As the life insurance industry contemplates the wave of disruptive automation crashing against its shores and self-image, robotics process automation (RPA) is looking more and more like the tide that lifts all boats.
A lesson from Poppy
Xchanging, a London-based global provider of insurance process services, partnered with RPA software platform developer Blue Prism to create Poppy the robot. Poppy successfully automated several steps in creating London Premium Advice Notes (LPANs) and submitting them to a central repository. LPANs are documents used by brokers to transfer premiums to their business processing unit or business process outsourcer.
In an excerpt from Service Automation: Robots and The Future of Work, published as part of a special report on RPA from Professional Outsourcing Magazine, authors Leslie Willcocks and Mary Lacity describe the processing routine for LPANs this way:
“The original process involved the customer sending Xchanging an unstructured data file. The file has to be opened and validated. The operator then has to collect additional data from a system called “Account Enquiry.”
“Next, the LPAN is manually created and, with supporting documentation, uploaded to the Insurers’ Market Repository. This is a high-volume process that the operators did not really like doing…”
Not long ago, LPANs existed as paper documents. For the most part, LPANs are now viewed as image files, but some still exist in their original form.
With Poppy, humans structure the data on each LPAN into a standard template and give it to a thoroughly trained Poppy to be read. The robot makes the decision to either validate or make an exception of the request. Poppy bounces the exceptions out for human intervention, completes the validated submissions and moves them to the repository.
Prior to Poppy, there were seven processes required to handle LPANs from the time of broker submission, all handled by humans. Poppy reduced the number of human-centered processes to three, and in doing so, illustrates the necessity for human beings in the routine.
Data capture requires human intervention, as RPA has not yet evolved beyond working with structured data. Analysis and resolution of exceptions remain a human process as well.
Willcocks and Lacity report that before Poppy it took days to process 500 LPANs. They report that a “properly trained” robot can process these documents, without error, in about 30 minutes.
Poppy eliminated the need for training, overtime, and workforce adjustment. The first-time completion rate for LPANs reached 93% by May 2015, less than a year after Poppy was used to automate the first four processes.
Taking the right steps toward RPA
Xchanging is committed to continuous improvement, and in automating the project Willcocks and Lacity found that continuous improvement beyond deploying RPA delivers more benefits overall.
Their other takeaways were that high volume, low-complexity repetitive tasks are best suited for robotization, that staffers (like those at Xchanging) can work well as teams with robots, and that robots can outperform humans on “quality, speed and error-rate metrics, but can only work at the pace the overall process allows it to work at.”
The authors outlined eight of the most important lessons they learned about automation throughout the Poppy process:
- Sponsorship, a project champion and piloting are a must for RPA projects
- A culture driven by innovation and technology makes RPA easier to adopt
- RPA should be embraced as part of the company business
- Before automating, processes should be standardized and stabilized
- Compliance with governance and architecture policies
- Build internal RPA capability to evolve, leverage scale and increase business value
- Train the robots in multiple skills
- Pay careful attention to internal communications
The Poppy case study illustrates how an insurance process provider with a culture of innovation and dedication to constant improvement can embrace and benefit greatly from automation, and it’s certain that life insurers can reap the same benefits. However, the inability of RPA platforms to handle unstructured data will remain an issue for the foreseeable future.