May 31, 2005
Editor's Note: This column by John Allen,
a New Hampshire-based consultant, is the latest in
a series of contributed editorial columns. Readers
interested in authoring future contributed columns
can click here to see the Guidelines for Editorial

Fixing Things:

The best way to measure performance is not just by counting the number of failures or adding up the warranty costs. A better alternative is to measure how the product consumes energy to perform each of its functions, and how successfully it avoids failures caused by energy lost to friction, vibrations, or heat. In short, it's a restatement of the second law of thermodynamics.

by John Allen

The warranty professionals who read Warranty Week generally think in terms of product warranty costs and extended warranty revenue on the income statement. To the consumers who for the most part don't read it, however, warranty is all about the last time they tried to return a broken computer, explain why their car won't run properly to a doubtful service technician, or wonder why they are being coaxed into buying an extended warranty for a washing machine when the commercial on television touted its quality.

One group thinks about cost and how to make money on stuff that may not work right, while the other thinks about how to make sure it works without paying extra.

Sometimes a company measures overall success without enough emphasis on product performance, leaning a little too much on the income statement's measure of its profitability. Such a company will blissfully sell products it is liable to later repair under warranty. However, it will discover no liability until warranty costs escalate. Then it will fear there are a lot more products of the same vintage, also at risk of failure. If enough customers are annoyed to the point where a story shows up in the New York Times, the company might decide to start talking to the engineers about fixing things.

A February 16, 2005 headline in The New York Times, "U.S. Opens Safety Inquiry on Lexus S.U.V.," in which Toyota said it knew about brake problems in the RX330 sport utility vehicle but did not notify customers or recall the vehicle, must have received some uncomfortable attention at multiple levels of the company.

Those who read the income statement might tell the engineers what to work on, how to do it, what the outcome should be, and when to finish the job. The trouble is, the engineers can't do a thing about the stuff already sold that is going to cause a lot of heartburn.

Two Kinds of People

This makes me think of the dialog in the movie "The Good, the Bad, and the Ugly," when the "man with no name" said, "There are two kinds of people in this world, my friend. Those with loaded guns" -- and we hear the loud click as he pulls back the hammer on his big Colt .45 -- "and those who dig" as he tosses the shovel to Tuco, who is looking down the barrel, finishing, "You dig."

The technical guys generally feel they are doing all the digging, with the cost guys holding the gun. To the technical guys, it is just a matter of who is under the gun on any given day, and the propensity of the armed to be trigger-happy.

I digress. My wife is my editor. As she read what I have written so far, she reminded me of the Jenn-Air kitchen appliances we bought because of the name. The igniter on the cooktop makes an annoying clicking noise after the flame is lighted. We had technicians investigate several times, while the cost was covered under warranty. One day, I happened to get a seat on an airplane next to a vice-president of Jenn-Air. I told him about our problem and he gave me a special number to call. "How lucky," I thought, until I found it was the same number we had been calling for service all along. Now we live with the annoying clicking. We would rather junk the cooktop and get a new one than put up with hours on the phone explaining the same thing over and over again in order to fix it. My wife wants a new cooktop, probably not a Jenn-Air. Jenn-Air will never know. (Today I might have better luck by going to

I suspect we did not buy the only cooktop that had faulty igniters. It might have been a rare enough problem that the service technician had never seen it before and might never see it again. Maybe he filled out a report that went to headquarters. Maybe it said, "No Trouble Found." He might have described it in such a way that it was compiled as one of the "trivial many" on a summary report.

Some common denominator in a disciplined warranty analysis might have given the engineers back at the headquarters a starting point. Often, a warranty report is a summary of whom to charge for what, and is of little help to those who must figure out what to work on.

No Trouble Found

Conflict can occur when warranty, reliability, and product performance are not strategically differentiated by proper analysis of warranty, returned goods, and field parts that have not failed in order to identify product performance and reliability problems. Many think they are doing a good job with all this data collection. Churning out reams of data and measuring thousands of field returns typically results in the most common field failure: "No Trouble Found" or NTF.

I prefer a little bit of good information instead of a lot of data, given that the information helps make good choices. The challenge is to get good information and avoid massive data collection. Data collection has become too easy. Don't let it stand in the way of finding good information. If you are swamped with data and uncomfortable with the outcome, you might want to think about making a change.

Warranty is an accounting metric. It tallies the cost of product failure, and/or the revenue and expense of extended warranties. Quality has come to mean just about anything these days. At least one company defines quality as "Customer Enthusiasm." That might work for the marketing department, but it won't do the engineers much good. They need a quantifiable metric in engineering units. A common definition of reliability is the probability that a system will perform its intended function over given circumstances for a specified period of time. That is a good starting point. It might help determine if one can bear the cost of a ten-year limited warranty on the powertrain of a line of vehicles or if they should cut back to five years/50,000 miles. It may tell the Director of Engineering that he has work to do, but by itself, leaves him in the dark as to what to work on.

In the February 15, 2005 issue of Warranty Week, the feature article describes the work of David Bien, Director of Reliability at Sub-Zero Freezer Company, and the SAS Institute Inc., to develop a software package to analyze field failures in conjunction with the engineering staff in order to identify opportunities for improvement. Maybe Jenn-Air has a similar approach by now, but Sub-Zero's approach sounds encouraging. And the Director of Reliability is in charge of figuring out what to work on. The next step is to convert every problem into a technical project, a process described in my article "The Model for Solving Technical Problems," posted at

Consuming Energy

I would say that if the design is sound, and important features and properties are effectively controlled in the manufacturing process, then the products should perform at the point of sale. I submit that the best way to measure performance is not just by counting the number of failures or adding up warranty costs. My preference is to think in terms of how well the product consumes energy to perform each of its functions.

