Dentsu pleases clients with campaign analysis through Tableau + Hadoop


Cut down reporting time from several days to several hours

Integrate data from several media channels

Creatively display and share advertising campaign data with clients

Dentsu is the number one advertising company in the Japanese market, managing advertising campaigns for a variety of local and global brands. To track campaign performance, Dentsu analyzes hundreds of millions of records each month. Before Tableau, Dentsu used SAS and SPSS to create reports from four types of data: TV, online, survey, and purchasing. Analysts visualized the data in Excel and distributed reports in PowerPoint. Now Dentsu produces reports using Tableau and Hadoop, allowing clients to see and interact with their campaign data—all in one place. In this video, Shigeki Yamazaki, Technical Director, explains how Dentsu is increasing client trust with fast and high-quality reporting in Tableau.


"Several days could be shrunk to several hours"

Tableau: What kind of customers do you serve at Dentsu? Shigeki Yamazaki, Technical Director: It’s an advertising agency. My job involves building data platforms at the data solutions center, creating business such as analysis with data visualization. When it comes to Dentsu customers, we have major clients from various industries placing ads with us. It’s wide-ranging, from the automotive and beverage industries to telecommunications carriers. Tableau: What was your goal when you first started using Tableau? Yamazaki: We want to deliver more effective services to clients by pursuing integrated analysis including mass communication. It’s not possible to take action on things that can’t be seen, so they should be made visible. Tableau: How has Tableau helped you communicate with different types of clients? Yamazaki: When it comes to places that hold a lot of data, traditional companies such as those in the manufacturing industry accumulate a vast amount of undisclosed data. While we wanted to take advantage of this, we were under pressure to look into the means to do so, the means to actually do so. Using Tableau for this, there have been many cases where we can aggregate data on the system and visualize and use it there and then. Tableau: What tools do you use to aggregate the data? Yamazaki: For example, by using a number of services such as Amazon Redshift and cloud services such as Treasure Data, it is possible to use mechanisms that can quickly and stably aggregate and process even a great amount of data, and run these without paying the initial costs. After becoming able to process large amounts of data, Tableau can be used to see it. Tableau: How have these improvements made an impact at Dentsu? Yamazaki: We were able to focus on thinking about the data. By being able to focus on it, the work that previously took several days could be shrunk to several hours by turning it into one process. We became able to produce higher-quality reports after repeating this process many times. Tableau: Several days to several hours—that’s great! How will Tableau continue to impact the company? Yamazaki: And I think the level of creative expressiveness is unique to Tableau. I think that we’ll think of data creatively from here on, not just from a marketing standpoint. We really became aware that Tableau can express the importance of what is happening now creatively, in infographics. This was the first chance we had to use it. While this improved the client satisfaction, more importantly, by using Tableau, we could deliver to clients in Tableau format by distributing Tableau Reader. It is great that clients can use can use interactive tools and have such flexibility, and using new tools and technology leads to new relationships of trust.

With Tableau, the work that previously took several days could be shrunk to several hours by turning it into one process.

Analyzing millions of records in Tableau

Tableau: Can you explain how you use data at your job? Shigeki Yamazaki, Technical Director: It’s an advertising agency. My job involves building data platforms at the data solutions center, creating business such as analysis with data visualization. There are lots of records each month, hundreds of millions or billions. Naturally we also manage these, analyzing the data in such records and sending reports. Tableau: What types of data do you analyze in Tableau? Yamazaki: Up to now we’ve analyzed a combination of four types of data using SAS and SPSS: TV and online data, as well as survey data and actual purchasing data. We visualize it in Excel and create reports in PowerPoint. However, all of this can be done on Tableau. This was the decisive factor for us to introduce it. We were quite shocked at a tool that could easily import and connect so much data and visualize it. Tableau: Want did your process look like before Tableau? Yamazaki: When we carried out data analysis, we used 80% of the total time to collect, store, accumulate and aggregate the data. From 70% to around 80%. However, with this taking up 80% of the total time, we only had 20% of time to consider the data. This was extremely limited. Tableau: And how has Tableau improved the process? Yamazaki: Consequently, using Hadoop for this, by using Tableau for the visualization part, we were able to exponentially increase this 20%. The quality of the reports themselves for the analysis results naturally increased.