysis, as well as the parameters of registration requests for these events, which allow grouping events by user sessions.

Key events to be recorded on the terminal devices include showing the interactive layer, click (activation of the interactive layer), the total time of interaction with the interactive layer and the abandoned applications (if there is a feedback form in the interactive layer). The total interaction time can be recorded as an event parameterized by time, or by recording the interaction event at equal time intervals (for example, every second).

In addition to the mentioned key events, for a more detailed analysis, additional events should also be recorded: viewing time (date and time, day of the week, time of day), click time (relative to the start of advertising content or time of display), switching between different information pages in the interactive layer, pressing interface buttons, pressing buttons on input devices (keyboard, mouse, remote control, touching the screen, etc.).

When registering events for subsequent analysis, it is important to have attributes that help group events by user session. Moreover, the presence of appropriate attributes is also necessary in order to ensure targeting by devices and / or by users. Thus, for successful analysis, it is necessary for each event to know exactly the identifiers of the session, terminal device and, if possible, the user.

It should be noted that only one terminal device and one user can correspond to one session identifier. Data about the terminal device and about the user can be part of the session identifier or can be transmitted separately. Given the existing technological scheme for servicing advertising materials, within the framework of the present invention, three methods for transmitting identifiers of user sessions, terminal devices and users are implemented.

The first way is to register events directly from the source code of programmable advertising materials (this is possible, for example, for VPAID and MRAID advertising, if the terminal devices have access to the advertising server for registering events). With this approach, on some devices there may be problems with reading device identifiers using advertising materials, since the transfer of unique device identifiers to third parties may be prohibited by the policies of some equipment manufacturers (such as Samsung SmartTV). In this case, instead of reading the device identifier, it is possible to generate it and write it to the browser cache used to display HTML materials, or to local storage (on SmartTV, this approach can give an acceptable result). Saving session and user identifiers is also possible there. Despite the obvious drawbacks of this approach, it does not require modifications to the video player or intermediate servers on the side of the site where the advertisement is placed (for example, on the servers of an online movie theater).

The second way is to register events through a video player. A video player of an application (for example, an online movie theater) usually has access to unique identifiers of devices and users, but does not control other parameters of events. However, the video player can automatically add these attributes, for example, in the form of HTTP request headers or query string parameters, when registering any events. As already noted, the transfer of unique device identifiers to third parties may be prohibited by the policies of some vendors. However, the video player can use hashing algorithms to hide the true identifier without affecting the further analysis needed for targeting. Adding attributes through a video player allows you to collect more accurate data, however, usually this method requires improvements on the side of advertising platforms (online cinemas, telecom operators, etc.).

The third method is to register events through an intermediate advertising server of the site. Not all advertising platforms have their own advertising server, but if it is present, usually the site has the opportunity to supplement the event registration requests with the necessary identifiers, similar to the method described for the video player. The advantage of this approach is that revision is required only for the server, which allows to reduce development time, despite an extensive fleet of terminal devices.

In all three methods, the user session identifier can be obtained either from the advertising server (during the initial issuance of advertising), or generated on the client side (using advertising materials or a video player in the terminal device or on an intermediate advertising server).

User Session Analysis

The history of events associated with one session, device or user can be used for further analysis. Grouped data allows you to better understand the behavior of individual users, identify abnormal sessions, fraud, etc.

In addition, when registering a sequence of events on an advertising server, analysis and full or partial reproduction of user sessions also become possible.

To target interactive layers, it is necessary to determine the success of a particular type of interactive content both within individual sessions and for a number of sessions of one device and / or user. The following describes general approaches to the interpretation of basic and additional indicators that can be combined to obtain a heuristic function for evaluating the success of a particular type of creative.

An event in one session does not in itself provide the required information. However, with respect to the device and the user, the total number of impressions makes it possible to understand whether enough data has been collected to draw any conclusions about the preferences of this user on this device or this user without being tied to the device.

The click event (activation of the interactive layer) in most cases indicates that the user is interested in the content. In one session, this indicates relative success, however, in isolation from other data, it does not make it possible to determine what exactly influenced this success. When analyzing a number of sessions, the ratio of the number of clicks to the number of impressions (for each type of interactive content and each type of advertising material) is the main factor in determining the types of interactive content (in particular, types and forms of dice) for future demonstration to the user.

