Issue link: https://industrialengineer.epubxp.com/i/905872
18 ISE Magazine | www.iise.org/ISEmagazine N Don't look back By Kevin McManus performance Now that the end of another calendar year is drawing near, more people are attempting to project how their teams, processes and organizations will per- form over the next 12 months. Sadly, far too many will use the conventional "last year's average plus or minus a per- centage" tactic to project performance. They won't think enough about the ex- pected impact of planned, let alone un- expected, internal and external systems changes. The potential for inaccurate projections looms large, as does the po- tential for misallocated funds and lost improvement opportunities. Better methods are out there. As a plant industrial engineer, I used good old regression analy- sis and spreadsheets to project monthly electrical costs more than 30 years ago. It was interest- ing to see how adding and removing dif- ferent factors affected one's effectiveness in predicting possible process outcomes. More importantly, it gave me insights into how you must consider a variety of factors when making future perfor- mance projections. How many different factors actually affect the outcomes pro- duced by your work processes? Had my IE manager at the time not pushed me toward such analysis, I may not have ended up so deeply enthralled with the world of systems thinking. As time went on, I learned more about how Monte Carlo simulations and other forms of modeling could "forecast the future" of certain work systems. Many times, I found that a moving range pro- cess behavior (control) chart, coupled with clean data, was one of my most useful performance projection tools. Sports provides some interesting ap- proaches for projecting future player (people) value and game (process) out- comes. Owners and managers use in- dustrial and systems engineering tools like regression analysis and Monte Carlo simulation for greater projection accura- cy. These same tools have been available to us for years, but we all too often use inferior ones like "percent bumping" or use process behavior charts and control limits incorrectly. Worse, we often use too much "dirty data" from the past to project future performance. Dirty data induces varia- tion into process behavior analysis. The sports world is making its projection ac- curacy gains not because the models and tools are more refined but because tech- nology allows for much more, and more accurate, data capture. Digital video and GPS tracking can capture every move- ment a player and ball make. Contin- ued data storage density and processing speed gains allow for data warehouse and algorithm variable expansion with- out bogging down the analysis process. However, we don't seem to be mak- ing the same gains in our organizations. Often, we simply fail to capture enough multidimensional data at the process level. In other cases, we comingle data from different processes, further mud- dying the analysis waters. Finally, we often fail to understand what the data is telling us about process capability and the probabilities of achieving future per- formance levels without making signifi- cant process changes. W. Edwards Deming said that "man- agement by results is like driving a car by looking in the rearview mir- ror." Deming was trying to get us to appreciate two things. First, we should use process data instead of outcome data to monitor and im- prove performance. This sounds simple, and yet in most organiza- tions process-level data use is not in the forefront. Perhaps more importantly, stop using your data to look back. Instead, use the vast amounts of data at your disposal to find new relationships between people and processes that can be leveraged to help sustain higher levels of perfor- mance. Who knows – higher levels of process understanding may lead to more effective changes in the future. Kevi cMa us is a performa ce improve- me t coach based i ai ier, Orego , a d a 35-year member of IISE. He has writte workbooks about perso al a d team effective- ess. McMa us is a lum i exami er for the Malcolm Baldrige Natio al Quality Award. Reach him at kevi @greatsystems.com. We should use process data instead of outcome data to monitor and improve performance.