Telephone 888-319-5852 Copyright 2005 by Productivity Press www.productivitypress.

com

Its Not Just the Processes; Metrics Also Make a Difference

Changing from the wrong metrics to the right metrics is just as important as actually improving your processes in any improvement strategy. Executives at Datex-Ohmeda learned that lesson several years ago when they began to implement quick response manufacturing (QRM) at their facility in Madison, Wis. They discovered that some existing metrics actually inhibited efficient operation, while switching to better metrics contributed significantly to increasing production and improving customer satisfaction. But to achieve the benefits, they had to address cultural issues, especially employees who did not want to let go of old ways of doing business. The biggest problem we had was that people thought we needed the old measures for operating reasons, or legal reasons, or regulatory reasons, and what we found out when we got into them was that was not the case, notes Paul Allen, finance director. Every measure we got into where people told us we had to do it, we found out we really didnt. Datex-Ohmeda, part of GE Healthcare, is the dominant player in the manufacture of anesthesia machines, with nearly 75 percent of the U.S. market and about half of the global market. The 1,100-employee facility in Madison makes about 9,000 machines per year, all built to order. That figure has grown significantly in recent years, up from about 3,000 in 1991. Allen explains that an anesthesia machine used to be something a hospital would keep for 20 years, but that changed with the growth of malpractice lawsuits. Now they want the latest technology, Allen comments. What weve done is build machines that are pretty foolproof. Its hard for an anesthesiologist to do something stupid. Now they replace them about every six years. The growth in demand and concern over possible loss of business to faster competitors prompted DatexOhmeda to pursue an improvement strategy. Executives learned of the Center for Quick Response Manufacturing at the University of Wisconsin, and became interested in QRM. That strategy is similar in many ways to lean manufacturing, but focuses more on low-volume, build-to-order operations. I dont think QRM is for everybody, Allen states. If you make two million of the same identical item, I dont know if it will help you a lot. QRM is aimed at an organization that has two characteristics: relatively low volume, like under 20,000, and also where you build to order.

Different Yardsticks

Allen believes metrics drive behavior; good metrics can drive the right behavior, and bad metrics drive counterproductive behavior (see sidebar, page 5). At Datex-Ohmeda, one metric the company discovered

was counterproductive was the number of units going out of the factory, or even going past a particular worker. For example, a worker might produce 15 or 16 units a day, and increased orders would prompt the company to ask the worker to produce 19 or 20. They would think we were doing it simply for our own benefit, and they would then have to do it every day, Allen recalls. We had the union on the side telling some employees, dont do 19. Youll have to be doing 19 forever, and then it will be 23 or 25. The company switched to telling workers the number of orders from customers that were not completed. We found that (workers) were very concerned about customers. Production actually went up in fact, way past what we hoped they would do, says Allen. Another metric Datex-Ohmeda used to use was on-timeto-promise-date, which Allen says is used by many companies. Lower-level employees faced pressure from top management to make sure a high percentage of orders were shipped by the promise date. The result, Allen notes, was that employees who took orders from customers would be so concerned about making sure they met the promise date, they would pad the promise date. They think we can get it out next week, but they would promise three weeks. As a result, our lead times just began to go out the window. (Lead time) used to be six weeks. Now it was 12 weeks. The metric was abandoned in favor of measuring the percentage of time the company can deliver an order by the customers requested date and we even had to be careful on that, because people on the phone can lead a customer, says Allen. He explains that a service representative might tell a customer delivery would take six weeks. The customer might accept that, so six weeks would be listed as the requested date, but in fact, the customer would have liked it earlier. Of course, in addition to changing metrics, DatexOhmeda made actual changes in its manufacturing processes. Two of the most significant changes were the creation of manufacturing cells (in place of a batch-and-queue system) and establishment of some subassembly lines and feeder lines, which helped reduce bottlenecks so you dont have the whole lead time strung out in one line, Allen notes. Improvement in production led to discovery of another problem metric. Datex-Ohmeda installs the machines it manufactures. Once cycle time was reduced, the service engineers who do the installations complained that they could not keep up with the increased pace. In some cases, machines were delivered to customers, then sat for extended periods awaiting installation. That delayed revenues, because customers did not pay for the machines until installation was complete. We found that in customer service, their measure was on how much service revenue they generated not installations. So they always did service first, Allen explains. We made them realize how much that was holding up payments. We were able to get them to change, a little bit at least, their