Depth of view, otherwise called screening (i.e. the fraction of the length of the video viewed by the user), allows you to understand how long the user had the opportunity to pay attention to the plate and the call to action. For example, if a user turned off the terminal device at the very beginning of the video, this display will not have a significant effect when analyzing this user's preferences.

The total time the user interacts with the interactive layer determines the user's interest in the contents of the interactive layer. This indicator can be used primarily to divide sessions into several groups: sessions with a short interaction time (the plate “worked”, but the further content turned out to be uninteresting), sessions with an average interaction time (the user became interested in the content) and sessions with a long time interactions (the user probably was not involved, for example, moved away from the device).

The number of transitions between information pages and other elements in the interactive layer allows us to analyze the user’s path and determine whether it coincides with the expected path and whether the user has experienced navigation problems. For complex interactive layers, this helps optimize the interactive layer in the future. However, the inventors do not believe that this information is critical for individual targeting; rather, it relates to assessing the overall effectiveness of the interactive layer.

The number of transitions to special pages, such as the contact input page for feedback (for example, phone, email address) or the final page indicating the application has been sent, also rather helps to understand the overall effectiveness of the interactive layer than to determine the preferences of a particular user regarding the types of interactive layers .

To determine the effectiveness of one session, the following formula is proposed that takes into account the above indicators:

равно 1, если была активация интерактивного слоя, и равно 0 в противном случае;where is the meaning

equal to 1 if the interactive layer was activated, and equal to 0 otherwise;

– функция с областью значений от 0 до 1, определяющая влияние досмотра на релевантность оценки эффективности для анализа; например,

может быть использована при линейной зависимости с отсечением по максимальному времени

;

- a function with a range of values from 0 to 1, which determines the impact of the inspection on the relevance of the effectiveness assessment for analysis; eg,

can be used for linear dependence with cut-off by maximum time

;

– функция оценки вовлеченности пользователя, основанная на остальных показателях взаимодействия пользователя с интерактивным слоем –

; в самом простом варианте можно считать

.

- a function of assessing user engagement based on other indicators of user interaction with the interactive layer -

; in the simplest case, we can assume

.

находится в пределах от 0 до 1.It should be noted that the value of the effectiveness of the session

ranges from 0 to 1.

Having determined the effectiveness of each individual session, it becomes possible to calculate the effectiveness of each type of interactive layer for a number of sessions of one device or user using the following formula:

,

,

– последовательность сессий (для одного вида интерактивного слоя, одного устройства или пользователя);Where

- sequence of sessions (for one type of interactive layer, one device or user);

– эффективность отдельной сессии s;

- the effectiveness of a single session s;

– время сессии, относительно последней сессии пользователя или текущего времени;

- session time, relative to the last user session or current time;

– вес сессии, основываясь на относительном времени

; например, можно использовать функцию

, где

– количество сессий, а

– константа.

- session weight based on relative time

; for example, you can use the function

where

- the number of sessions, and

Is a constant.

также находится в пределах от 0 до 1.It should be noted that the value

also ranges from 0 to 1.

Interactive Layer Selection Method

Based on the efficiency values for each type of interactive layer, it is possible to determine the most effective types by sorting them according to the calculated efficiency indicator.

For large impressions, storing all metrics for all sessions can be difficult and inappropriate. But even with the technical ability to store the entire history, calculating the effectiveness of the types of interactive layers for each device or user in each session can affect the throughput of advertising servers.

при расчете эффективности того или иного вида интерактивного слоя для конкретного пользователя или устройства дает возможность избегать ресурсоемкого пересчета эффективности при добавлении информации о новой сессии. Это позволяет быстро обновлять показатели без повышения вычислительной нагрузки и без необходимости в хранении всех показателей для полной истории взаимодействия с устройством или пользователем.Optimal weight function selection

when calculating the effectiveness of a particular type of interactive layer for a specific user or device, it makes it possible to avoid resource-intensive recalculation of efficiency when adding information about a new session. This allows you to quickly update indicators without increasing the computational load and without the need to store all indicators for a complete history of interaction with the device or user.

