Mastering AI in the DX Era: Trends and Possibilities in Marketing AI Utilization

Following the wave of the "Third AI Boom," it has been five to six years since the adoption of AI (artificial intelligence) emerged as a business trend. As of 2022, the so-called "AI boom" as a buzzword appears to have subsided, yet AI is once again gaining prominence as a key technology within DX (Digital Transformation). This article explores the latest trends, possibilities, and applications of AI, primarily within the realms of marketing and customer experience.

What Happened to the "AI Boom"?

Before looking at marketing, let's grasp the trends in "AI utilization" itself over the past few years.

The peak of interest in AI adoption was around 2016-2017. Many will recall the news of DeepMind's AlphaGo defeating the world Go champion. At that time, numerous AI startups emerged, collaborating with large corporations on proof-of-concept experiments, dominating news media coverage daily.

However, around 2019, "AI skepticism" and "AI implementation failures" began to surface. Information sharing from a more realistic perspective increased, highlighting issues like "AI not delivering cost-effectiveness" and "pitfalls of AI adoption."

Just when it seemed the AI boom might be cooling down in 2020, the global spread of COVID-19 and the simultaneous surge in DX (digital transformation) brought AI back into the spotlight. This was because AI proved useful in various pandemic scenarios: replacing physical operations like handwriting and visual inspection, and contributing to infection control through contactless communication like avatars and chatbots. Advanced behavioral data analysis and personalized communication, discussed later, are also areas where AI excels, and its applications have broadened within the DX boom.

While this trend appears to continue as of 2022, one key difference stands out compared to the boom around 2016-2017: AI is now increasingly viewed not as a "magic wand" capable of anything, but simply as one tool that enhances business value when used appropriately.

Unlike a few years ago when the phrase "AI-powered XX" carried value—for better or worse—simply using AI no longer inherently elevates the value or expectations of an initiative. We've entered an era where the value of AI adoption is measured based on various business factors, including cost-effectiveness and development/operational frameworks.

Overview of the Current State of AI Utilization in Marketing

With this backdrop in mind, let's examine the current state of AI adoption in marketing.

Here, we break down the marketing process into the following major steps:

1. Market/User Understanding
2. Strategy Formulation & Targeting
3. Production/Development
4. Delivery (Customer Touchpoints )

We will introduce representative AI application approaches for each of these steps.

1. Market and User Understanding

AI is fundamentally a "machine built on large amounts of training data to classify or predict data." By classifying and predicting various consumer and market data, or gathering user feedback through novel approaches, previously unseen consumer insights surface.

For example, using "natural language processing" technology to analyze the language humans use daily makes it possible to identify trends and make predictions from vast amounts of text data (such as social media posts and survey results) that were previously impossible to analyze.

Case Study
・TREND SENSOR
・mindlook
・TexAIntelligence

AI is also actively utilized in the field of "demand forecasting," which predicts market trends and the sales performance of a company's products. Demand forecasting is applicable not only to marketing but also to various fields such as production planning and logistics, making it a major theme in DX.

Examples
・Michishiro℠
・AITC (AI Transformation Center) AI Demand Forecasting Project

In a slightly different approach, solutions are emerging that utilize conversational agents (chatbots) to enable "interviewing a large number of users and quantitative analysis" – tasks that were difficult with traditional interview surveys. Technology is changing the very nature of user research.

Examples
・Smart Interviewer

2. Strategy Development & Targeting

Strategy development—determining "who to target and what message to deliver"—is the core of marketing. Here too, AI proves its worth.

The most common approaches involve user data analysis and segmentation. Increasingly, companies are implementing platforms like CDPs (Customer Data Platforms) or private DMPs (Data Management Platforms) to collect and manage data, thereby retaining their 1st Party Data. By training machine learning models on this data, they can predict users with "higher purchase likelihood" or "higher churn rates," enabling the development of communications tailored to specific target attributes.

Even beyond metrics directly linked to KPIs like purchase likelihood or churn rate, more companies are building AI to predict clusters like "users who like music," "users with children," or "users highly interested in business," using these insights to shape communication approaches.

Such data-driven customer prediction and classification, previously only possible through rule-based methods (e.g., classifying users who visited a site x times as interested), can now achieve high-precision classification by applying machine learning technology.

3. Production & Development

One pitfall in digital marketing is that the more detailed the strategy design, the more massive the production costs become, eventually reaching a limit on the budget and resources that can be allocated. As mentioned above, the more various data and AI are leveraged to refine targeting, the more patterns of creative assets and messages increase. Furthermore, considering the increasing complexity of customer journeys and the diversification of touchpoints (social media, web, video, TV, print media...), the required creative patterns explode exponentially.

AI-powered creative automation is the key to overcoming this challenge. Tools for automatically generating various creatives—copywriting, banner ads, landing pages—are emerging, enabling the creation and deployment of numerous creative patterns with fewer resources than ever before. Solutions are also emerging that focus on replacing labor-intensive tasks like resizing, rather than generating everything automatically.

Case Study
・ADVANCED CREATIVE MAKER

Additionally, AI exists to predict the effectiveness of generated creatives, enabling not only mass production but also the ability to filter out the most effective ones.

Example
・MONALISA

Here, the following two points are crucial.

The first point is that not everything can be fully automated by AI. At least as of 2022, automatically generated creative is not perfect. There are no AI systems that can generate content with 100% human-level quality—there may be awkward phrasing or layout flaws. Therefore, it's necessary to identify which parts of the production and execution process can be automated and which require human involvement, starting with partial implementation.

