The world’s largest companies use data analytics to increase their revenue and keep up with the changing business world. But how does data science relate to simulation modeling, and what are the cases for the implementation of this interaction, primarily concerning value for the business? The United Services Automobile Association (USAA), a Fortune 500 group of companies, has answered these questions with real-life solutions.
Getting business value through modeling
To make the right decision and benefit from it, executives need to answer the following questions, specific to their business:
- What are the company’s options in terms of money investment?
- Where is the breaking point/capacity limit of the business?
- What are the likely outcomes/business impacts if certain actions/decisions are considered?
- Is the business agile enough to handle sudden shifts?
Methodologies such as data-mining or machine learning do not respond to these questions. USAA analysists found the answer in AnyLogic simulation modeling. It goes beyond analytical modeling, and combines business processes with assumptions that executives make. Simulation modeling is used to visualize system behavior, processes inside the system, and their aftermaths, and prescribe a solution. This approach explains why the system will act in a certain manner and explores a wide range of outcomes.
Case #1: Call Centers Management
USAA owns large call centers with highly complex infrastructures. The USAA representatives wanted to model the call center framework in order to optimize the headcount and the scheduling and routing of calls, by using aggregated data. These steps aimed at improving call center overall utilization and customer satisfaction rates, as well as lowering the abandonment rate.
Call center simulation models are quite widespread. However, because of deficiencies in utilized approaches, these models were neglected. Some of the deficiencies are listed below:
- Low granularity in terms of time and skills/groups of people
- Missing effects of abandonment behavior and multi-skilled sales representatives
- Very simplistic routing strategy
- Inability to take into account call and sales rep attributes, attribute-based routing, and individual agent behaviors
Solution and Outcome:
The AnyLogic model represented the incorporation of calls, sales representatives and their skills, routing and abandonments in details. What-if analysis was performed to find a suitable solution. The insights which were gained from simulation and optimization acted as a basis for improvements in the call center working process. For example, customer service index rose significantly due to reduced wait time, while the abandonment rate dropped down, which increased the revenue. Due to the changes in the working process, it became possible to cut hiring and training costs.
The company has been using the model for several years, and is still using it, applying modifications to reflect the changing environment. Representing the stand-alone contact center, the model can be expanded in the future into the entire call center eco-system.
Case #2: Investment Planning
Companies are trying to plan investments, while facing the problems of prioritization and placing them on an annual roadmap. USAA challenged these issues with the AnyLogic simulation tool and created a model with a high level of abstraction on how the investments could be prioritized.
In the model, the capability roadmap visualized possible investment plans, while interdependencies between them exposed costs, benefits, and possible risks of each investment plan.
When a multiple portfolio of investments was created, modelers simulated the operations of the company with these investments and players, and analyzed what the expenses, revenue, and profitability in the long-term period would look like, and which resources might be under stress.
The AnyLogic simulation modeling approach helped reduce operational risks and find out where these risks might surface. It also enabled USAA to see the benefits of each investment plan and see the prospects of each plan in a 12-15 year period. This strategy provided the company with a roadmap to follow, and facilitated performing proactive mitigation strategies.