April 19, 2005

Warranty Conference,
Part Seven:

Warranty fraud is a big problem for both manufacturers and warranty administrators. What's surprising is how deliberate and organized their schemes have become.


When the Warranty Chain Management conference kicked off last month in San Francisco, one of the keynote speakers noted how he had started out thinking that warranty fraud was a small problem -- a series of gray areas and exaggerations that didn't amount to much. It took some convincing, but he said he was now certain that warranty fraud was not only a big problem, but was something practiced by organized criminals.

What was even more shocking was the source of most of the warranty fraud. It wasn't the buyers, nor the sellers. And it certainly wasn't the manufacturers. Instead, it was the service providers. See that lonely Maytag repairman? He's claiming to have fixed appliances that don't even exist. You know that guy who came to pick up your broken printer? He claims to have fixed every electronic gadget in your house.

During a panel discussion held the morning of March 3, William Fung, HP's warranty process and systems manager, suggested that the warranty fraud problem was more widespread, and its perpetrators more organized, than perhaps most people realize. It's not just customers looking for free service the day after their warranty runs out, or technicians doing unnecessary repairs to drive up service revenue. What Fung described were enterprises whose core activity seemed to be warranty fraud, and whose impact ran into the millions of dollars per company per year.

Measuring Warranty Fraud

Fung said the Association of Certified Fraud Examiners estimated that in 2004, an average of 6% of a given company's revenue was consumed by some type of fraud or abuse, warranty or otherwise. BJSI and Asset Management Solutions narrowed it down even further, estimating in a 1997 study that warranty fraud cost the high-tech industry $1.9 billion in 1996 and would cost $3.4 billion by 2000.

Internally, Fung said, HP has more recently confirmed that 6% to 8% of its warranty claims are fraudulent, lending some weight to the findings of those external studies. To put that in perspective, for HP alone that equates to an estimated $142 to $189 million in warranty fraud per year, and is about the same as the profits generated by the sale of 15 million printers. Or to put it another way, HP would have to sell another 15 million printers per year to make up for warranty fraud losses.

However, Fung noted that warranty fraud is really nothing new. He said that while preparing for his WCM presentation, he looked back at some old reports, and even in 1996 and 2000, fraud levels were being estimated at 6% to 8%. Why hasn't anything changed since then? That has a lot to do with the scale of the job, he said.

"There are issues, and there are challenges that are very big," Fung said. Success is by no means assured. So the effort could end up raising costs without reducing fraud. There's a very strong temptation to let things be -- to continue with business as usual and to absorb fraud losses as a cost of doing business.

In addition, efforts to combat warranty fraud will need company-wide and high-level sponsorship, Fung said. Such a sponsor may require more concrete predictions of the outcome than are possible with such a "fishing expedition," he added.

Fung said the perpetrators tend to fall into three distinct levels of aggressiveness in terms of the way they capitalize upon warranty system weaknesses:

  • Level 1. Opportunistic Fraud: The general population of employees and partners will fall prey to opportunities created by the perception of weak control environments. The acts are sporadic and segmented. These frauds do not have a significant financial impact and would likely remain under the threshold of fraud detection tools.

  • Level 2. The Slippery Slope: Warranty program abuses become institutionalized when the perception of being caught is low. Examples include: Certified technicians vs. uncertified technicians; a 2 part call booked as 2 separate events. As these frauds increase in scope and frequency, the financial impact becomes significant and they become more detectable.

  • Level 3. Systematic Fraud: Certain opportunists target companies with sophisticated and systematic frauds and abuses to maximize profits and satisfy their established customers - both legit and illicit. These frauds have a significant impact and should be detected by fraud reduction tools.

The ones that are really dangerous, he said, are the ones who commit systematic fraud. "Those are companies or individuals whose whole purpose is to commit fraudulent activities. They understand the company's process, and they understand the loopholes, and their whole purpose is to attack those loopholes and make a profit out of it."

For instance, they may know how to time claims so they don't arouse suspicion. They may understand that the only real piece of information they need to get a claim paid is the serial number. They look for what Fung called "leakages" in the system, and once they find one, they keep using it.