и добавлении сессии с эффективностью

и временем

общая эффективность пересчитывается по следующей формуле:For example, when choosing

and adding sessions with efficiency

and time

total efficiency is recalculated according to the following formula:

.

.

The value of the performance indicator of the type of interactive layer, close to 1, most likely indicates that this type of interactive layer is well suited for this device or user. However, it is worth noting that a value close to 0 does not give confidence that this type of interactive layer does not work at all. Indeed, if the user has never met this view or met it for a very long time, the indicator can be 0. Therefore, it is important to take into account the total number of sessions, as well as the age of the last session with an interactive layer of this type.

The interactive layer targeting mechanism uses performance indicators to decide which interactive layer to show the user (if there is appropriate variation in the advertising materials).

The simplest method of targeting is to choose the type of interactive layer with the highest performance indicator. If there are several types with equal indicators, the type is selected randomly (for example, with a uniform probability distribution). It should be noted that with this approach, some types of creatives can never be selected, since each type of interactive layer without sessions has an efficiency of 0.

In order to avoid the situation with unfair distribution of types of interactive layers, it is proposed that each of the available types be assigned the probability with which it will be selected, based on performance indicators:

,

,

– неотрицательный коэффициент, обеспечивающий минимальную вероятность для видов креативов без истории;Where

- non-negative coefficient providing a minimum probability for types of creatives without history;

– время самого недавнего показа для вида

;

Is the most recent show time for the view

;

– дополнительный вес, определяющий «устаревание» показателя эффективности;

- additional weight, which determines the "obsolescence" of the performance indicator;

– количество доступных видов для выбора.

- the number of available species to choose from.

Thus, the species with the highest performance indicator will have a greater probability of choice, but species with an unknown or low indicator will also have a chance to “rehabilitate”.

In FIG. 1 presents a generalized structural and functional diagram of a system for managing interactive video advertising according to the invention. The system comprises an advertising server, optionally, an intermediate advertising server and one or more user terminal devices.

In FIG. 2 shows a diagram of the interaction of elements of a system for controlling interactive video advertising according to the invention. The VAST designation here refers to the aggregate data associated with video advertising in accordance with the VAST standard. Downstream arrows indicate the flow of data related to video ad management. Downstream arrows indicate the flow of data related to the recording of events related to video ad consumption. The dashed line indicates the data stream of the video ad itself.

In FIG. 3 illustrates an algorithm for an interactive video advertising control method according to the invention. Moreover, the registration and grouping of events, analysis of user sessions and the choice of an interactive layer are performed in accordance with the principles of the present invention described above.

An example of a practical implementation of the invention described below is based on the use of data related to one user. It should be obvious that this is only to simplify the understanding of the invention, and data relating to many users can be used in the invention.

Over the course of approximately one month, several dozen displays of interactive advertising were carried out on the user's terminal device. These impressions applied to three ad campaigns:

- advertising campaign No. 1 (RK 1) contained one type of creative - a stop banner with the call "Learn more";

- advertising campaign No. 2 (RK 2) contained two types of creatives (information button-plates with calls “Get instructions” and “Order”), which were selected for demonstration to the user at random, without using the targeting mechanism;

- advertising campaign No. 3 (RK 3) contained three types of creatives (stop-banner “Find out more”, a button-button “Learn more” and a button-button “Order”), the choice of which for demonstration to the user was carried out in accordance with this invention.

It should be noted that on a relatively small data set it is difficult to judge the effectiveness of the method, since for a given user, the click motivation can be any and may not depend on the type of plate. For a more accurate assessment of the method, it is necessary to conduct full-fledged testing for a sufficiently large audience, dividing it into groups, in one of which random targeting can be performed, and in the other - targeting based on the described method. To exclude other factors, the test must be carried out as part of one advertising campaign and on one advertising platform.

However, in practice, it is important that the method is workable for the average user, therefore, considering the history of the selected user representative, you can make certain assumptions about the effectiveness of the method.

In FIG. 4A and FIG. Figure 4B presents a table containing the history of impressions on the user's terminal device, including information on the date, type of advertising material, the presence of a click, and the estimated values of weights (probabilities) for advertising targeting. From the table you can see how successful impressions (i.e. impressions that led to a click) affect the distribution of weights, while types of creatives without clicks have non-zero weight, and the weights of successful sessions gradually decrease over time (every week).

