Effective Data Use Is Key! The Latest Digital Marketing to Meet Customer Needs

The digital transformation in marketing is gaining momentum as use of smartphones, tablets, and other smart devices rapidly increases. As the types of customer contact points diversify, how will new technologies, including AI and quantum computing, change digital marketing? This article introduces the latest digital marketing strategies enabled by Fujitsu technologies.
[Fujitsu Insight 2017 Digital Marketing Keynote Report]

Greater Customer Diversity Increases Digital Marketing's Importance

Iwao Nakayama
Chief Evangelist, Global Marketing Department,
Corporate Executive Officer,
Fujitsu Limited

We use smart devices as everyday tools. Many companies use digital technologies to create customer contact points and offer a wide range of information and services through the same. Today, companies can even personalize information for individual users. Digitization has created bilateral information flows between companies and consumers, enabling mutual use of each other's information.

To be more specific, take the example of purchasing a car. In 2000, consumers visited a car dealer seven times on average before purchase. Now, this figure is only 1.5 times on average. Consumers use smartphones to collect the necessary information and narrow down their choices in advance, so they may visit a dealer only to do a test drive.

In the area of digital transformation, which is accelerating worldwide, many companies promote digitization as a marketing theme. According to a 2016 Gartner survey, 37% of CEOs chose digital marketing as a key investment area for the next five years.

Not Sharing Marketing Data May Increase Man-hours

While digital marketing continues to evolve, marketing techniques find themselves in chaos. Our study revealed that nearly 4,900 companies worldwide offer digital marketing tools, and over 5,000 kinds of related solutions are sold. Meanwhile, many companies have digital marketing tools but do not know the combination that best solves their work problems.

Offices confront a heap of problems. One such problem is not sharing marketing data. In many companies, marketing data is scattered over multiple divisions, such as the sales division, the marketing division, the contact center, and the e-commerce division. Even though digital tools have been introduced, more man-hours may become necessary if each tool has a different data input procedure.

In addition, employees are completely exhausted by top management demanding data-driven marketing, which involves integrating various types of data. In divisions using multiple tools, much of the workforce takes time to integrate data from various tools. As a result, many companies face the digitization dilemma that they are not able to run the PDCA cycle, an essential element of marketing measures.

Case Studies on Successful AI Use

Given such circumstances, how should we use various types of data in marketing? At Fujitsu, we use the FUJITSU Human Centric AI called Zinrai, a systemized AI service, in our digital marketing.

Case Study 1: Targeted direct email doubled the number of customers (in-house implementation)

In email marketing, conventionally we first defined the target segment based on user attributes and then sent direct emails to that segment. Refining the segment definition required many staff members to perform cumbersome manual tasks, such as establishing and testing hypotheses.

We used our Zinrai AI to identify target users more accurately and then send direct emails to them. More specifically, Zinrai learned the relationship between attribute data and website access data (cookies) to autonomously identify the target segment with high accuracy. As a result, identification accuracy improved and the number of customers doubled while marketing costs remained the same.

Case Study 2: Quadrupling the membership signup rate by visitors to a makeup information website

Many makeup information websites use member information to recommend products that match visitors' attributes. These websites are often designed with a focus on cosmetic items; as a result, they fail to effectively establish product-article connections. Thus, viewers do not regularly visit such websites.

To address this issue, our client created a mechanism to recommend the best makeup items and styles to users based on face photos that they uploaded. Employing deep learning technology, Zinrai learned 50,000 pieces of face image data and the makeup procedures best suited to each facial shape. It also learned and generated knowledge on about 800,000 makeup simulation results. After updating the website with this recommendation feature, the membership signup rate for visitors quadrupled, and the rate of website re-visits doubled.

Recommending the best makeup items and styles using deep learning technology

Case Study 3: Purchasing behavior prediction for a shopping mall

For retailers such as shopping malls and convenience stores, predicting customer purchase behavior is an important factor to increase profits. We use compact sensors to detect what customers see and then have Zinrai analyze which products attract customer attention as well as which ones they pick up. This analyzed data can help increase sales by encouraging retailers to rearrange shelf displays, adjust product prices, and feature selected products as recommended items.

Zinrai also harnesses one of its strengths, an image processing technique, to understand how crowded a shopping mall is. Taking advantage of this capability, retailers can send out information matching shopper attributes via digital signage and apps at the correct time. Zinrai can also improve store-specific supply/demand prediction by referencing additional data, such as the weather and events in surrounding areas.

Deep Tensor: Fujitsu's Deep Learning Technology That Covers Social Networks and Business Relationships

Deep learning can make use of text, numeric, voice, image, and video data. In addition to these types, Fujitsu's unique deep learning technology, Deep Tensor, understands whole graph data to determine the optimal answer.

Graph data refers to data connected to form real-world structures, such as relationships in communications, business, and social networks (personal relationships) as well as structures like those of chemical compounds (element bindings). Deep Tensor converts and expresses such graph data in the form of a tensor so that AI can learn it. Possible future applications include detection of illegal financial transactions, traffic congestion easing, improved transportation efficiency, and shortening of new medicine development times.

Deep Tensor

Case Studies: Digital Annealer Solves Digital Marketing Problems

It is expected that in 20 to 30 years, the singularity will arrive--the point in time when AI reaches and then exceeds human intelligence, and we will witness a dramatic evolution of computers. It is predicted that in the near future, quantum computers will emerge and change our daily lives and business dramatically. However, quantum computers are still in the research stage, and we have a long way to go before practical use begins.

To create higher quality lives and society, we at Fujitsu have developed an innovative, futuristic computer called Digital Annealer using quantum computing technology. With the mission of solving various types of social issues, Digital Annealer has been designed to specialize in solving combinatorial optimization problems, which require finding the optimal combination from among a vast number of possibilities.

The world of digital marketing has many combinatorial optimization problems. Here are some examples of what Digital Annealer can do.

A car manufacturer's website

With Digital Annealer, a website can autonomously personalize information for each visitor. By assigning attribute data (e.g., age and gender) to each website part, the optimal content is autonomously generated and displayed. If a website consists of 6 parts, such parts can be arranged in 720 ways. If there are 20 parts, there are approximately 2.43 quintillion ways. Digital Annealer facilitates instant website display with the optimal content.

Flexible trip recommendations

Digital Annealer can recommend a place to visit now or in the future to individual customers and then instantaneously prepare a corresponding travel itinerary. Moreover, the ultimate trip personalization system can be created by detecting information such as train accidents, news, and weather forecasts at destinations on a real-time basis, flexibly adjusting the travel itinerary, and then recommending the completed itinerary to the user.

Flexible trip recommendations

As described above, Fujitsu has a rich lineup of unique technologies to sophisticate your digital marketing strategies. We hope to create a fun, interesting world of digital marketing with you as a step toward the singularity era. We look forward to hearing from you!

  • Iwao Nakayama Chief Evangelist, Global Marketing Department,
    Corporate Executive Officer,
    Fujitsu Limited