The second point is optimizing the production process itself. To fully leverage these tools, you must reevaluate your existing production workflow. For example, instead of outsourcing everything to an agency, handling some tool operations or final adjustments in-house can dramatically improve efficiency. Alternatively, encouraging creative staff to use the tools and building a production process that accumulates usage data allows that data to be collected, leading to further tool improvements. By evolving these processes behind AI adoption, the value of AI itself increases.

Case Study
・CXAI

4. Delivery (Customer Touchpoints)

AI can also be leveraged at the contact points where messages tailored to specific targets and created creative assets are delivered to end users. This can enrich the experience and deliver messages more efficiently.

For example, tools exist that change the message displayed on a website (landing page) based on data-driven customer clusters (groups classified by similar customer characteristics).

Case Study
・Microscope

AI is also being used to enhance advertising effectiveness in mass media, where detailed segmentation was traditionally difficult. In the television industry, solutions that predict viewership ratings for various user segments and serve tailored ad creatives to optimize cost-effectiveness are rapidly expanding.

Example
・SHAREST
・RICH FLOW

The Future of AI in Marketing

As we've seen, AI is being utilized across various areas of the marketing process. At the same time, AI technology itself is evolving daily. What changes might technological progress bring in the future?

I'd like to briefly predict future trends from the following four perspectives.

1. Advancements in Generative AI
2. The Emergence of Virtual Talents
3. Advances in Optimization
4. Ethical Considerations

1. Advances in Generative AI

In the previous chapter, we noted that AI-driven creative generation has not yet achieved sufficient accuracy. However, cutting-edge algorithms are rapidly closing this gap. Below is an example of AI-generated copy verified by the Dentsu Group, producing remarkably natural text. If this trend continues, full automation will eventually become feasible.

<Example of AI-Written Text>

○ Baumkuchen Slogan
・Our thin-layered Baumkuchen is fluffy, moist, and gently sweet.
・Made with 100% natural ingredients, 100% handmade, 100% fluffy and moist, using Japan's finest Baumkuchen techniques.

〇Café Latte Slogan
・An exquisite café latte made with premium milk. Its rich depth and sweet-tart flavor are irresistible.
・Made with delicious coffee beans, it's a tasty café latte where you can enjoy the aroma too.

〇 Lemon Tea Slogan
・A tea-time treat-style lemon tea featuring a subtly sweet lemon flavor. Its richly aromatic blend and pleasant sweetness enhance your relaxation time.
・A mixed tea characterized by carefully simmered tea leaves and the sweet-tart flavor of citrus.

*Brand names have been removed from the generated catchphrases.

Furthermore, the scope of what can be generated is expanding. For example, facial generation technology, famously known as "deepfakes," is gaining attention. By generating non-existent individuals and utilizing them as advertising assets, it promises numerous benefits: personalization, increased pattern diversity, simplified rights management, and streamlined filming/editing processes. Not only faces, but also natural motion synthesis is now possible, and voice synthesis has reached a remarkably natural level.

Combining these technologies could soon enable things like "automatically generating and delivering video ads where a fictional character tailored to each user delivers an optimized message."

2. The Emergence of Virtual Talents

Using the various generation technologies mentioned above makes it possible to create not only fictional characters but also virtual representations of real people. These are known as virtual talents. In September 2021, " Virtual Wakadaishō " was born, featuring the singing voice of actor Yūzō Kayama recreated by AI.

Employing virtual talents enables the creation of advertising materials without filming and the generation of large volumes of video ad variations, which were previously difficult. While not explored in depth here, if a fully virtual world like the "metaverse" becomes mainstream, such virtual characters will likely become increasingly important in future communication.

3. Advances in Optimization

In marketing, one key theme has been how to effectively combine online and offline strategies—the so-called "marketing mix"—to optimize return on investment. AI holds the potential to support this.

For example, it could automatically generate integrated online-offline plans to optimize reach for highly specific, granular target audiences. Rather than relying on past survey data or experience, it could optimize future activities based on demand forecasting and predicted media exposure. Such borderless optimization should become possible.

So, will human planners and creators become obsolete? No, marketing isn't that simple. The true creativity lies in how to outsmart people's behaviors and thoughts with limited budgets and drive action. If an era arrives where "average marketing" is automated, the next frontier will likely be a battle of ideas once again.

We'll leverage various AIs to tackle what was previously impossible. Or we'll outsmart AI itself to create unprecedented value. That might be what we call "Creativity in the AI Era."

4. Ethical Considerations

As AI adoption deepens, ethical considerations grow increasingly critical. We've seen cases where AI chatbots made inappropriate remarks leading to shutdowns, or where AI-driven hiring processes unfairly disadvantaged certain applicants. In Japan, controversy arose when AI predicted job seekers' "job offer rejection rates" and shared that information with companies.

In marketing, consider this hypothetical scenario:

• AI automatically generated creative content containing discriminatory expressions or phrasing.
• AI-based classification based on discriminatory factors caused disadvantage to some users.
• AI predictions and personalization became overly granular, causing user discomfort.

Because AI makes predictions and judgments in place of humans, it carries the inherent risk of producing such undesirable outputs. The risks associated with AI are not solely the responsibility of developers; companies utilizing AI also bear ethical responsibilities. What risks exist? How can they be prevented? How should they be addressed if they occur? Establishing guidelines for these matters is also required.

We have now overviewed the current state and future of AI utilization in marketing. While the initial frenzy has subsided, AI continues to command attention. Regardless of industry or size, every company must correctly understand both the usefulness of AI and the dangers inherent in its usefulness, and learn to harness it effectively.