This series introduces trendy and popular themes from among those covered in Fujitsu Research Institute's information magazine Chisounomori by interviewing consultants who are working on actual business projects. The third theme is "Digital Marketing."
Market Trends in Digital Marketing
― Digital marketing has become a popular topic, but few customers seem to be seriously working on it. What are the market trends?
Internet business accounts for 10% of the retail industry's total sales, while the remaining 90% has not yet been digitalized. However, clearly sponsors of terrestrial commercial broadcasters have been shifting from traditional advertising to digital advertising. For example, CyberAgent started an Internet TV station.
Even if the same content is delivered, terrestrial broadcasting can only give the audience rate, while on the Internet we can identify all user behaviors. Therefore, advertising budgets currently spent mainly on TV may shift to Internet advertisements. In Japan, annual TV shipments total 10 million units, while those of PCs and mobile phones number 50 million. This shows that the Internet is overwhelmingly advantageous in terms of providing more touch points with customers. If the Internet's value as an advertising medium continues to increase, the digital shift will accelerate whether we like it or not.
― Ms. Ando, you are in a position to support customers on-site. What do you think?
Ando: When I see the increase in businesses that do not have brick-and-mortar stores, such as Internet banks and MVNO (Mobile Virtual Network Operator), I feel that the digital shift is progressing. As the Internet has become a major battlefield, we have no choice but to acquire and nurture customers mainly through the Internet. Actually, the number of companies that have established a data management platform (DMP) * has been increasing, and they are struggling with how to optimize their DMPs. So, I believe customers need ICT services to solve this optimization problem.
Takagi: As for the positioning of the Internet, the role of conventional media is to let customers flow in through advertisements while Internet media takes charge of the processes thereafter. The current focus of the Internet is how to optimize and increase the targeting accuracy after customers flow in. However, when the Internet takes over TV's role, I think Internet media will get a bigger piece of the pie, not just the small piece of pursuing targeting accuracy. Also, in the process of targeting individual customers, customers will be optimally classified into data-driven ** segments. It is only a matter of time before part of current marketing operations that rely on people are replaced by machinery. Therefore, I think companies that lag behind with respect to such trends will also fall behind in terms of arming themselves with IT.
A Mechanism to Develop Fans by Bringing Users Together
― Recently, companies are creating fan sites to obtain customers' trust. In marketing terminology, this is described as a communication strategy or an increase in the number of touch points. Please tell us the story of the MVNO you work for.
Ando: In the case of MVNO, it is difficult to get users who already have a smartphone to sign another contract for a new line. However, if nothing is done, they may move to other companies. Company A directs new users to its community site, which serves as a touch point. Existing users provide support on how to use the products and their advantages, and users can also exchange ideas with each other on the site. Low costs are achieved by having users support each other, which also attracts new users. At the same time, the users' sense of contributing to Company A increases their loyalty to Company A and prevents estrangement. However, as there is no concept of conversion *** for community sites, I think that going forward we need to deliver different messages catered to each individual user to make them into fans.
Takagi: Though viral marketing did not become popular ****, we can see a variety of things mathematically--for example, to whom and at what timing we should send information. Finding the optimal timing and spreading information at that timing is called influence maximization. To that end, users must be brought together. Big companies always have fan sites and user groups where users have ownership, and a mechanism such as influence maximization harnesses the power of data and computers, not manpower. I think this is an important technique.
― US major league baseball teams rent out stadiums and hold events for fans to talk about their teams. If Ichiro suddenly appears there, tweets rapidly spread in a chain reaction. Those who read the tweets can get information that they cannot obtain from the mass media, such as what Ichiro thinks. In this way, what had previously seemed distant becomes familiar, and fans start to think that they want to come see games more often. As a result, the profits of US major league baseball teams have tripled over the past twenty years, while those of Japanese baseball teams remain flat.
Takagi: That is the epitome of viral marketing, isn't it? If a player plays a game alone, that player can only enjoy the game in a limited amount. However, when multiple players compete against each other via the Internet, the world of play grows of its own accord, and a far greater number of users are generated than what happened at the beginning. In this way, I think it is essential to bring many users together.