One of my engineering friends said to me one day, "John, I can build a machine that won't break. I cannot build one that won't wear out." If a machine wears out, it can wear out only in the course of performing one of seven functions, which I use as a starting point for nearly every technical investigation.

Most engineers would recognize this as a restatement of the second law of thermodynamics. Stated simply, if you want a machine to do some useful work, (one of the seven functions) you have to have an energy supply. The energy you pay for and the useful work can be measured in common units. Joules work nicely. Once you compare the numbers, the useful work is always less than the energy you paid for. The law says that some energy is going to be lost and you don't get to change the law! The lost energy results in things that the engineers would rather not have happen, like heat, friction, vibration, etc.

These are the reasons machines wear out and annoy customers. If we could get the work out equal to the energy in, we could theoretically build machines that never wear out and never break. That being said, it follows that the best way to measure performance is to measure how much energy is consumed in doing useful work and comparing it to how much energy you paid for. Reduce the wasted energy, and machines will last longer. Once again, it's the law! I have demonstrated this approach and use it almost exclusively in my approach to helping clients solve technical problems around the world.

Preliminary Investigation

Danny Hakim, the reporter who broke the Lexus brake story in The New York Times, wrote: "The company has not notified customers because it says it does not think the problem is widespread and no injuries have been linked to it... The government inquiry was prompted by complaints from 10 Lexus owners... The investigation is in the preliminary phase. It will become more serious if enough evidence of a potential defect is found... A technical service bulletin sent to dealers recommended replacing the vehicles booster if the problem occurred."

Looking at warranty data at this point of the investigation would be a waste of time. Brainstorming what might be wrong is of similar low value. An investigation which begins by asking which of the seven machine functions is failing, followed by building a function model of how the system is supposed to work based upon the second law of thermodynamics will get the investigation past the preliminary stage. Further investigation will reveal what matters, and get it fixed based upon the principle of delivering energy to create useful work -- not sitting around comparing measured parts to the print in order to figure out who is to blame. Engineers respond well to the model of delivering energy to create useful work. It is intuitive and is based upon what they were trained to do.

There have been billions of dollars spent on improving warranty and quality over the last 20 to 25 years. Nearly every company has a quality program based upon some model of improvement, likely copied from the guy down the street. Sure, there have been improvements, but I am not convinced that there was a fair return on investment. I suggest a far more effective and simple approach is to just start with the second law of thermodynamics. Save the money you spend on consensus management and team building, hoping it will result in better quality. Spend some of it on good science.

Now we have a new starting point: performance is defined as how well a machine provides its independent functions at the time of delivery; and reliability is defined as how long a machine performs its independent functions well. The units will be in terms of work or energy, a system based not solely on counting broken parts or computing the probability of system failure.

A more effective strategic approach should include the following:

  • Are our products breaking or wearing out? There are independent reasons for each. (By the way, more often than not, warranty costs include more failures from machines that break than those that wear out.)
  • Which of the seven functions are the most common failures? Have we already engineered our way through this in another area and can we emulate (copy the answer) instead of invent our way out of this?
  • How do we make the warranty system support engineering and vice-versa, making warranty data part of a system?

John Allen

John Allen, based on Rye, New Hampshire, specializes in the application of engineering principles within the context of product performance, product durability, and reliability improvement in manufacturing companies. John's clients have included General Motors, Pratt & Whitney, General Electric, Rolls Royce, US Steel, Delphi, Phillips Consumer Electronics, Siemens Electronics and others. He has worked in foundries and chemical companies, and on precision machining, plastic molding, precision assembly and extrusion and blown film.

For years, Allen taught Shainin Statistical Engineering seminars, was an annual presenter at the Shainin Symposium, and was a keynote speaker at the Daimler Chrysler 2003 Black Belt Symposium. John spoke at a conference of West Coast Universities, delivering a speech, "Emulation and Invention: Strategy, Structure and Tactics." He was recognized in January 2002 by Shainin LLC as the first Shainin Fellow for his contributions as Director of Technology, for his ability to teach clients, for solving technical problems and for training consultants.

John's reputation is based upon his ability to efficiently assist clients in solving complex technical problems. There are three projects that solidified his reputation. The first involved defects in cast aluminum heads. Scrap was in excess of 50%, much of it not found until after machining or installation in the vehicle, causing assembly line stops and warranty problems. John's small team reduced the scrap to less than 1%, eliminated warranty claims for the defect, and virtually eliminated machining scrap. The project required only several days of John's time over the course of a few weeks.

The second project was at a smelter, which suffered from broken furnace electrodes for more than 50 years. The life of the electrodes was unpredictable, requiring frequent furnace outages. Over the course of one year (it took six to eight months to manufacture and consume an electrode) the life of the best electrodes was extended by 40%, broken electrodes were eliminated, and this type of furnace became one of the most productive in the world. The additional cost to produce these efficient electrodes rose by about $3 from a base price in excess of $5,000.

The third project looked at the air seals in aircraft engines. Previously, another team, over the course of two years, reduced scrap from 40% to 20% and stated that no further progress could be expected. Then John and his small team took over. At the end of the project, scrap was less than 1%.

John can be reached at +1 (603) 436-4495 or at

NEW Customer Service Companies
SAS eNews ad
ServiceBench for Service Administrators
4CS iWarranty
Sign up for a free subscription to Warranty Week:
     subscribe     change of address     unsubscribe
Related Articles From Warranty Week