Intelligent Decision making is key to any business. The entire process of decision making requires tremendous amount of time, effort and attention. And this cannot be achieved without the guidance of management and strategy consulting firms. It’s crucial for managers to understand the business, gather information, investigate alternatives, zero down on the best route and make decisions. In this entire process, the managers spend more than 80% of their resources in searching for relevant data and less than 20% in making decisions. With Brevo decision-making tool, the goal is to 1) reduce the effort required to collect and present the information and 2) provide ample space to discover patterns and analyze them to answer business questions.
For instance, a Manager in a sales company that sells XYZ product must do a what-if analysis on the following:
- Sales forecast to plan the business in coming years.
- What-if the profit increases?
- What-if hire more employees?
- What-if spend more on marketing? And so on.
This information can be helpful to make business decisions:
- How much money needs to be invested?
- What departments need to be investigated?
- How much has to be spent on resources?
The approach was not just to build an application which will answer these questions for decision making, but also assist the users with meaningful insights at every step, and induce users to act on the system by automatically providing insights into the data. Such insights and patterns are the typical ML use case.
However, when we come to ML for businesses the classic approach looks like the below chart. We formulate a specific problem, gather data, explore, model it and present. But it doesn’t solve the problem fully. The UX, on the whole, is not being addressed.
“Despite its name, there is nothing “artificial” about this technology—it is made by humans, intended to behave like humans, and affects humans. So if we want it to play a positive role in tomorrow’s world, it must be guided by human concerns.”
– Fei Fei Li
It was important to humanize the complete process to recognize and address the pain points or rather the opportunities where the system can be made intelligent. We designed a decision-making tool Brevo Value Driver Tree by solving the ML problem together with the design thinking process.
Why did I choose SAP CP Services?
To ease the process of decision making and to implement solution obtained by following design thinking methodology, leads to developing an intelligent tool which helps decision makers to visualize, predict and analyze various business scenarios.
- Key Influencer algorithm provides a list of influencers in their descending order along with their ranks, and their impact based on the selected target. This helps in building a decision driven tree automatically. Positive and negative influencers are identified by the system and user could further drill down into and analyze the impact.
- Multiple Linear Regression Algorithm is used to derive the relationship between the impacting measures. What-if analysis can be performed by varying one of the inputs, describing all possibilities and probabilities.
- Time series forecasting is used to analyze trends, cycles, and fluctuations in time series data. Utilizing historical data and forecast length, it predicts future values using the SAP HCP predictive services. Thus, this functionality assists business users in making more data-driven decisions, especially when combined with their business planning processes.
- Multiple services from SAP Cloud Platform Experience Maker was used to successfully build the solution. Interactive prototypes using Fiori Design principles for the solution were built using the collaborative tool SAP BUILD. With SAP Web IDE we were able to deploy to cloud, test, develop and build the application. GIT was used to do source code management, keep track of changes and coordinate on the files with multiple users. SAPUI5 provides rich UI controls, consistent user experience, support of jQuery doesn’t show up any constraint on integrating third-party libraries. Additionally Sap Theme Designer, Fiori launchpad and SAP Cloud Platform Portal services were also used to spruce up the look and feel of the solution.
What were the challenges?
- Building an intelligent application: When building an intelligent application, the system is expected to present events, data and action points to the users instead of users looking for them. We tried achieving this by tackling design thinking and machine learning simultaneously. Every UX pain point was looked into with ML first approach and this helped us to transform pain points into opportunities for a smart application.
- Finding the right algorithm: When building a tool for a wide range of use cases, it’s always important to get the right algorithm which provides optimal results for all the use case. Different algorithms had to be tried, optimized and tested for a variety of data sets.
- Responsive interaction of application: When building apps with a wide range of features, such as impact analysis sliders, ad-hoc calculation of nodes etc., the mobile-first strategy doesn’t always work. The features that go into a mobile device had to be well-thought, so it doesn’t require unnecessary development effort.
Using these SAP Cloud platform services we built a solution- the Brevo Value Driver Tree. It is a representation of the business model that links a business value (what managers or stakeholders are concerned about e.g. profit) to a set of drivers (any variable that can affect the value). We can run simulations in one area to see how it will impact other areas. It eases making daily business decisions by providing visualization of scenarios in the tree structure with the target value and impacting values compared to other methods. It is also easily shareable with other decision makers.