Telephone 888-319-5852 Copyright 2005 by Productivity Press www.productivitypress.com

Good vs. Bad Metrics

As finance director at Datex-Ohmeda, Paul Allen has developed lists of two categories of metrics, which he labels measures we like and measures we dont like. The measures he likes include: Lead time. Unfulfilled customer orders. On-time to customer request date. Cash-to-cash timing (meaning the time from purchase of materials to payment by the customer). Instances outside of minimum/maximum stocking levels (where suppliers are responsible for stock levels). First-pass rate (of quality). Out-of-box failures (with subcategories, such as need parts and didnt work.) Customer complaints. Warranty cost as a percentage of revenue. Actual material cost per unit (NOT standard cost). Actual expenses vs. budget. Safety index (which measures unsafe situations, not accidents). He also likes a balanced scorecard covering issues of finance, process, customer intimacy and employee development. Measures Allen does not like include: Daily output. On-time to promise date. Anything related to standard cost. Allen is particularly vocal in his opposition to absorption accounting, which involves the calculation of standard cost. He offers two alternative absorption methods: Record the absorption/overhead-application at the time of shipment. Only record the variable costs, excluding fixed cost. As a longer-term alternative, he recommends abandoning the use of standard costs for accounting. He notes that modern computer systems allow quick calculations of actual costs, making standard costs unnecessary. However, he concedes that standard costs may still be useful as goals or targets, and as reference points for analysis. priority and not just be single-minded about trying to get service revenue.

any visibility from the time shipped to the time the customer was using it. As always, culture change was difficult. When we told direct laborers, you build one or two and thats it, they say, what do you want us to do then? We didnt want them to build ahead because they would probably build the wrong thing. We actually had to give them a shopping list of things to do (when they werent building machines) clean up your area, go to training, work on employee safety. This idea of stop, we dont want you to do anymore today was a real counterintuitive thing, and a big culture shock, he recalls. At the same time, he notes, we were very surprised at how much our direct labor force was way ahead of us in terms of being responsive to the customer, and knowing whats right and whats wrong. If you can use their knowledge, they have so much more knowledge than anybody thinks. One example was the need to find ways to staff cells, and schedule overtime, if necessary, when people were out sick or on vacation. Initially, supervisors handled this task, which Allen says didnt work very well. Ultimately, the plant manager told the employees in the cell to work it out themselves. We thought it would be a disaster, Allen admits, but production and quality in that group just skyrocketed. As top managers worked to shorten lead time, one surprise was that our sales and marketing operations didnt see any value at all in shipping earlier, says Allen. He explains that many orders came from hospitals building new facilities, with the order being placed well in advance of when the machine would be needed. However, other customers with existing facilities did want fast delivery. It was surprising to me how much resistance we got out of sales and marketing. But once they did get on board, they got on board in a big way, he comments. For other companies embarking on a similar journey, Allen offers two pieces of advice. First, he urges operations managers to make sure people outside operations are also involved. Second, he stresses that finance people need to really question everything that they hear. You hear something is required. You hear they cant get by in accounting without doing this. You have to challenge all of that.

Takeaways

Data and Culture Shock

One of the challenges involved in switching to new metrics is gathering the data needed for the new measurements. Allen notes, A good example is lead time. We dont have good systems to measure just how long it takes from the time we get the order to the time we ship it. We didnt have

Beliefs that non-productive metrics are required by laws, regulations or operations are usually wrong. Getting people to abandon metrics used for a long time is a significant challenge. New metrics may require new data-gathering processes.