April 11, 2006

High-Tech Warranties:

Designing products that can be fixed easily or replaced quickly will reduce warranty expenses. What's required, though, isn't so much better software as better teamwork and planning. Meanwhile, a hunt for patterns and anomalies also works for fraud detection.


At last month's Warranty Chain Management Conference in Las Vegas, electronics and computer manufacturers reported some of the same challenges and problems as manufacturers in more mature industries. If there was anything approaching a common theme to the suggested solutions, it was that people and procedures are frequently more in need of an overhaul than any of the actual claims processing software.

At Sun Microsystems Inc., the problem was finding a way to create an honest and complete picture of the company's warranty and service costs, and to build that into the product from the outset. At Sony Electronics Inc., the problem was the high cost of warranty work and the low levels of customer satisfaction it produced. At Toshiba America Consumer Products LLC, the problem was knowing how to tell the difference between an unusually efficient and productive repair center and a fraudulent and corrupt repair center.

Cynthia Zamora, senior director of finance for global customer services at Sun Microsystems, gave a high-level presentation about the benefits of warranty to a business computer manufacturer. If she were making automobiles, she might have spoken about the value of early warning systems. If she were making consumer appliances, she might have spoken about the value of quality, reliability, and warranty data to product designers.

Instead, she noted, "Sun is a business-to-business high-tech company. We are not in the consumer space. So a lot of things that we're hearing at this point don't have much relevance to us."

Business vs. Consumer Computing

One difference she noted with the consumer end of the computer business was a reluctance at Sun to push customer self-replacements of parts, given the increased complexity of the systems in question. What's more common is for Sun to design remote diagnostics and monitoring into the systems, so its service staff can quickly determine what parts may need replacement. With a consumer brand, it may be more important to design side panels that open easily and hard drives that snap into place. At Sun, it's more important to get the unit back in business as soon as possible.

It is therefore essential, Zamora said, that service people be involved from the outset in product engineering and design. "You have to design for serviceability and diagnosis," she said. "We've found our best success when we have a very cross-organizational team involved in product design."

That approach feeds into a closed loop process, she explained, in which the next step in the process is to estimate future warranty costs for the soon-to-be-launched product. After the actual launch, real warranty costs can be measured, and compared against expected levels. And then, after the product has been in the market for a while, it's time to take a look at how product design or service delivery can be improved to reduce costs or raise customer satisfaction in the next generation of the product.

"It isn't just financial metrics," Zamora said. "It isn't just the hours and time costs to meet your contractual commitments. Are you doing it in a way that satisfies customers? It isn't only the warranty itself. It's how you deliver that warranty."

Zamora then explained a little about how Sun develops its warranty cost estimates. The first consideration, she said, is to take a look at engineering data such as the mean time between failures, which gives the company some idea about the likelihood of repairs over a given time span. Then Sun matches that against marketing information such as the proposed terms and conditions of the product's warranty -- not only its duration in months or years, but also in terms of the promised response time for a repair. Some customers are content with twenty-four-hour response. Other customers are willing to pay for eight-hour response.

From the outset, Zamora said, Sun is very careful to include all expected warranty costs in the product design and serviceability discussions. "That's your call center. That's your field support. That's your spares. That's your freight. Everything that goes into providing a warranty goes into the cost of a product," she said.

Supplier Warranties

With suppliers, Zamora said it's important to know as much as possible about the likelihood of failure of a given component, because in some instances it may make more sense to buy the components with very short warranty periods or perhaps without any warranty at all. That way, Sun can pass the savings on to customers, and absorb the cost of repairs itself.

Elsewhere in the computer business, a purchasing manager might be encouraged to find a lower-cost component supplier, without regard to whether those cheaper parts might lead to more frequent and costly repairs down the road. Likewise, an engineer might find ways to cut the cost of manufacturing that increase the cost of service later on. Not at Sun. Reducing the cost of manufacturing or purchasing is not an end in itself. It has to be considered alongside the cost of support services.

In terms of measuring warranty costs, Zamora said the key is ongoing monitoring. It's important to establish a baseline for a new product as early as possible, so that unusual spikes in certain types of failures and repairs can be detected quickly. It's also important for Sun to understand all the components of its own warranty cost structure so that it can more accurately measure acquisition candidates, Zamora said.

"It helps you evaluate acquisition targets to see if the data you're receiving during initial exploratory conversations is reasonable. It helps you understand their business model better, and it helps you understand if those costs are perhaps not representative of the way you if you acquire them would represent the costs. Not that it's right or wrong, but there are definite differences in how companies do their accounting and how they put costs into one bucket," she said.

Is Broke, Needs Fixing

At Sony Electronics, the U.S. consumer products sales arm of the Sony Corp., the task handed six years ago to Mike Ehlers, the company's vice president of service, was to repair a warranty repair system that was clearly broke and in need of fixing.

Back in 2000, Sony Electronics concluded that its service operations needed a major overhaul. Costs were high and customer satisfaction was low. Usually, Ehlers said, high service costs correlate to good customer service and low costs correlate to poor customer service. But not at Sony.

