How to Reduce Warranty Costs Using Tools That Predict Product and Process Performance

April 26, 2019
8 min read

Looking just at the auto industry, GM has recalled roughly 3 million vehicles this year in the U.S. alone. In 2009, GM recalled roughly 2.2 million vehicles.

The burden of warranty costs

Recent news offers a reminder that huge opportunities still exist to reduce warranty costs that drain manufacturers’ bottom lines.

Consider that during the week of June 7, Chrysler announced recalls of 35,000 Dodge Calibers and Jeep Compass vehicles for potential accelerator pedal problems, 289,000 Jeep Wranglers for potential brake fluid leaks, and 285,000 Dodge Caravan and Chrysler Town & Country minivans for a sliding-door hinge problem that could potentially wear through wire insulation, short circuit and lead to a fire.

WinSPC Means Lower Costs and Higher Quality

WinSPC is software to help manufacturers create the highest quality product for the lowest possible cost. You can learn more here.

A few days later, General Motors announced a recall of more than 1.5 million cars and trucks because of a defect in the windshield wiper fluid system which could result in a fire (two-thirds of these vehicles were recalled in 2008 for the same problem).

Looking just at the auto industry, GM has recalled roughly 3 million vehicles this year in the U.S. alone. In 2009, GM recalled roughly 2.2 million vehicles. Last year, Ford recalled 4.5 million vehicles due to a defective switch that could also lead to fires. Earlier this year Honda recalled over 400,000 vehicles for a brake pedal problem. And of course, the extent of Toyota’s accelerator pedal problems have been widely publicized.

These numerous recalls translate into a heavy burden on corporate earnings. It is difficult to accurately estimate the total cost of the thousands of vehicle recalls occurring on an annual basis. There are over 76,000 vehicle recall records in NHTSA’s database covering the years of 1966–2008. The largest 10 recalls included 55.5 million vehicles. If dealers were paid $100 per repaired vehicle, then the 10 largest recalls would have cost nearly $5.6 Billion dollars. In 2008 alone, the U.S. auto industry spent an estimated $14.5 Billion on warranty costs.

Of course, the burden of substantial warranty costs is not limited to the auto industry. Recall costs for consumer products in the U.S. excluding automotive recalls are estimated at more than $700 billion annually (according to the Consumer Product Safety Commission). Warranty costs of large U.S. manufacturers typically average 2 percent of revenue. So, for every $1 billion in revenue, a company spends a daunting $20 million in warranty expenses.

One clear trend is a rise in recalls among drug manufacturers. There were 1742 drug recalls in 2009 as compared to 338 a decade earlier. Of recent note, Johnson & Johnson shut down a manufacturing facility and recalled 136 million bottles of Benadryl, Motrin, Tylenol, and Zyrtec for children due to contamination and too-strong dosage levels. Additionally, Merck, Genzyme, and GlaxoSmithKline are all struggling with quality issues in vaccine production leading to many shortages.

In addition to direct warranty costs, the costs of lawsuits associated with any one recall can be oppressive. Each lawsuit typically costs a manufacturer millions of dollars in legal fees and losses, and in some cases, tens or hundreds of millions.

The cost most difficult to quantify is the loss in goodwill and repeat business. offers a great venue for customers to review electronics after purchasing them. The customers’ remarks on this site are often damaging to a company’s brand. Nearly every complaint concerns product failure due to flaws in the design or manufacture of the product. Other potential customers listen closely to these customers’ complaints.

Clearly, huge opportunities exist to increase profitability and customer satisfaction, if the root causes of reliability problems can be identified and addressed.

The Manufacturing Problem—Product and Process Failures

In its simplest description, manufacturing plants typically take some raw materials (steel, plastic, powders, etc.), process them (heat, press, stamp, mold, form, etc.), and then assemble different pieces together. The parts produced by this manufacturing process are supposed to be identical. However, that outcome remains theoretical as no two parts are manufactured exactly the same. Variation exists in every manufacturing process; it is a natural occurrence, although many manufacturers don’t appreciate the impact of this fact.

