Harish Boini

I had the privilege to use data visualization to enhance a business process, create self-service analytics and generate increased revenue of $2.93 million over eight months.

The skilled and knowledgeable sales managers struggled with using walls of Excel sheets to guide their actions in the field. Through layered inquiry, I understood the needs of the sales managers, their directors, and the executive team and developed a suite of Tableau dashboards.

Through the power of visualization, these dashboards help uncover physicians' prescribing trends and characteristics. These were highly interactive dashboards with links amongst dashboards and had user-level permissions to account for HIPAA restrictions. The sales managers were able to tailor their conversations to be targeted, relevant, and more effective than pitching the full set of the companies' capabilities. This was a tremendous time saving for the managers and provided key information to leadership in a more real-time fashion.

I provided full end-to-end service for this project from requirements gathering, data collection, data cleansing, ETL process setup using Alteryx, Tableau dashboard development, training, support, and subsequent product version enhancements.

Working in operations, the contact center is always thought of as a cost center needing a reduction of calls. However, this initiative flipped that thinking into making contact center an asset. If members are calling anyways, why not get additional data points from them. Data that was often inferred at best from transactional data could now be sourced directly from an automated member survey following a call with a representative.

This became a new capability of our analytics team where I would gather business hypotheses, generate survey questions, collect data from direct member post-call surveys and conduct the revenant analysis. Due to the large volume of calls, we would run a new survey every month.

The business was able to use this data to make direct product and preference changes, knowing that there would be alignment with members' needs, leading to more operational efficiencies. This process was also used to inform members through carefully worded questions. For example, asking members if they were aware the contact center was open 24/7 lowered the peak call volume at 7-8 AM for members waiting until business hours to call. With more even call distribution, peak hour staffing could be reduced.

The business often wants to understand if one event is related to another. This is straightforward when events are linked with a unique ID, such as an order and a return. However, when a member has multiple orders, web sessions, and phone calls, how can we tell which call may have been tied to which order. In addition, how do we make sure that we are not duplicating the data any tying a single call to more than one web session or one order with multiple calls.

I enjoy solving these challenges by using creative SQL logic in a multi-step process where I apply appropriate business rules and then look for the time difference and match records on the nearest events. In other cases, I develop logic to factor many to one relationships, i.e., three orders and only call translates to 1/3 call per order.

While this type of data matching is not exact, it can provide valuable insights. One study demonstrated that 80% of calls were made within 10 minutes of having a web session. This, in turn, led to a detailed analysis of what members were doing before calling and how to contain them digitally. Another study showed how member security preference changes impacted their propensity to be victims of fraudulent events.