Problem Context: A healthcare-based Revenue Cycle Management firm found a major point of failure in the beginning of a long process. In essence, front-line workers were taking customer insurance information and manually entering it into their systems. The amount of data entry was substantial, and one small error with a minor ripple effect could cause a wave of larger problems downstream that involve multiple departments.
Solution: After some needs analysis meetings, it was determined that training would be an appropriate intervention. However, it’s important to note here that while I was in support of some training, I voted much more heavily to develop a system that could use AI image recognition to scan a customer’s insurance card and automatically enter the information into the appropriate areas of the system to ensure accuracy and to greatly reduce data entry time. This idea was seen as too complicated and time-intensive to pursue, so the training solution was the sole focus. To that end, I recommended an e-learning that simulates the system, adding highlights and Just-In-Time popup boxes around relevant fields as needed. It would be similar to a screen recording, but with interactivity and allowing learners to, in essence, have a “sandbox” environment for the software which would otherwise not exist.
Result: After completing, implementing, and assigning the e-learning module, the Learning & Development department ran post-test statistics to determine the effect of training on data entry error rates. There was in fact a 20% reduction in errors, so by that metric alone, the training was a success. However, software simulation e-learnings are the most complex kind to build, and as the only Instructional Designer with e-learning development experience, the entire development responsibility fell on my shoulders, creating poor time-to-value. In addition, an egregious amount of collective hours between myself and other team members were spent discussing the solution – more than I felt was reasonable. Given this, I thought that the 20% improvement on error rates was not a great return on investment. Compared to all aggregate human resources costs to develop the training, it likely would have actually been less expensive to pursue developing the AI image recognition software to potentially eliminate the error rates to virtually zero. What I identified as an opportunity for improvement after this experience had to do with my agreeableness; after initial rejection, I did not push the AI solution hard enough, and allowed the conversation to be steered away from an innovative solution, towards a more conventional solution instead. Still, this experience was a great stepping stone towards my continuous improvement of being a better communicator.