Variation from part to part sometimes means that parts don’t conform to customer requirements. Everyone understands this principal and the scrap or rework it produces. But equally important (and far more insidious), are parts that do conform to specifications but still vary in their performance. Although these parts are within specifications and considered “good,” the variation within specification can still lead to product failure in the field and warranty costs for the manufacturer.

This concept is easily understood with a practical example. If every part of the 2002 Model X washing machine were made identically, then each would fail with the same amount of usage. But actual usage shows that they fail at dramatically different times–some fail early and some last for a long time. This is largely due to variation within the manufacturing process.

There are three primary reasons for product failures:
1. Inadequate engineering (design shortcomings)
2. Variation in production (so parts perform differently for consumers)
3. Customer abuse (misuse of a product)

Based on observation over years of analyzing product field failure, it is clear that the first two reasons are far more prevalent. But these two reasons are also the most amenable to improvement. The fields of applied mathematics and statistics are well-armed with quantitative methods available for predicting product performance and for predicting the output from manufacturing processes.

If a manufacturer can predict performance, then that manufacturer can also prevent poor performance. Unfortunately, manufacturers needlessly suffer from failed new product launches, high internal scrap, and premature failures in the marketplace because they are not taking the necessary steps to predict performance and prevent field failure.

Best Practices for Reducing Warranty Costs

Product Performance

Even a modest investment in developing physical models can have a significant positive impact on product performance. Physical models are equations developed from actual data (collected under a variety of strategically determined conditions/scenarios). Collecting physical data from testing can be expensive, but there are many optimization methods for minimizing the amount of data required while maximizing the amount of information produced.

Physical models can predict product performance under infinite sets of conditions, so engineers understand how their products will perform in a wide variety of environments (i.e., identify robustness). Furthermore, the predictive models can uncover design flaws quickly, so design adjustments can be made before a failure-prone product enters the marketplace.

Since many problems only manifest with product usage over time, reliability methods are critical. But this crucial testing is often shortchanged. With ever shorter product development cycles, reliability testing time becomes constrained. Luckily, techniques such as accelerated life testing and degradation modeling provide an efficient and effective means for predicting product reliability over the useful life of aproduct.

Process Performance

As mentioned above, it is impossible to completely eliminate variation in a manufacturing process.

However, useful tools exist to detect variation, identify its root causes and mitigate it. The cost of assessing performance through 100 percent inspection is usually prohibitive, so the most common and cost effective means to predict process performance is through the use of Statistical Process Control (SPC). Effective SPC requires the right chart, the right sample size, and the right sampling plan performed on the right characteristics. And the SPC data should be collected and utilized on a real-time basis, not stored and underutilized. A plant without proper SPC is a plant that reacts to problems—rather than preventing them.

(Click here for a discussion of common misapplications of SPC—and the consequences.)

In addition to monitoring individual key process characteristics for potentially harmful changes, plant-wide production data can be monitored and mined in real-time. Opportunities for the largest cost savings (through variation reduction) may be easily identified. And…variation reduction will only have a positive effect on warranty levels.


Costs due to poor quality and reliability (warranty, recalls, scrap, customer dissatisfaction) continue to plague many companies and present a huge opportunity for improvement. Appropriate quantitative methods for addressing the most common root causes of product failures (inadequate design and excessive variation) must be more widely understood and adopted. Product designs must be verified with physical testing that yield models that predict product performance and reliability under a range of operating conditions. Real-time SPC must be implemented on key process characteristics that are critical to achieving successful outcomes.

Steven Wachs, Principal Statistician
Integral Concepts, Inc.

Integral Concepts provides consulting services and training in the application of quantitative methods to understand, predict, and optimize product designs, manufacturing operations, and product reliability.