Colopl and Dentsu Inc. Analyzed Location Data

In May of this year, Colopl and Dentsu Inc. jointly launched the location-based marketing service "miraichi". This time, we spoke with Mr. Yukiteru Sakai of Colopl and Mr. Kenichi Amami of Dentsu Inc., who are involved in the service's development and provision, about what insights location-based big data can offer for marketing challenges, as well as the features and usage methods of "miraichi".

Developing High-Value-Added Services with Colopl and Dentsu Inc.

——Colopl is known for smartphone games, so why did you release a location-based marketing service?

Sakai: To ensure our smartphone games provide a pleasant experience for many players, we analyze massive amounts of data in real time to adjust game balance and events. While the success of hit games often highlights creativity and development skills, we're also a company with strong analytical capabilities, backed by robust data mining environments and personnel.

Leveraging this analytical strength, we launched a location-based big data utilization project with KDDI. Initially targeting local governments, as the scope gradually expanded to retail and distribution industries, we recognized that adding Dentsu Inc.'s marketing expertise could create an incredible service. That's how "miraichi" began.

Amami: Dentsu Inc. supported the development of this service by advising on how best to present Colopl's location-based big data utilization to clients, given the diverse marketing challenges they face daily.

When explaining the service to clients, Colopl and Dentsu Inc. members often visit together, collaborating as business partners.

Sakai: Through our collaboration with Dentsu Inc., our client touchpoints have increased dramatically. We feel the scope of application is expanding daily—location-based big data can solve problems we never imagined, and conversely, it uncovers issues clients hadn't previously noticed.

Understanding people's movements, understanding people's characteristics, and connecting them to the next strategic move

——What are the key features of "miraichi"?

Sakai: It's a reporting service that utilizes location-based big data obtained from au smartphone users with their consent (*). It allows us to track people's movement patterns over time—including their departure points for any given area, gender, age, duration of stay, and visit frequency. We can also analyze where people flowed to after visiting a target area.

※Colopl, commissioned by KDDI, analyzes data processed into an anonymous format based on the locations of au base stations used for communication/calls and the current location information of au devices. Reports are created incorporating Dentsu Inc.'s marketing expertise. Location big data is never provided to Dentsu Inc., and the data contains absolutely no personally identifiable information.

 

Amane: Since we can analyze dwell time and visit frequency, if people who habitually visited that area stop coming, we can identify them and understand where they've gone instead.

——Could you tell us more specifically about the actual analysis content?

Sakai: Since we have this opportunity, let's analyze the movement of people during the Halloween season in Shibuya and Roppongi, which recently saw significant excitement, using "miraichi" as an example.

Amane: It's a bit odd since the city is already fully in Christmas mode (laughs). But since it's such a widely discussed event, I think it's easy to visualize.

Visualizing Halloween Crowd Movement

Sakai: First, Figure 1 shows the analysis of when people started gathering in the Shibuya area.

<Figure 1: Number of people entering Shibuya by time slot: Top: Overall / Bottom: 20s>

Sakai: On a typical day (blue line), there's one peak influx around noon. But on Halloween itself (red line), the peak influx occurs around 4 PM. This trend is especially pronounced among those in their 20s, showing a clearly delayed peak influx compared to usual days.

Amane: I understand the image of people flowing in from the afternoon for lunch or shopping on a typical day. So, where were the people who came to Shibuya on Halloween before 3 PM?

Sakai: <Figure 2> shows where Shibuya visitors were staying before 3 PM.

<Figure 2: Locations where people who visited Shibuya after 3 PM were staying before 3 PM>

Sakai: We can see they were spending time in other areas like Minato Ward, Shinjuku Ward, Chuo Ward, Chiyoda Ward, and Setagaya Ward.

Amane: So there's a significant number of people who had lunch or went shopping elsewhere before heading to Shibuya. It's interesting that many people plan to spend Halloween night in Shibuya even if they have daytime commitments.

Sakai: Next, let's look at the inflow times to Roppongi. Figure 3.

<Figure 3: Number of people entering Roppongi by time slot: Top: Overall / Bottom: 20s>

Sakai: Similar to Shibuya, there's an influx around 4 PM, but there's another influx around 10 PM.

Amane: Even after Halloween's peak excitement, it's possible some people are still moving around to enjoy the festivities. This movement is especially noticeable among those in their 20s.

Sakai: Figure 4 shows the analysis of movement patterns between areas for those in their 20s.

<Figure 4: Trends of Visitors Traveling Between 2+ Areas (20s)>

Sakai: You can see a high frequency of movement between Shibuya and Roppongi.

Amane: Recalling the day, it's conceivable people evacuated from Shibuya to Roppongi because Shibuya was too crowded to walk or felt intimidating. Alternatively, similar to hopping between different spots before coming to Shibuya, they might have changed locations to enjoy time with different friends. Looking at actual Twitter posts, both types of people seem to exist.

Sakai: Figure 5 shows the outflow time analysis for Shibuya and Roppongi.

<Figure 5: Number of people leaving Shibuya by time slot: Top: Overall / Bottom: 20s>

<Figure 6: Number of people leaving Roppongi by time period: Top: Overall / Bottom: 20s>

Sakai: While Roppongi shows a clear peak around midnight, just before the last trains, Shibuya has a peak around 8 PM, but the departure times appear more dispersed. Research shows that many visitors to the Roppongi area come from nearby locations like Minato Ward, Shinjuku Ward, and Meguro Ward.

On the other hand, visitors to the Shibuya area include people heading back to places like Funabashi City in Chiba Prefecture via the Chuo/Sobu Line, or Machida City and Kawasaki City in Kanagawa Prefecture via the Odakyu Line. The fact that many people have earlier last trains seems to be reflected in the dispersed departure times.

Amami: Halloween has gained widespread popularity in the last five years or so, coinciding with the proliferation of social media and smartphones. As analyzed by Dentsu Inc. Wakamon, this trend shows that a segment of young people—who maintain multiple communities via social media and employ the communication technique of "switching personas" to expand these networks—spent Halloween hopping between venues until the very last train, as evidenced by actual behavioral data.

——Thank you for sharing these specific analysis findings. This seems applicable for understanding corporate marketing challenges and planning next steps.

Amami: As some reports noted, Shibuya especially sees people gathering spontaneously, making it hard to predict when the peak excitement will start. While this applies to such sudden events, "miraichi" excels at quickly capturing and responding to environmental changes—like new openings at nearby large facilities, impacts from primary transportation shifts, or the effects of your own store renovations or large-scale promotions.

Sakai: That's right. By leveraging location-based big data, you can even track customer traffic to competing stores within the same commercial area. This allows you to visualize the market share battle between your store and competitors within the broader market context, significantly improving the accuracy of your situation assessment. As with game operations, if you can't correctly determine whether an implemented strategy succeeded or failed, and why, you can't prevent missteps. We often receive praise for this aspect.

"miraichi" – Connecting to the Next Move

——Finally, could you share your outlook for "miraichi" moving forward?

Sakai: "miraichi" evolves daily, absorbing diverse needs from companies in retail, distribution, railways, theme park operations, and urban development. We're often surprised ourselves by the high potential of location-based big data. Accumulating these use cases will solve more client challenges. We want to expand this virtuous cycle further.

Amami: By combining not just location data but also asking data—like from Smart Answer (Colopl's smartphone-exclusive survey app)—we can gain deeper insights into visitors. Relying solely on memory-based questionnaire surveys often misses crucial details. Actual behavioral data frequently uncovers unexpected challenges and discoveries. If you're interested, we'd love to discuss how we can help.