When you envision the planetary trajectory, what will come to your mind? Some disorganized lines, or a one-dimensional picture whose arrangement varies with longitude? This is one of the simplest and most classic cases of data visualization.
Graphic of longitude variation from Toledo to Roma, drawn by Michael Florent van Langren (1600–1675) in 1644 and deemed as the first (known) statistic graphic in the world.
Data visualization tools can make data and tables interesting. Then how to grasp data visualization technology and apply it to everyday advertisements? Tony, from hdtMEDIA, shared his experience on the 25th T Day (Technology Day).
What is data visualization?
Data visualization refers to a process of displaying a large quantity of data through graphic images of different forms, positions, sizes, values, orientations, colors and textures, and finding unknown information through data analysis and development tools.
Its basic idea is to consider every data item in the database as a single graphic element and turn reams of data sets into data images, while displaying the attribute values of data in the form of multi-dimensional data, so that users can observe data from different dimensions for further analysis and study. Effective communication of information is one of the most important advantages of data visualization.
Theoretically, data visualization usually involves the display of second-class data (latitude and measurement). So, Tony thinks that it’s not difficult for him to understand and apply data visualization tools. With the right tools and templates, data visualization can be very easy!
Currently, such tools mainly include Echarts, DataV, D3, Kibana, Tableau and Quick BI. Most of these tools not only support drag-and-drop operations, but also support data display and analysis, thus able to satisfy the needs of enterprises for business intelligence. Some can also generate h5 pages, so that users can view and spread them on the mobile terminal.
The application of visualization to advertisement
Advertising involves the supply side, advertisers and agents. And in this process, to make the right decision, they all need to monitor traffic, bidding, exposure and clicks. Among current advertisement systems, data display still exists in the form of two-dimensional tables. If data visualization is used, data will be clearer and more direct and the systems themselves will also be more useful.
In his sharing, Tony focused on the needs of advertising in different phases and displayed how well data visualization technology is applied to advertising. Of course, all this is based on the real-time collection and processing of data, because without real data, data visualization is only superficial progress without substantial ability.
Before advertising: determine the traffic and exposure of the platform
Usually, agency companies estimate the traffic of large platforms, but to accurately evaluate the differences of platforms is much harder. In addition, smaller traffic platforms in the market are too many to be known and accurately evaluated. A bubble model is used to import data on the traffic distribution of platforms. Bubbles of different colors and sizes represent the traffic of different media. This helps rapidly and fully learn about the traffic distribution of different platforms.
After the traffic is determined, the exposure of traffic platforms is to be evaluated. The total exposure of the PC and mobile terminal, the exposure of platforms in different channels and the correlation between such impression and platform traffic are too complex to be explained in simple words. But they are key to platform selection in advertising.
In the data visualization model, the sunburst graphic presents a tree structure and radiation layout. The circles represent dimensions and in the same circle, different sizes and colors represent different values and classifications.
Take the above question. The core of the graphic represents the total impressions, the next represents different PC terminals and mobile terminals, and the next platforms. This simple graphic can clearly show the total distribution of exposure. A comparison of two rainbow explosion graphics helps understand and evaluate the traffic and impression of platforms in different channels.
During advertising: the impression and click monitoring on platforms
If we are to make a radar map with traffic, impression, click, bidding and costs as dimensions to monitor the advertising process, this graphic is supposed to be distributed in a concentric manner, i.e., high traffic brings high impressions and high clicks. Meanwhile, the number of bidding and costs also reach the maximum on this platform. Therefore, the total costs of advertising reach the maximum.
The reality is much more complex though. Traffic platforms with a low pv may have good impression; the bids completed may be many, but the total costs are relatively low. Thus, when an ad is delivered, a rational decision is made based on the traffic, impression and click of platforms.
Real-time monitoring is the ideal state for advertising monitoring. In data visualization, real-time dynamic display with regions as the latitude is applied the most. Take impression. When the real-time impression of regions is displayed in a single map, the advertisement performance in regions can be compared in real time. Of course, on the basis of data collection and processing, data visualization can also monitor in real time customized advertisements of other latitudes.
Advertising prediction: visualization based on advertising data analysis
Compared with display, analysis and prediction are more important in data visualization. After an advertisement is delivered, a large amount of data, including time and effect, is generated. Graphics that are made based on the calendar and present the advertising needs in different time and of different brands to predict advertising tendency help clients seize the opportunity in the next round of advertisement delivering.
“Data visualization may use many technologies, but with a mature model, these technologies can be learned through simple training even for common users. This is very useful to advertising and we will add more visualization to our system.”