AI Application Areas in Communication Strategy
― I feel that AI is likely to enter the world of bringing users together.
Takagi: On Chiebukuro (a Q&A site) and similar websites, those who ask a question cannot choose who they want to answer the question. However, it should be possible to suggest appropriate respondents by matching the question's content and the questioner's historical data using AI. As the relationships between social network users can be clearly displayed in graph form, we can generate a variety of ideas by bringing users together. For example, if users use Twitter not only to chat but also to talk about more weighty topics, they may perceive more value in tweeting.
― What do you think about the gap between social networks such as Twitter and machine learning ***** ?
Takagi: Using social data for marketing will become indispensable in the future; however, what to extract varies significantly. For example, at present, people determine the keywords they want to make hit based on an expected rate calculated only from numerical values. To optimize targeting by capturing human behavioral intentions, more sophisticated AI is required, not just simple machine learning for classification.
Linkage between Social Networks and DMPs
― To ascertain behavioral intentions, a proposal to understand consumers can be made by combining Do-Cube® ******, Fujitsu Research Institute's customer behavior analysis service, with a DMP as a set.
Ando: Companies want to extract only the background and situational information, such as the reasons for purchasing products and behavior after purchasing, from social networks, because such information can be used directly to understand products' points of appeal and to set up advertisements and landing pages. Do-Cube is good at extracting the reasons for purchasing products as well as how they are used from social networks. The problem is that Do-Cube has not been linked with DMPs.
Takagi: I think such linkage is possible and should be automated. Using recent computational AI, it is possible to react with mechanical responses: for example, "If this word or intention is detected, handle it this way." Although it is troublesome to extract important data from social networks, once this has been done, the subsequent processes are easy.
― That is what Do-Cube is good at, isn't it?
Takagi: It is hard to extract (important information) from social networks. However, it is easy to replace what humans do intellectually with calculation. I think the final form is to mechanize what is beyond that.
― A DMP can be configured with a data-driven approach; however, I thought the world of social networks requires human interpretation. If social networks can be digitized and connected (with a DMP), we can do many things.
Takagi: For example, in the AI Project "Can a Robot Get into the University of Tokyo?" the robot answered world history questions and completed the writing tasks. Suppose a difficult question that starts with an explanation and then, based on that, asks "What do you think?" The computer interprets the assumption part, searches for related articles, and then integrates and organizes them into an answer. The computer recognizes the preamble and the main subject, and then finds out what is being asked through iterative processes. For example, in a social network, it is sometimes the case that what one wants to say appears in the latter half of a sentence, such as "Although A was not good enough, B was." In this case, if the sentence including the former half is simply averaged, the preamble occupies a larger part. However, computer-based interpretation of social networks is getting closer to human sensibility. Now, machines can perform a considerable part of the work, although the final part is left up to humans.
― In digital marketing, the accuracy of approaches to individuals will rapidly improve in the future. On the other hand, if consumers feel that they are always being monitored, the effect will be just the opposite. What is essential is to adopt a viewpoint of customer friendliness. For example, when summer vacation comes, recommended books are arranged on bookstore shelves as "recommended books," but few bookstores can explain why those books are excellent and recommended. Customers will put their trust in bookstores that give them correct information and enrich their lives by making them aware of what they did not know. In this way, we would like to promote our business from a consumer-friendly point of view. Thank you very much for joining us today.
*: DMP (Data Management Platform)： A platform to optimize action plans for advertising delivery through centralized management and analysis of big data accumulated on Internet servers and in a company's website log data.
**: Data-driven: To take the next action based on data obtained from effect measurements, etc.
***: Conversion: The final result obtained on a website. Actions desired by the business owner taken by visitors to an advertising website or other company website, such as registering for a membership, requesting information materials, or purchasing a product.
****: Viral marketing: An Internet-based promotion method that encourages users of a product or service to introduce it to their friends and colleagues.
*****: Machine learning: An AI research theme that is a technique or method to recreate humans' natural learning abilities by means of computers.
******: Do-Cube: A blog search service for quickly and easily collecting consumer voices from sampled blogs representative of overall consumer opinions.
Chisounomori (Focus Series) Digital MarketingDigital Marketing in the Future
The details of this article are available in PDF format.