Why maintenance planning’s most popular KPI is an awful metric
The most popular KPI for maintenance planning, Plan vs. Actual, is a counterproductive KPI. It usually worsens planning performance (and consequently plant performance).
Doc Palmer is the author of McGraw-Hill’s Maintenance Planning and Scheduling Handbook and helps companies worldwide with planning and scheduling success. Visit www.palmerplanning.com or email [email protected].
What is Plan vs. Actual (PvA)? Plan vs. Actual (also called planner accuracy) is the most commonly used KPI to measure planner performance. It is the comparison of the planner’s labor hours estimate to the actual labor hours charged to a job. For example, a planner estimates a job should take three hours, but it actually takes six hours. The PvA would be 3/6 = 50%. (Some people measure it as 6/3 = 200% and interpret it accordingly.) But what does the score mean? Who is “right?” Is the planner right or is the craftsperson right? The planner is making an assessment without actually dissembling anything. Will there be any discovery work after the craftsperson dissembles the equipment? The planner can make an educated guess, perhaps aided by history and PdM input, but maintenance (especially corrective maintenance) does not have the frequent repetitions and minimal variations of factory assembly-line work.
Furthermore, the planner has no control over who will be assigned the work in the field. Will the supervisor assign the slowest person or the fastest person? Should the supervisor try to speed up a slower person? Should the supervisor try to slow down a faster person? Will the supervisor assign the least-experienced person or the most-experienced person? Should the supervisor give the least-experienced person extra time or more help? Was it cold or raining that day?
In real life, maintenance actual field times vary widely. The above example of 50% was for a single job. More commonly, the PvA compares the sum of all the estimated labor hours for a period (usually a month) with the sum of all of the actual labor hours charged to those jobs. Some companies instead measure the “absolute value” of the difference of individual jobs. (Also, as a side note: PvA is not a “labor utilization” KPI, which is a percentage of paid or worked hours that are actually charged to work orders. That’s another KPI altogether.) But management, as usual, wanting a school grade of 90%–100%, deems the planner with a 50% score to have poor performance.
The math: Let’s say a supervisor assigns a 10-hour shift to a craftsperson with three jobs estimated to take five, three, and two hours. The craftsperson charges two hours, six hours, and one hour, respectively, to the jobs. The total planned hours are 10 and the actual hours are 9. The PvA is 111%. It looks like the planner is fairly “accurate.” But the planner, being concerned about a “score,” might change the three-hour job estimate to six hours in the future. So the next round of estimates is five, six, and two hours. The actuals this time are five, six, and two. The PvA is now 100% and it looks like the planner is even more “accurate.” The truth, though, is that this practice is slowly increasing estimates over time in an unusual fulfillment of Parkinson’s Law.
Parkinson’s Law states, “The amount of work assigned will expand to fill the time available.” (Cyril Parkinson, “The Economist,” Nov. 19, 1955.) Because of Parkinson’s Law, craftspersons are more likely to take extra time allotted to individual work orders than to finish them early. In addition, craftspersons are more likely to charge the full time to the job merely knowing that planners are being scored on accuracy. Furthermore, planners are more likely to increase estimates based on actuals than to decrease them. And finally, planners are likely to get better PvA scores by generally giving extra time on jobs.
In the above example, say that the five-, three-, and two-hour estimates (10 total hours) were in fact very reasonable. But the actuals with the PvA scoring system encouraged planners to raise the estimates to five, six, and two hours (13 hours). Afterward, keeping the estimates at 13 hours means the plant continues to spend 13 hours of actual labor to do 10 hours’ worth of work.
This extra labor spent is grossly counterproductive considering that the much larger purpose of planning is to help maintenance increase its productivity and work quality. Maintenance productivity, reflected in work orders completed per month, suffers because an extra three hours of labor is not spent on extra work. And maintenance quality, reflected in MTBF (mean time between failures) or some similar quality KPI, suffers because the plant would have spent those three hours on proactive work worth 10 times the value to the company’s bottom line. (Extra work is always proactive because a plant will always cover its reactive work. So by definition, any extra work is proactive.) But the plant in this example just did not have those extra three hours to grease a minor bearing squeak, adjust a coupling, or check an oil level.
This article is part of our monthly Palmer's Planning Corner column. Read more from Doc Palmer.
Why is PvA so popular? For one thing, it does seem to make sense. For another thing, it is hard to come up with a list of appropriate planner performance KPIs. Nevertheless, it is worse than not having any planner KPIs. PvA is counterproductive because it encourages planners to inflate estimates to get better scores, which leads to less work being scheduled.
What are some good KPIs for planners? Forget KPIs for a minute. Consider hiring qualified planners, persons with good data-organizing skills, good communication skills, and preferably good craft skills, as well. Nevertheless, the KPIs I like for planners are mainly: Are they using plans from the CMMS job plan module and not planning each job directly on the work order? Do they have a minimal unplanned backlog? These two KPIs help ensure that the previously selected qualified planners are running the Deming Cycle to plan enough of the plant’s work to improve quality and planning enough of the plant’s work to supply the scheduling program to increase productivity. If you must use PvA, keep it in the background and use it as a third reference if you are checking overall performance – certainly not to grade the planners’ performance.
Planned versus Actual is a terrible KPI for planners. Be mindful of Dr. W. Edwards Deming’s insight that “a bad system will beat a good person every time.” I also remember another good piece of advice once shared with me: “Be careful to mind the store and not just mind the score.”