Patterns of Fraud

One authorized service provider, Fung said, submitted warranty claims using serial numbers they secretly gathered during legitimate customer site visits. Over a 24-month period, they frequently re-used those serial numbers to submit additional bogus claims. HP estimates it lost $1 million to this company's fraudulent activities.

Another authorized service provider would swap refurbished printers for new units, and would then return the new units to HP as warranty claims. "They would actually print their own proof of purchase," Fung noted.

Fung also said some authorized service providers seem to have figured out which components aren't tracked alongside their system unit by serial number. So, for instance, they might say they replaced a hard drive, and they may return a broken hard drive while making a warranty claim, but it might not be the hard drive that was in that unit at the time of purchase. But that seems more like a Level 1 or Level 2 type of abuse.

Another pair of companies sent staff into computer retail storefronts in search of floor models from which they could copy down the serial numbers. Worse, Fung said, each seemed to share the serial numbers they gathered with the other company. Over a span of 12 months, these scammers cost HP an estimated $2 million.

One time, Fung rode along with an investigator who was looking at an HP authorized service provider in California who had sent in 3,000 warranty claims the previous year. One would think that a shop with that much warranty work would be very busy, but it was dead quiet outside. They didn't go in, though. Instead, they drove to the listed addresses of a few of the shop's largest customers. And they were all in shopping malls, in those mailbox rental places. In other words, both the service provider and the customers were bogus, existing for no other purpose besides warranty fraud.

When he returned to the office, he sounded the alarm. Warranty fraud wasn't all what he'd call Level 1 or Level 2, which would be somewhat easy to dismiss as a cost of doing business and a cost of keeping both customers and technicians happy. No, there were racketeers and corrupt organizations out there whose primary function seemed to be warranty fraud. It surprised not only Fung, but also some of HP's top executives.

The impact of warranty fraud spreads beyond just the manufacturer. Fung cited a Rand study completed in 1999 which found that every $1.00 in warranty fraud actually costs the victim company $2.80, because of secondary losses. For instance, stolen parts may be resold, costing a company once for the initial fraud and again for the lost parts sale. And then there's the cost of investigation, additional security, and insurance, plus the cost of reduced (legitimate) customer satisfaction. Customers also pay for fraud -- by Rand's reckoning an additional $2.40 through higher prices and other steps taken to offset fraud losses. And then the industry as a whole suffers an additional $1.00 loss through reduced revenue and other factors. So according to the Rand study, every $1.00 in warranty fraud is multiplied by 6.2.

Proactive Fraud Reduction

To prevent warranty fraud, Fung said companies need an end-to-end fraud reduction solution that integrates the business policy, processes and information technology solutions and utilizes proactive monitoring tools. This would allow companies to balance the goals of a predictable positive experience for customers with the requirements for aggressive controls and boundaries that seek to eliminate fraud and abuse. Key characteristics of proactive fraud reduction include:

  • Well-defined entitlement policy and process: Establish and enforce warranty terms that are clear and simple. While warranty entitlement must first serve the business requirements and customer demands to support the value proposition, all warranty terms must be verifiable and able to be implemented.

  • Well-defined authentication process: Develop unique tamper-proof and reliable identification that will designate key parts and components as genuine.

  • Accurate and consistent warranty entitlement: Uniquely identify products, embedded components and date of sale. The information must be recorded and maintained to facilitate claims decisions. Customer-facing tools should be secure and ensure consistent entitlement response occurs regardless whether claim is submitted directly or through authorized service providers.

  • Closed loop RMA tracking and validation: Establish vigorous entitlement validation at customer touch points such as the call centers and implement inspection processes at return centers for the validation of the physical parts and confirmation of warranty entitlement.

  • Realistic warranty terms and conditions: Ensure terms and conditions can be adequately fulfilled by the service delivery organization or authorized service providers. If terms and conditions are beyond the capability of our delivery partners, it essentially opens up an opportunity for fraudulent activities.

  • Pro-active fraud detection procedures and tools: Establish processes that utilize specially-tailored automated reporting tools to identify and report anomalies and other activities that are potentially problematic, such as repetitive submission of claims with the same serial number.

