Marketing Data Analysis: The Front Lines. The Real Reason Marketing Accuracy Isn't Improving? (Part 1)

To enhance marketing precision, data analysis is naturally effective. In recent years, advancements in technology like AI have made it possible to measure and track data that was previously unmeasurable. Furthermore, the emergence of "big data analysis," which combines vast amounts of this data, has increased the frequency of new discoveries, making data analysis increasingly important.

However, while some companies effectively leverage data analysis for their marketing, others attempt it without seeing results... We increasingly hear voices expressing a gap between the widely touted reputation of data analysis and their own company's efforts. It's undeniable that this disparity in outcomes is beginning to emerge among companies.

Therefore, this time we interviewed Mr. Hiroyuki Sato of DENTSU CROSS BRAIN INC., a data scientist with extensive experience. We heard about the latest trends and challenges in marketing data analysis from his perspective on the front lines.

What kinds of inquiries are data scientists receiving now?

Q. This might be a simple question to start with, but what kinds of inquiries do you most frequently receive from companies right now? Are requests like "We want to do big data analysis" increasing?

Sato: Well, to be honest, the scope of inquiries is incredibly diverse (laughs). That said, I think they broadly fall into three main types.

First, Type A: "We want to start utilizing big data, but we don't know where to begin. We need consulting on the initial steps." Then, Type B: "We've built the infrastructure to handle big data, but we don't know how to actually use it. We want guidance on effective utilization methods." Finally, Type C: "We have the data collection infrastructure and are already doing data analysis. However, we don't feel we're utilizing it effectively on our own, so we want external opinions."

I think we can divide the level of engagement companies have with big data analysis into roughly three phases based on their current stage.

Q. Among these three types, is there a particular phase that has seen more consultations recently?

Sato: This is purely my personal impression, but I think it varies significantly by industry. Companies directly facing consumers, like retailers or e-commerce, tend to be more Type C. They're already doing analysis but lack confidence, feel the accuracy is low, or want to tackle new areas but lack the skills.

In contrast, manufacturers tend to fall more into Types A or B. While very few companies have "absolutely no data infrastructure," it's still quite rare to find companies where data from various departments is integrated and usable. Based on my gut feeling, about 40% of the companies we consult with fall into Types A or B.

Q. Why have manufacturers recently started pursuing big data analytics? Unlike retailers or e-commerce firms, they don't directly face consumers, so companies that previously showed little interest in big data analytics have suddenly become keen. What changes do you think drove this shift?

Sato: No, it's not that they weren't interested before; I believe the interest itself has existed for some time. However, without direct consumer touchpoints, they couldn't obtain what we call "user data" until now. But with recent technological advancements, an environment has emerged where data can be collected even without direct consumer contact. Of course, conducting their own research is still an option, but companies can now acquire user data by utilizing data clean rooms provided by major platform operators or through partnerships with other companies. Consequently, regardless of business type or supply chain, many companies are starting to engage in big data analysis.

Why isn't marketing precision improving despite data analysis efforts?

Q. With so many companies tackling big data analysis, there must naturally be a gap between those succeeding and those struggling. What problems do you think the struggling companies face?

Sato: There are certainly various factors, but from what I observe, most often they seem to be suffering from "narrow vision." For example, when tackling digital marketing, there are various metrics commonly discussed, right? For ads, it's CPA; for websites, bounce rate; for email newsletters, open rates or click-through rates, and so on.

Everyone is certainly working hard to chase these metrics or identify differences through A/B testing. However, it's all too common to see cases where they fail to "take a bird's-eye view." From the perspective of data scientists like us, the real value lies in looking at all these numbers holistically to pinpoint where the truly significant problems lie.

Q. I myself have operated a website in the past, chasing metrics like traffic numbers and conversion rates. But the more I looked, the less I understood. I ended up just staring at the numbers, unable to translate them into improvements. It's embarrassing to admit.

Sato: I completely understand. What's even more concerning is that surprisingly often, people don't actually know what the metrics they're looking at represent. For example, if you ask a digital marketing manager, "What exactly does the conversion rate you're checking in Google Analytics measure?" very few can answer correctly. Many people don't truly understand how that metric is measured in the first place, or what its meaning and nature actually are.

Take a recent case, for example. A client offering a subscription service consulted us. Their concern was: "Our churn rate dropped during the pandemic. But once things settle down, won't it just jump back to its previous level? We want that analyzed."

So, I immediately looked at the client's data. From my perspective, it turned out that "the churn rate hadn't actually decreased during the pandemic." What I mean is, while the "churn rate" figure itself appeared to have dropped, coincidentally, the number of new users acquired had fallen just before the spread of COVID-19. If you think about it, the time when people most often cancel a subscription service is right after signing up. Therefore, the more new users you acquire, the higher the churn rate inevitably becomes. Conversely, if new user acquisition drops at a certain point, the churn rate will also decrease at the same time. However, without looking at this underlying dynamic, they were only focusing on the fact that "the churn rate had decreased" and concluded, "Due to the impact of the pandemic, digital services became popular, which resulted in a lower churn rate." Naturally, this led to subsequent "measures to reduce the churn rate" being completely off the mark.

Regarding marketing data analysis using big data, Mr. Sato explained that many struggling companies suffer from "narrow-mindedness" and may lack the ability to "view metrics holistically." He further noted that insufficient understanding of the fundamental meaning and nature of the metrics being handled can lead to distorted data interpretation, resulting in misguided marketing initiatives. In the upcoming Part 2, we will delve into the essential qualities and attitudes required for data-driven marketing analysis and explore the secrets to enhancing marketing precision.

• The Cutting Edge of Marketing Data Analysis: The Real Reason Marketing Accuracy Isn't Improving (Part 2)