Thus, this invention allows for more effective targeting of video ads based on data obtained from the history of user sessions than other methods of selecting the type of video ads for display to the user. In particular, this invention provides a higher conversion of video advertising than a random choice of the type of video advertising or the absence of a choice (i.e., demonstration of video advertising of only one type).

It should be noted that in the above description reflects only those actions that are most essential to achieve the objective of the invention. The specialist understands that for the functioning of the described system it is necessary to perform other necessary actions, for example, connecting the equipment, initializing it, launching the corresponding software, transmitting and receiving commands and confirmations, exchanging service data, synchronization, etc., the description of which is omitted for brevity of presentation.

The devices and their parts mentioned in the description, drawings and claims are hardware and software, while the hardware of some devices may differ, partially coincide or completely coincide with the hardware of other devices, unless otherwise indicated explicitly. The hardware of the devices may be located in different parts of other devices, unless otherwise specified explicitly. Software parts (modules) can be implemented in the form of program code contained in a storage device. Data processing may be concentrated in one computing facility, or it may be performed in a distributed manner in multiple computing facilities.

The sequence of actions in the description of the method is illustrative and in various embodiments of the invention, this sequence may differ from that described, provided that the function performed and the result achieved are maintained.

The features of this invention can be combined in various embodiments of the invention, if they do not contradict each other. The embodiments described above are provided for illustrative purposes only and are not intended to limit the scope of the present invention as defined by the claims. All reasonable modifications, upgrades and equivalent replacements in the composition, construction and principle of operation of the present invention, made within its essence, are included in the scope of this invention.

This invention is described in the context of its use in interactive television. However, to a certain extent, it can be used in other areas involving the activity of a user of video advertising, in particular in online advertising.

Claims (22)

1. A method of managing interactive video advertising, which includes the following actions: the user’s device generates an interactive layer request and sends it to the advertising server; in the ad server form a set of interactive layers for subsequent selection; in the ad server, based on the historical data of past sessions, the expected performance of each interactive layer from the set of interactive layers is determined; in the ad server, an interactive layer is selected based on the expected performance; transmit data of the selected interactive layer from the advertising server to the user's device; the user device receives the transmitted data of the interactive layer; in the user device, an interactive layer is shown to the user; events associated with user actions regarding the displayed interactive layer during the current session are recorded in the user’s device; transmit event registration data for the current session from the user's device to the ad server; the ad server receives the registration data of events of the current session; in the advertising server, based on the registration data of events of the current session, at least one indicator of the effectiveness of the interactive layer of the current session is determined; the advertising server updates the expected performance of each interactive layer from the set of interactive layers, taking into account at least one indicator of the effectiveness of the interactive layer of the current session. 2. The method of claim 1, wherein the user session identifier obtained from the advertising server is used to register events of the current session. 3. The method according to claim 1, in which the user session identifier generated in the user device is used to register events of the current session. 4. The method of claim 1, wherein the user session identifier generated in the intermediate advertising server is used to register events of the current session. 5. A system for managing interactive video advertising, comprising: a user device configured to request an interactive layer and send it to an advertising server, receive data of an interactive layer from an advertising server, demonstrate an interactive layer to a user, register events related to user actions in relation to this interactive layer, and transfer recorded event data to an advertising server; an advertising server configured to determine the set of interactive layers for subsequent selection, determine the expected effectiveness of the interactive layers from the set of interactive layers, select an interactive layer based on the expected effectiveness, transfer data of the selected interactive layer to the user device, receive data from registered events from the user device, their saving and analyzing, determining at least one indicator of the effectiveness of the interactive layer of the current session and updating I am the expected performance of each interactive layer from the set of interactive layers, taking into account at least one measure of the effectiveness of the interactive layer of the current session. 6. The system according to claim 5, in which the user session identifier received from the advertising server is used to register events of the current session. 7. The system according to claim 5, in which the user session identifier generated in the user's device is used to register the events of the current session. 8. The system of claim 5, further comprising an intermediate advertising server, in which the user session identifier generated in the intermediate advertising server is used to register the events of the current session.

Images