  • Strong warranty compliance and investigations: Implement adequate controls and the means to enforce them. Policies and contracts must consistently reference and enforce the entitlement terms and conditions.

Same Story in Major Appliances

John Estrada, a co-founder and the chief operating officer of ServiceBench Inc., turned the focus from computers to major appliances. Surprisingly, the pattern of fraud and abuse seemed quite similar, despite the differences in product lines.

Estrada said he has doubts about the accuracy of warranty fraud loss estimates, because if it could be accurately measured, it could be eliminated. He said he's seen estimates of fraud's impact being as high as 10% to 15%. But assuming that 6% to 8% is close to accurate, and assuming that product warranty claims and extended warranty premiums are together around $39 billion, and you're talking about a $2.3 to $3.1 billion problem for manufacturers and claims administrators.

Estrada said research into the nature of fraud reveals that four behavioral ingredients are needed to facilitate fraud. As outlined in the Dec. 2004 edition of the CPA Journal, in an article entitled "The Fraud Diamond: Considering the Four Elements of Fraud," these ingredients include:

  • Incentive: I want to, or have a need to, commit fraud.

  • Opportunity: There is a weakness in the system that the right person could exploit. Fraud is possible.

  • Rationalization: I have convinced myself that this fraudulent behavior is worth the risks.

  • Capability: I have the necessary traits and abilities to be the right person to pull it off. I have recognized this particular fraud opportunity and can turn it into reality.

    Occasionally, ServiceBench will come across instances of blatant warranty fraud where there really isn't any possible defense. "We've seen cases where for some of our clients, some of their service partners will go into a home, write down the serial numbers, and submit claims against those," he said. "We've seen cases where people are adding parts that weren't actually part of the repair."

    One time, he said, an appliance manufacturing executive got a phone call from his own company's warranty administrator, asking how his recent repair went. It turned out that a technician had recently been in the executive's home to fix a completely different product made by a completely different company, and had copied down some extra serial numbers and had submitted claims for those also. So what started as a courtesy call from a warranty administrator quickly escalated into a fraud investigation.

    Fixing Everything in Sight

    Another time, a technician was called to a retailer's storefront to fix a single broken display unit. But the warranty claim he submitted alleged that he'd fixed all the units the retailer had in stock. Estrada said that since that happened, some of his clients have insisted that all stock repairs must be pre-authorized.

    Then there are the instances where technicians try to take advantage of a loophole. For instance, a client may have checks in place that look for duplicate claims within a rolling 45-day window. So a dishonest repairman who knows this might wait 46 days and resubmit the same batch of claims again.

    And then there are the more subtle instances of fraud, such as when a service technician arrives at a location and finds that there's nothing wrong with the unit. Maybe the plug was loose, or maybe the customer had no idea how the unit worked. So the whole service call was a big waste of time that should have been handled over the phone. But who's going to pay for the waste of time? Maybe the technician will find something wrong that merits a claim after all.

    Despite all the anecdotes to the contrary, Estrada noted that the vast majority of the 18,000 service technicians who do occasional work for ServiceBench and its clients are very honest and hard-working people who want to operate their businesses in both a profitable and legal manner. "I think it's the exception rather than the rule," he said, and warranty administrators have to keep this in mind, lest they make it too difficult for all the honest technicians to process their paperwork in an effort to catch the crooks. And you don't want to spend $10 to save $5.

    Estrada said that fraud detection is best done either up front as a claim is submitted, or after the warranty work has been completed. Detecting and denying it up front has the advantage of avoiding the chore of paying a claim and then trying to recover the funds. But doing it after the fact has the advantage of allowing for a more thorough investigation, because at least in theory the unit is back in service by then and the customer is once again happy.

    He mentioned customer surveys as a good way to keep tabs on technicians, not only as a way to combat fraud, but also to measure customer satisfaction. Or more precisely, a survey that on the surface appears to be measuring customer satisfaction is subtly asking questions such as "was this work actually done?" If the survey comes back "return to sender, address unknown," that in itself is a good indicator that something is amiss.

