In 2017, Spark Thinking, an online education brand focusing on children's logical thinking, was officially established, relying on small class live classes, real interactive AI classes and other forms, Spark Thinking combines the teacher's inspiration and guidance with animation, games, interesting teaching aids and other carriers, and guides and cultivates children's logical reasoning and problem-solving ability at all levels. Up to now, the cumulative number of students has exceeded 700,000, and it has become a leading enterprise in the domestic online education industry. Among them, "digital intelligence" is one of the bases that carry the rapid growth of spark thinking.
To a certain extent, online education itself is a data-native industry, but most online education companies cannot fully realize the potential of data value on their own. Spark Thinking is aware of this, and has made a scientific choice, and finally realized the successful release of data potential and detonated business growth.
So, where is the explosion point of the digital and intelligent upgrade of spark thinking?
Inclusive data consumption, igniting all employees to watch and use data
The core product involved in this scenario: DataWind
Core Introduction
Why: The core point of digital operation is that the enterprise has a good level of data consumption. Whether it is the data team, the business operation team, or the internal functional team of the enterprise, the use of BI systems in the past cannot bring more inclusive data consumption to the enterprise. In the past, Spark Thinking had built its own BI system, and the highest number of monthly active users of the system at that time was about 300. Due to the high threshold and poor user experience of the original self-built data system, in most cases, only data analysts and product R&D teams will use it frequently, and employees of the business operation team need to master certain code capabilities if they want to use it. Now: Based on the vision of realizing self-service business analysis around Volcano Thinking, DataWind, a digital intelligence platform of Volcano Engine, bids farewell to the SQL capability requirements of the original self-developed BI system by accessing multi-source heterogeneous data, so that all business personnel of the company can query data and draw charts through the drag-and-drop operation interface, and form different business topic dashboards. Let business people see and consume data.
Customer site
The number of monthly active users in the data system has increased by 2 times
With the visualization and intelligent analysis functions provided by DataWind, the digital intelligence platform of Volcano Engine, even business operation employees who do not know the code language can directly see and download the data they need from the data dashboard, and make the next refined operation actions based on these data. For example, in order to pursue a better student experience and a higher completion rate, Spark Thinking will add a management intervention process in addition to the platform process. In the past, this required the middle office to download an Excel list from the BI system, and then the team members needed to break down the list from the region to the team to the sub-teams and front-line managers—a lot of employees' time was consumed on table splitting and distribution tasks. However, with DataWind, front-line business managers can directly see the class details and absenteeism details of the corresponding students in their teams from the data dashboard, and then ask the instructors to remind the students to make up for the missing courses and get data feedback. On the one hand, team members no longer need to stare at the Excel sheet to split down the level, on the other hand, front-line teaching managers can also simplify their work from the data, and get feedback on the implementation effect from the data, so as to devote more time to teaching and management.
Thanks to the easy-to-use product features, more and more employees within Spark Thinking are becoming data consumers: data shows that the number of monthly active users in the data system is now around 800, which is twice as high as before; In the third quarter of last year, the Net Promoter Score (NPS) for internal employees also rose to 0.7 from 0.2 previously. "Learning Health" increased by 5 percentage pointsThrough DataWind's visual Kanban capability, Spark Thinking can directly display the multi-dimensional learning data under the learning health index to the teaching managers, so that each business team can see whether their courses have really promoted the improvement of learning health through a piece of data, and the middle and senior management teams can also let each sub-team learn from each other's successful experience and promote the improvement of the overall teaching quality. It is precisely because of the low threshold of the Volcano Engine Digital Intelligence Platform product that the data use threshold within Spark Thinking is lowered, the middle and senior management team leaders can gain timely insight into the excellent examples in the teaching process and reproduce them, and the teaching managers can adjust the course content and teaching strategies in a timely manner according to the data-guided decision-making application of the dual role, the health of Spark Thinking's learning situation has increased by 5 percentage points, and it has also led to the improvement of student retention.
DataWind shows the growing trend of academic health
Efficient business decision-making ignites the scientific nature of business strategy
The core product involved in this scenario: DataFinder + DataTester
Core Introduction
Why: When exploring new services, if you follow the traditional AB experimental method of spark thinking, you need to wait for another month or even a few months for traffic precipitation, and if you use other testing methods, you will be limited by complex processes and more resources. Now: DataTester is able to perform A/B experiments with small sample sizes, reducing the feedback cycle required by traditional A/B tests on a monthly basis to a weekly one. With the help of DataFinder and DataTester, Spark Thinking's marketing team can monitor the progress of experimental data and analyze differences in user behavior paths.
Customer site
- The success rate of new business user registration experiments increased by 30%.
The team in charge of the new business tries to use DataFinder and DataTester to perform the whole process of self-service operation on the business homepage, even if the traffic is lower than the traffic required for regular AB experiments, they can still start the experiment through DataTester, and track the experiment process and result data in real time. At the same time, DataFinder can also help the team gain insight into the user's activity path in new business scenarios and conduct further analysis. In less than a month, Spark Thinking's new business has obtained experimental results, and the success rate of the winning registration model can be increased by nearly 30%.
Everyone can consume data, and everything can be data-driven, which is becoming a daily routine under the digital and intelligent upgrade of Spark Thinking, which also coincides with the new model of enterprise digital intelligence upgrade launched by Volcano Engine. According to Zhang Junying, vice president of Spark Thinking Technology, the data flywheel can provide a new perspective on business and provide a large number of digital intelligence product support, so that enterprises can predict the future through data based on the data flywheel. This is particularly important for enterprise management who need to rely on data to make scientific decisions, and at present, Spark Thinking, whether it is the front-line management of the business or the core management of the enterprise, has become one of the main groups of data consumption. He gave an intuitive example: for example, if you want to predict the company's business volume in 3 or 5 years, one logic is to make some assumptions about each process and each link from front to back according to the company's business flow, but the final calculation result of this method will have a big error with reality; Another logic is to look at the business as a whole and turn it into a pool model with growth and loss, and the assumptions made on the basis of this model can be more accurate. The crux of the latter logic is the need to transform the horizontal enterprise business process into a vertical overall model, a process that requires a large amount of data to be analyzed from multiple perspectives. With the help of the data flywheel, it is easy for the management of the enterprise to see the business from all angles, especially with the assistance of a large number of digital intelligence products, such as data analysis and computing, which can help the management better predict the future of the business through data.
At this stage, the same data flywheel solution that has been practiced by Spark Thinking has been launched.