He said the culture that existed at the time within the Sony service department viewed repair costs as a good thing. It brought revenue to the "service silo," he said, which was what kept all the service people employed. So it was something of a repair culture, where the focus was on service company profit maximization. Failure-prone products were a good thing, because they generated lots of warranty work.

"We wanted to change that," he said. And so, a business process re-engineering committee was chartered, bringing together people from finance, service, operations, and human resources, with the support of top management. The first decision they faced was whether or not they should just simply outsource the whole service offering to a service company.

"The first thing we realized was if we outsource this operation, we will be outsourcing an inefficient operation," Ehlers said. "That didn't make any sense. So we decided not to do that. And we first said, 'let's fix it.' And we learned some things along the way when we fixed it that were actually pretty interesting."

No Outsourcing for Sony

One of the lessons learned, Ehlers said, is that outsourcing didn't make sense for Sony. It would have put Sony's service operations into an "outsource prison," he said, from which escape would be difficult. It would have been tough to bring the process back in-house, and even tougher to switch from one service provider to another. "No matter how good your contract is," he said, "it's hard to get out of."

So instead, Sony decided to "in-source" the process. That meant taking the repair process, where possible, out of the hands of the networks of authorized service centers that are so prevalent in the consumer ends of both the computer and audio/video businesses. That also meant, in Sony's case, cutting down the number of service centers owned and operated by Sony from 22 to eight, to five, to three, and finally to one.

The first thing the company did, Ehlers said, was to begin centralizing all its repair operations for everything but computers and televisions in a facility located in Nuevo Laredo, Mexico, just across the border from Laredo, Texas. In that location, which was already operating as a refurbishment center as well as a tape manufacturing plant, Sony found it had a technically skilled labor force that was willing to work for half to one-third the wage rates that prevailed north of the border. Meanwhile, the actual border was very close, and the cross-border trade regulations were very favorable, so logistically there was little downside to the move.

But there were other advantages of centralization, Ehlers said. First, it allowed technicians to hyper-specialize in key product areas, working on, say, just PlayStations, or just CD players. Second, after going from 22 repair centers to one, Sony found that there was less idle time per worker, because there were fewer slow periods. This helped to reduce labor cost even further by boosting efficiency. Previously, a handful of service centers might have been unusually busy while the others were unusually inactive. With one service center, volumes didn't fluctuate as much.

Repair vs. Replacement

The next step Sony took was the most radical. After centralizing in Nuevo Laredo, it replaced its repair-and-return strategy with an exchange strategy. Sony found that when a product was repaired at one of its 22 internal service centers, the process usually took between 14 and 16 days. When the repair was instead performed by an authorized service center, it usually took seven to nine days. But when the defective product was simply exchanged for a refurbished unit, the turnaround took only four to six days on average.

The exchange strategy also cut costs, Ehlers said. Let's say that Sony found that products being fixed at an authorized service center cost it X dollars. Products getting fixed internally cost 30% more than X, he said. But the process of replacing a product with a refurbished unit after an examination of the broken unit in Nuevo Laredo was found to cost only half as much as X, Ehlers said. And customer satisfaction was measured to be higher.

For products selling for under $200, Ehlers said, an exchange policy like this probably makes more sense. The idea is to quickly replace the broken unit with a refurbished unit, then to fix the returned unit and put it aside, eventually shipping it to the next customer who sends in a broken unit. Fixing the same unit and returning the same unit to the customer merely lengthens the delay and increases the cost. And customers don't seem to mind getting back a different unit.

Of course that strategy isn't going to work for computers, where lots of personal data, applications software, and settings are installed on each unit. So the computer product lines still use the authorized service center model, as do Sony televisions. But most of the company's video game, home video, and audio products are now shipped to Mexico for replacement and refurbishment. And the number of authorized service centers still doing warranty work for Sony has been reduced by 600.

Better Warranty Data

Ehlers said the quality of Sony's reliability data also improved under the new centralized exchange strategy. Before, any of up to 2,300 authorized service centers did the warranty work and sent in the claim form. After, all the data was already in Sony's systems, where it could be analyzed to spot trends and anomalies. Ehlers said problem detection delays fell from a matter of months to a matter of weeks.

The biggest problem for the new strategy, he said, was that it forced the customer to ship the product to Sony, rather than dropping it off at the nearest local repair facility. However, a solution was found, and it actually increased the number of possible drop-off points. To initiate a product replacement, a customer can now bring their broken Sony product to any of 3,600 United Parcel Service locations, which will pack and ship it to Nuevo Laredo.

The result has been a 60% reduction in the Sony service operation's selling, general and administrative expenses between 2000 and 2005, Ehlers said. And the average turnaround time has fallen from 12.4 days to under five days, so customer satisfaction has risen. "We've also been able to reduce warranty fraud quite a bit," Ehlers added.

To Catch a Thief

Fraud was a common thread running through several of the presentations. During the second day of the WCM Conference, a panel discussing efforts to reduce warranty fraud was kicked off by Scott Schock, the senior warranty manager of Toshiba America Consumer Products LLC.

