Getting Performance Data

Rate x Time = Production was the topic of my last post. The application of several data analysis techniques makes it possible to identify opportunities for improvement and to accurately forecast the value of that opportunity.

The catch is the techniques and formula require data to be collected and processed. This can seem like a significant problem. We have to collect a complete and comprehensive amount of data for the analysis tasks to be worthwhile. In this article we talk about how to do this in cases where the data does not already exist. While this can seem like a daunting task, it is actually pretty easy.

I know because we used these techniques in a 1500 person company that had a limited number of pre-existing sources of data. It was a service company and the claims and call center teams had automated systems to collect data. In some departments, there was no data being collected and stored in a computer system. Since our need was urgent, we used the data that was available and we created new solutions where necessary. In some cases this literally meant having staff make marks on paper which were collected daily by management and hand entered into spreadsheets. Remember the goal, we want task level performance data by individual for each job type for each team/department/functional area of the company.

With the large variability in the type of data that was available we focused on the Rate component first. The number of data elements being collected can be very small.. We weren’t initially interested in how long people were taking to do a task. That was a nice second generation solution. We were interested in understanding how many value added tasks each person was doing per day. This could be reduced to a simple count of meaningful tasks. If we use claims as an example we are focused on how many claims each person pays per day. In addition, we are interested in understanding how many adjustments each person is processing. In a claims department claims paid is production, and adjustments are rework. The paper form that we used to collect information could be very simple. It might look like this:

We asked staff to make a mark to indicate a completed task. At the end of the day the manager for that team would collect all of the sheets and input the results into a team spreadsheet. This is obviously not everything that a claims person is doing, but it is their value producing tasks.

We did a similar exercise for every team in our business. Managers worked with their teams and a core group of data/process people to determine what tasks were important. In the initial scale rollout we had 77 managers and roughly 1500 employees. I played a coach/quality control role to ensure that teams weren’t trying to to something that was needlessly complex.

For the paper process the simplicity of the form meant that staff were likely not to be confused about what task to choose. It also meant that we could begin the baseline data collection 30 days after starting the project. The data for each team was aggregated and plotted to create productivity distributions.

We ran a 30 day baseline collection period. We ran the analytics at the end of every week, and at the end of the period we published our first baseline and our savings forecast. The numbers were pretty eye popping – we forecast a 10% reduction in total expense. The baseline period results were used to set the expected performance standards for each type of work.

Once the baseline was established we began our data driven performance management process. It took about 4 months for the managers to achieve the forecast productivity levels. This meant we were saving $800K/month.

All of the work was done without having to wait for a technology solution to be created or installed. It did require management behavior change and clear communication with our staff about why we were changing the performance management and appraisal processes. These can be changed fairly quickly if management has a clear idea of what needs to be accomplished and is committed to getting the results. You could start now and achieve your run rate savings in time for the 4th quarter of this year.

Next time we will talk about how to collect time data. In future articles we will provide you with a detailed case study where you can try these technique for yourself.