posted 12-18-2003 06:55 PM         

By Ralph Kisiel
Automotive News / December 15, 2003

Nearly 250 new or redesigned vehicles will hit the North American market in the next three years, and rules that took effect Dec. 1 will provide the government - and product liability lawyers - with a wealth of information about postlaunch problems.

Many in the industry fear that this avalanche of warranty and customer service data will provide lethal ammunition for product liability lawsuits. And they're arming themselves with IT tools designed to find and fix new-vehicle problems faster.

They are called early warning systems, and they use software to make sense of the mountain of information - from the dealership, assembly plant, owner surveys, customer call centers, public records and replacement parts sales data - that can spot a troublesome part or vehicle system.

Key to this is the ability to analyze information from dealership service departments. Data from mechanics - their details about the thousands of problems diagnosed and fixed daily and recorded onto service bay computers - contain a gold mine of data for spotting problems.

Software can scan key words, spot patterns, identify trouble and keep isolated incidents from snowballing into widespread defects.

Tight lips

It is an emerging technology, and automakers won't talk much about it. A General Motors (gm.com) spokesman, for example, says "GM is not looking to go public with what we are doing in this area."

But in August 2002 - two years after the first phase of the highly publicized Ford Motor Co. (ford.com) recall of Firestone tires - Ford saw the benefits of early warning technology. It faced another tire recall, this time with tires from Continental Tire North America (continentaltire.com). The tires were prone to vibrate, wear unevenly and, in some cases, separate from the tread.

Continental voluntarily recalled 600,000 tires on Ford Expedition and Lincoln Navigator SUVs. Neither company will talk about the event. But at the time of the recall, Ford, which used IBM technology (ibm.com), credited each company's early warning system for spotting the problem.

Kevin Mixer, research director at AMR Research Inc. of Boston, says the companies were able to pinpoint the problem in less than a year. Typically it takes 18 months to identify a problem in the field, he says.

IT-based early warning systems are expected to play a larger role in helping the industry battle rising warranty costs as well as complying with the TREAD Act. Congress passed the TREAD Act in 2000, after the Ford-Firestone debacle. On

Dec. 1, automakers began reporting a wide variety of warranty and customer service data to the federal government. The goal is to prevent recalls the size of the Ford-Firestone tire campaign.

AMR Research (amrresearch.com) estimates that liability claims are likely to triple with the public dissemination of TREAD Act data. The company has just completed a study that recommends automakers and suppliers put early warning systems into place to reduce liability exposure and reduce warranty expenses.

The North American auto industry pays out $12 billion annually in warranty claims, AMR Research estimates. That works out to about $712 for each vehicle sold, based on North American light-vehicle sales in 2002.

True, automakers already analyze warranty claim data. But that analysis is typically limited to structured information, such as diagnostic trouble codes.

Going to the root

Those codes don't always capture the cause of a problem, says Barton O'Brien, vice president of business development at Attensity Corp. of Salt Lake City (attensity.com).

Attensity would not identify the automaker clients using its text analysis software. But O'Brien points to a manufacturer that identified the cause of clogged fuel filters just days after beginning to use the software.

For months, trouble codes kept pointing to the fuel filter. But in its analysis of service technicians' comments, the software isolated key words and phrases that revealed the real problem: A supplier was drilling mounting holes in the fuel tank, and shavings were falling inside. Eventually, those shavings clogged the fuel filter.

The software helped pinpoint the problem by picking out phrases such as "with plastic shavings" and "from fuel tank assembly" and "from manufacturing."

"It ended up being a fairly major problem," O'Brien says. "They were replacing fuel filters, fuel valves, fuel screens. We enabled them to hear the voice of the technician in a summarized way."

Attensity is not the only company vying for auto industry business. In March, IBM Corp., of Armonk, N.Y., and software vendor SAS Institute Inc. (sas.com) of Cary, N.C., began selling its version of early warning software. Their software analyzes data from a range of sources, including service bays, the assembly plant, customer surveys and call center contacts.

Wealth of information

"The text data is the difficult data to really get your arm around," says William Lang, IBM's vice president of global automotive solutions. "But the text data actually is today the data that is holding the wealth of the most information."

Three automakers and one Tier 1 supplier are using the IBM-SAS product, Lang says. He declined to identify the parties.

"What you are trying to do is reduce the lead time in finding out that there's a problem in the field," Lang says. "There is a lead time before they see a trend and can react, and it's in the 200-day range."

Compressing the time to spot trends will lead to reduced warranty costs, higher customer satisfaction, and safer vehicles, Lang says.

But don't expect the automakers to publicize their successes.