    As with the executive who was puzzled by the courtesy call from his own warranty administrator, a survey asking about a product that was never repaired is bound to raise suspicions. Perhaps a survey could ask who paid, to detect instances where the technician was trying to bill both the consumer for the service call and the manufacturer for the warranty claim. Or perhaps a survey could ask how many trips it took to repair a given unit, looking for discrepancies in the labor portion of the claim.

    Another post-processing approach would use business intelligence software to spot patterns or anomalies in warranty claims, looking for indications of fraud. For instance, why are there so many problems with ice makers in one particular city? Is it a problem with the ice makers, or with the service provider in that city? Similarly, why are so many of a particular service provider's warranty claims for labor only? That could be a training issue, or it could be an indicator of fraud.

    ServiceBench Analytics

    To illustrate how this works, Estrada showed the audience screen shots from a dashboard control system available to ServiceBench clients. One graph plotted the number of claims per service provider during a given month against the percentage of claims that were labor only. For all service providers, the average was around 23% of claims being for labor only. And the average number of claims was around 40 per service provider per month. But as the graph below shows, there were five or six dots far above the average for number of claims, and two of those also were far above the average for percent labor only. Those are the service providers who merit a closer look.





    System Snapshots

    After this panel discussion, combined with the previous day's detailing of claims exaggeration problems in the automotive industry, audience members seemed to be mostly convinced that warranty fraud was a big problem in several industries. Aaron Nielsen, president of BlueRealm Solutions Inc., said it seemed to him to be one of the major themes of the whole conference. But even he was somewhat taken aback at how organized and deliberate some of the fraud efforts have turned out to be.

    "With the kinds of dollars that are attached to these claims, it does seem like an area that could be vulnerable to be exploited by organized crime," he said. "I was surprised to hear companies are believing it's at that level already. But if you step back and look at it, certainly it has the potential to be susceptible."

    Nielsen said most of the warranty fraud detection tools he's seen are focused on systematic fraud. They look for patterns or anomalies, for instance spotting a service provider who's billing for 40% more parts than any of the others. But they don't do much to look for the one-off sorts of fraud -- what Estrada called subtle and exploitive fraud and what Fung assigned to Levels 1 & 2.

    "There's a trend now towards these independent service providers as opposed to a couple of large nationals," Nielsen said. "So with these one-off guys, I think you're more vulnerable. These individuals could try to inflate their prices, or bill you for additional parts, or bill you for parts that weren't replaced."

    BlueRealm's software is designed to take "before" and "after" snapshots of a computer's configuration, which a warranty issuer can then compare to an invoice. If the warranty claim says a CD-ROM drive was replaced, that would be confirmed by the changes in the "after" snapshot. And it would confirm that the right part was used.

    "By capturing that information, we can help expedite technical support process, or do parts determination reports," Nielsen said. "There are ten different things we can do with that information, but one of them is to be able to audit technicians."

    Somebody's Watching

    The question, of course, is how hard do you push? Do you want to catch the guys who use four bolts but say they used five? Or do you want to catch the criminal enterprises who use no bolts but say they used hundreds? Estrada noted that the definition of integrity is doing the right thing even when nobody's watching. But sometimes, the integrity-deficient need to be watched. So perhaps the best solution of all would be to make it seem as if somebody is always watching, even when they're not.

    Perhaps somewhere out there, a dishonest repairman is reading Warranty Week and is thinking that maybe copying down serial numbers from display units isn't such a good idea. Maybe they're thinking that their scams which use bogus customer addresses will be defeated by something as simple as junk mailings. Perhaps they'll worry that their fraudulent behavior will be flagged by a graph in some warranty analytics software.


    Go to Part One: Opening Keynote Speeches
    Go to Part Two: AIAG Early Warning Standard
    Go to Part Three: Extended Warranties 101
    Go to Part Four: Customer Loyalty
    Go to Part Five: National Regulatory Trends
    Go to Part Six: Auto Warranty Fraud
    This is Part Seven: PC & Appliance Warranty Fraud
    Go to Part Eight: Warranty Metrics





  • AMT Warranty Corp.
    Fulcrum Analytics
    Sign up for a free subscription to Warranty Week:
         subscribe     change of address     unsubscribe