Toshiba, he said, still relies on a network of authorized service centers. Warranty claims are submitted electronically through a third-party warranty processor, who first validates the data in the claims before passing it on to Toshiba. Then Toshiba processes the claims for payment.

"So the majority of these claims we never visually see," Schock said. "They just go right through the system. Unfortunately, that does open up some doors for fraud."

To combat this, Toshiba has developed an anti-fraud system that relies on three ever-more intrusive levels of fraud detection. It begins with simple random sampling of the warranty data, and a search for obvious statistical anomalies. Next it moves on to a manual review of claims. The third and most intrusive round includes an on-site audit and inspection of the repair facility.

"I'm not going to give you all my secrets," Schock said, but in general what the company is looking for in the first round of fraud detection are deviations from the expected averages and ratios. For instance, a service center might use a higher-than-average number of parts per claim. Or it might have a lower ratio between claims and serial numbers than most of its peers.

It could be something as simple as a higher-than-average percentage of claims that contain parts, Schock said. He displayed a graph that showed the distribution of claims that included parts, and said there was a better correlation with the place that did the work than there was with either the model number or the type of failure. What this says to him is that certain repair centers tend to replace more parts than others, simply to inflate claims.

Working Too Fast

He also showed how repair time data can be used to detect fraud. One service center, he said, was completing 62% of its warranty repairs in two days or less. That meant it was completing most repairs in less time than it took many other service centers to merely order parts. In fact, for all service centers, less than 20% of repairs were turned around in two days or less. The median, he said, was closer to four or five days. So why was this one service center fixing products so fast?

The answer, not surprisingly, was that most of the customers and their resulting warranty claims were made up. They simply didn't exist. And it was detected because the warranty work was being done too fast. "Under our previous theory, where everyone is honest and is just trying to make a living, we would have given this guy an award, because his turnaround time was outperforming all the other service centers," Schock said.

Next Schock showed data for eight service centers that compared the number of claims to the number of units. At one extreme, an unnamed service center reported 84 unique serial numbers for every 100 claims. At the other extreme, another reported 97 serial numbers for every 100 claims. Four of the other six were at 93/100, and one was at 96/100.

Toshiba decided to audit the service center with the 84/100 ratio, he said. What they found was that this particular service center was submitting more than one warranty claim per product -- one when parts were ordered, and one when parts were actually installed. It wasn't so much fraud as it was poor training, and was amicably resolved.

Ironically, the service center with the 97/100 ratio was the same one that was completing 62% of its claims in two days or less. The reason why this particular service center didn't have too many duplicate claims per serial number was because it was careful not to re-use the same serial number in multiple fraudulent claims. So ironically, he was caught because he was too far on the seemingly good side of these averages and ratios.

Another Toshiba account also had a very fast turnaround time, and it seemed that the televisions were failing almost as soon as the customer got them home. They also had a very low percentage of claims that included parts. Toshiba investigated, and found that the account was providing free home deliveries and installations to the customers, and was generating a partless warranty claim to compensate itself.

Calling Customers

Schock also said Toshiba randomly calls a small percentage of its customers by phone after the repair has been completed. The investigator says it's a customer satisfaction survey, but it's really more of a customer verification survey. Does the customer exist? Do they own a Toshiba product? Was it recently repaired? If any of the answers are no, or if the phone number doesn't even exist, then there's a high likelihood of fraud.

But the surveys also uncovered a different kind of fraud. Schock said roughly 80% of the calls checked out as legitimate claims. Of the remaining 20%, some of the customers reported that yes, they had initiated a claim, but no, the repair had not been completed. That was because for the product in question, parts were covered under warranty for a year while labor was free for just the first 90 days. So the customer had declined the repair once the cost of it was clear.

"So if the customer refused the labor, why am I still paying for parts?" Schock asked. The answer was that the service technician submitted the claim anyway, even though no work was done. But all the other information was correct: the name and address of the customer, the model number, and its serial number.

Sometimes the streets don't exist, or the phone numbers don't check out in the reverse-lookup phone books. When such basic details don't check out, it usually indicates fraud.

The Final Step

After detecting such a pattern, Schock said Toshiba will attempt to schedule an on-site audit and inspection of the repair facility. He said the company's contract explicitly give it the right to make such a visit. But usually, such an audit is contracted out to an independent forensic accounting specialist, and usually it's a last step before the repair facility's contract is terminated and/or the case is marked for legal action. And usually, the visit never takes place, because the repair facility declines.

One time, however, the repair facility didn't think it would be caught, so it allowed the inspectors in. Schock said they discovered not only that the company was submitting fraudulent warranty claims to Toshiba, but was also duplicate-billing the customers for the same work. "Can you imagine that? Not only are you stealing from me, you're also stealing from my customers," he said.





Assurant Solutions Shed The Light
Fulcrum Analytics
GWSCA First Annual Conference on Service Contracts
Sign up for a free subscription to Warranty Week:
     subscribe     change of address     unsubscribe