Artificial Intelligence That Evolves Marketing

July 20, 2016

Artificial Intelligence That Evolves Marketing

Tomohiro Takagi
Professor,
School of Science and Technology, Meiji University 
Do you remember the term “big data”? At the beginning of this decade, the media often used the term, yet we seldom hear it these days. In reality, big data has been growing significantly and is used in various fields together with the utilization of numerous web services. Here, machine learning is used as a method and is a branch of artificial intelligence (AI). However, this is only a small part of the potential of AI.
AI Is Expected to Further Streamline Marketing

These days, marketing activities have been data-driven and are changing drastically. When we hear the term “marketing,” many of us think of sales marketing. However, marketing means broader corporate activities that include creating products and services that customers are truly looking for and delivering information so that customers can understand the value of the products and services, and obtain them. For these activities, the marketers make graphs by using statistical analysis tools to analyze the gathered data. Data is shown graphically because the data needs to be visualized and shared with the related parties as they are the ones who devise marketing strategies. Simply put, machines work on the basic part of data processing and humans work on the creative aspects.

When there was comparably less data volume to collect, the target of the information delivery was the mass market, and the advertising media were mainly television, newspaper, and magazines, this structure was fine. However, with the passage of time more effective marketing has become necessary and the target of advertising has become segmented away from mass marketing to individuals as it is more effective to market to a narrow target and deliver information that is tailored to individuals. For example, when someone is looking for something on a website, advertisements of products that interest that person are displayed. This is because it is more effective sales-wise to show items that appeal to individuals rather than showing the same items to everyone.

This marketing that delivers information to individuals is attracting attention because it was difficult to do this using conventional advertising media, such as television and newspapers, and web technology has been developed to realize these marketing methods. Many customers leave their “footprints” on the web and that large amount of data is leveraged for business. Some examples include online advertisements that are displayed in accordance with recommendations and web browsing in electronic commerce. Massive amounts of data are processed in real time and highly optimized, but there is still a ways to go until the data is intellectualized.

Increase in the Application of AI
That said, marketing utilizing AI is only a part of the area of promotion when seen from the entire marketing process. Most other marketing areas still depend on time-consuming conventional methods, such as questionnaires and aggregation, and are not utilizing real-time data-driven marketing methods. A huge amount of labor and time is required to process data conventionally and show it graphically, and then devise marketing for each target based on the data. The data of course cannot be processed in real time. Although some marketing uses POS or other online data, data utilization appears to be way behind compared to the real time processing system described above, and automatic intellectual processing is not performed.

Basically, in current-day marketing, data is utilized effectively only partially downstream and humans are still working upstream to create marketing actions through trial and error. I believe a primary role of AI is to provide functions that penetrate the intellectual level.

For example, I am experimenting with replacing intellectual work performed by humans through trial and error with machines in my lab. Many people seem to have seen that advertisements on the screen change depending on the user in real time. My lab, in conjunction with web advertising companies as part of joint research between industry and academia, is researching a system to improve the preciseness of this function and change advertisements to the most optimal expression, even for the same product, depending on the individual. We are trying to make a system that automatically generates advertising phrases depending on the target to deliver optimal advertisements with a high matching accuracy.

Let’s suppose that we can obtain the footprints of one billion people on the Internet as information. My lab is developing a system that automatically calculates where products should be injected for better sales based on this data. These systems enable clarification of which attributes can segment potential customers the most (for example, customers jumping on a specific campaign is more important than gender and age) more than segments vaguely obtained conventionally through intuition like gender and age, and can place products that have greater requirements into optimal segments. Thus, optimal advertisements and promotions can be calculated automatically. Simply put, by analyzing and calculating purchasing behavior automatically, not only advertising but all other marketing activities including the segmentation of consumers, product development, and promotions become linked and optimized for data driven marketing. Furthermore, we are trying these approaches in various other marketing elements. We are replacing the tasks and judgment that marketers perform with their senses and intellectual activities aggressively with machines.

To Make AI Truly Beneficial for Humans

These days, AI is being developed in various fields. It was big news that one of the world’s top Go players lost a match to AI. Some say that robots with AI will revolt and humans will be conquered in the near future. However, as a researcher who is researching AI, I believe that this will not happen as humans are too intelligent for this to happen. Humans process information flexibly, appropriately, and instantly, even in casual daily conversations. Actually, this is a very difficult task, and AI has not achieved this level of ability yet, and in fact it will be very difficult to achieve in the future as well.

The merit of AI is that it efficiently processes tasks that are intelligent yet time consuming for humans. One task is the processing of a large amount of data accurately and calculating optimal results quickly. Our expectation for AI in the marketing field has increased so that we can use big data more effectively. However, some people are afraid that their privacy may be transparent when customized advertisements are delivered based on information that is their footprints. However, people in the industry that handle information, including myself, are very careful about how far we should encroach on individual privacy. Our advertisements will also have no value if people feel scared or upset. In a time when there are so many unnecessary advertisements for uses, we aim to make advertisements that deliver beneficial information to users. Moreover, this idea is not limited to advertisements but can also be used for marketing activities across the board. Ideally, we would like to develop AI that provides pleasing and beneficial marketing for users.


* The contents of articles on M's Opinion are based on the personal ideas and opinions of the author and do not indicate the official opinion of Meiji University.

Profile

Tomohiro Takagi

Professor, School of Science and Technology, Meiji University

Biography:
Tomohiro Takagi is one of the most influential authorities in the world in fuzzy theory—one kind of computational intelligence—and is well acquainted with marketing theory. He is active in the latest research, development, and commercialization of products that relate to the highly precise recommendation engine, and highly precise targeting and digitalization of the entire marketing process while being involved with technical and business strategies, new business planning, development, and actual product development in industries. He has worked on various joint and contract research with major domestic corporations and US oil capital, and has achieved the highest level of achievement in competitive international workshops. Currently a Professor at Meiji University after working as a research fellow at the Department of Electrical Engineering Computer Science, University of California Berkeley, Matsushita Electric Industrial Co., Ltd., and as the Program Officer at the Japan Society for the Promotion of Science.

Research Fields:
Computational artificial intelligence, data driven marketing, and web science

Research Themes:
High level intellectualization of data driven marketing

Degree:
Doctor (Engineering)
 

Page Top

Meiji University

Copyright © 2015 Meiji University. All Rights Reserved.