How eflexgroup.com Sped Up Its Claims Process Using Lean Six Sigma

In an earlier article I discussed how eflexgroup.com (eflex) took the plunge into Lean Six Sigma. I’ll now focus on our first project.

Our company began its journey of implementing Lean Six Sigma and continuous process improvement in May 2007. eflex was growing rapidly, averaging about 30 percent revenue growth per year. As our expertise in the world of employee benefits, health flexible spending accounts, dependent care accounts and health savings accounts brought success, it also pointed out the weaknesses in our processes. We promised claims paid within two business days but were averaging between seven and 10 days at that point.

Lean Six Sigma was going to help speed up the claims processes for eflex. Interestingly enough, while some people fear using the methodology’s complex statistical tools (a potential roadblock for Lean Six Sigma deployments), for us, ignorance was bliss. We didn’t know that we were supposed to be afraid of these tools, let alone know which ones to use. With our first Lean Six Sigma project we didn’t do a chi-square, paired t-test, ANOVA, factorial DOE or even a Larssen-Russe transformation. (OK, I made up the last one.) We started simply.

Looking at How to Delight the Customer

First, we sat and talked about what it was our customers wanted. Although we advertised claims paid within two business days, as a company we have never been satisfied with meeting minimum requirements. Our CEO, Ric Joyner, challenged us to ask what would delight the customer. We decided that we wanted claims processed within 24 hours of receipt because that is what our customers wanted.

Examining the Problems in the Claims Process

As we were casting about for direction, we settled on a fishbone diagram (or Ishikawa for those who wish for the jargon). We carried out structured brainstorming in order to come up with possible causes for slow processing. A picture began to emerge of the difficulties within the claims department.

The picture of our problems then became much clearer when we created a detailed map of the claims process (right, click on image to enlarge). We realized that the manager was doling out work to individuals according to what she thought they could handle. As long as each processor understood what needed to be done with each claim, the process moved forward.

When the processor did not understand what needed to be done, the claim would be re-routed back to the manager. The manager would eventually look the claim over and send it back with an explanation — a task that she had to juggle among routing other claims to be processed, managing the check printing details, and completing additional duties. Finally, the original processor would process the claim. If it were really complicated, the manager would simply process the claim herself to save time. Can anyone say non-value added?

We also knew that we were supposed to measure things. We needed a baseline, so we set out to measure how quickly people could process claims as things were. This certainly scared some of our staff. People wondered if they would be eliminated if they were deemed too slow.

Here was an opportunity for some critical engagement with the work force that was responsible for this process. They needed to understand that we needed to know how quickly they could process "before" so that we would know if they were faster "after." Once they understood and trusted the motivation for the measurement, they began to open up about what would make it easier for them to work more quickly.

As the staff spoke with us, we discovered that most of the claims we received were ready to be processed and paid as is. A much smaller percentage of them required more documentation, advanced knowledge, or additional contact with the customer (we didn’t know what a Pareto chart was yet).

Redesigning the Process

This discovery led us to adopt a variation of the U.S. Marines’ ideal of "Every man a rifleman." For eflex, this became "Every staff member a claims payer." If necessary, everyone in the company ought to be able to process the simplest of claims, and if we needed to hire extra staff we should be able to pull them off the bus stop outside our conference room window and teach them to pay claims in an afternoon!

Accepting this as our goal, we redesigned the process to support it. Instead of one manager pushing claims out to individuals as she saw fit, the new process involved a team of experienced processors classifying claims and forwarding them to a holding area for each classification. Each class has standardized work and each staff member is certified to work on various classifications.

Now there is a pull system: a processor logs into the holding area for his or her level of certification, takes up to a prescribed maximum, processes the claims, and pulls the next batch. If the processor processes according to his or her training and the standard work, mistake proofing is built in (no, we didn’t know what a poka yoke was — it was just common sense).

Before implementing the new process, we tried it out. We measured how quickly claims in each of the classifications could be processed, and all of them were faster than the old process.

Most of the improvements to the new process came from our frontline workers. They told us what body of knowledge was required for each classification. They helped us see potential failures ahead of time and helped us to find solutions. We got buy-in from them because we honored their knowledge and contribution to the company. Our claims processors take pride in serving our customers, and they did not like it when they were falling behind.

Beyond training and re-training our claims processors, we did make everyone in the company a claims processor. Every employee, from the C-suite down, had to take the claims training for the lowest classification and pass the test with a score of 100 percent. This gave us a great resource should we end up behind schedule with processing. It also improved the customer service we were able to give in our call center and increased first call resolution. Each year we require everyone to take a refresher course and pass the test again.

Committing to the New Process

Now, the implementation of the new process was not instantly a stunning success. For the first two weeks it bombed. We were slower and less accurate than we had been. People had many questions about how to do things the new way.

Several meetings were held along the way to ensure that everyone stayed on the same page. We kept talking. The CEO remained committed to sticking with the new process because he trusted the data and knew that our old way was not fast enough. After two weeks we started getting faster and more accurate. Within a month we were consistently hitting our target of averaging paying claims within 24 hours on average.

A Retrospective

So, looking back, we used fishbone diagrams, detailed process maps, and basic manual timing of claim processing. It is a list of very basic tools, but coupled with the DMAIC framework, very powerful. A bunch of untrained, non-belted "practitioners" used a book to guide them and saved $400,000 in the first year in a company of only $4 to $5 million gross annual revenue.

We learned to listen to the frontline workers, the power of a detailed process map, and that making decisions based on tested data rather than gut feelings produces consistently better results. It deepened our commitment as a company to incorporate Lean Six Sigma into the way we would do business from that time forward.