Why isn’t Cadillac or GM’s vehicle reliability better than it is?  Okay, Cadillac has been having good survey results in initial and four-year quality.  I get that quality and reliability have improved, and that they are near the top of the automobile industry.  But why aren’t they better than they are?

Cadillac’s Dealers are independent operations.  Every warranty repair on every GM vehicle is entered into the Dealer database, so that a claim can be made to GM for the warranty repair.  These warranty repairs cost GM Millions of dollars a year in lost, wasted cost, and decreased customer satisfaction.

Why can’t GM simply analyze this warranty data, determine what parts are having reliability problems, change the design or the supplier, and have reliable parts?

The warranty database currently has information from various sources. The PRTS+ database has information that the GM engineers have entered with regards to problems that they have detected. The QWIK database has problems that dealership mechanics have entered with regards to customer complaints and their root cause of a single vehicle.

But THINK about the systems that we have available in Cadillacs right now.  The vehicles themselves have thorough self-diagnostics.  They also have snapshot capability, as well as GPS and OnStar satellite communication.  The second a diagnostic trouble code goes off in any Cadillac anywhere in the world, a signal should be sent via OnStar, and a red light and warning siren should sound at the Cadillac factory at Lansing Grand River.  Immediately the cause of the problem should be diagnosed, identified, and resolved.

OnStar:  “This is OnStar. We have detected a problem with your vehicle; is that correct?”

Customer:  “Uh, I see a red light on the dash, yeah”

OnStar:  “Thank you.  I have made a reservation at the nearest Cadillac Dealer. I am downloading the turn by turn navigation to get you there.  I have alerted them that you are enroute, and your Service Expert will be Michelle.  She is awaiting your arrival.  I have a Cadillac Engineer standing by to review the Technician’s report as soon as it is ready, and we will get you on your way.  Do you want me to stay on the line while you drive to the Dealer?”

This 52 page thesis from Jelani Ellington for MIT in 2005 addressed some internal roadblocks that GM has that keep it from harvesting the flow of valuable information: [Thesis paper]

GM’s approach:

Engineers at GM pull data from the warranty database to see failure trends. This warranty is segregated by vehicle line. The warranty data is further segregated via labor codes. The data is then placed in two Pareto charts. One chart is based upon cost per vehicle (CPV) while the other chart is based upon incidents per thousand vehicles (IPTV). The charts are then used to determine which vehicle warranty problem to work on first. There is no one method that is used across the organization. Some vehicle lines use cost per vehicle to work on problems first whereas other use incidents per thousand vehicles to determine which problems to address first. The only common thread is that all vehicle lines place safety issues above either cost or frequency.

Fail: This process seems to have FAIL written throughout.  It lacks immediacy, and depends on the Customer experiencing the problem, debating it with the Dealer, and many customers together becoming dissatisfied before an issue gels together to be bad enough for the Engineers to identify it months later as a problem.  FAIL.

Success: Every time a problem comes up in every car, a light should go off at Cadillac and an on-call Engineer respond.

Once the management team decides which failure to be worked on, GM assigns a warranty engineer to fix the problem.

Fail:  only fix things after you make a lot of customers unhappy

Success: fix every customer’s problem

After the failure is found, the engineers use the RedX procedure. Engineers are given 30 days to root cause an assigned failure.

During the solution phase of a problem, the warranty engineer informs the Designing Engineer of the root cause of the problem. The design engineer then, likewise, has up to 30 days to come up with a solution to the problem.

Implementation is the part of the process where the solution is carried out. Implementation time varies greatly. Implementation time can be of a very short duration or can span a period until the next model year.

The final phase of the PRTS+ process is feedback. Feedback is due no later then 180 days after implementation. Feedback is design to close the loop and make sure that the new design is working properly or to maximum efficiency.

Fail: no immediacy; every 30 days Cadillac will have sold another 15K vehicles, and enraged another group of Customers

Success:  This step is what engineers do BEST.  Identify the problem to them, and let them analyze it.  But move the Engineers back up to the START of the problem.

In the current GM Warranty system, “a minimum of 92 days will pass before a problem that a customer experiences becomes known at General Motors“.  FAIL.

At General Motors, design engineers do not have access to the warranty database. FAIL.

If all diagnostic trouble codes that were set on any GM vehicle, along with a DTC snapshot, were reported automatically to a central database, the Reliability Engineers can cut months and Customer frustration out of the picture.  “Ninety percent of General Motors’ warranty problems are solved by reproducing the failure with GM vehicles or in GM plants.”  Give the Warranty Engineers the info on what faults are happening, in real time as they happen.

The paper concludes:

There exists a wide variety of time to discover defects in the field. Reducing the measurable standard deviation of time from introduction of a vehicle to the field to assigning engineer for root cause of a problem to implementation of the solution is the key to solving the problem. There is not real time data transfer from the field to the corporation. If this link did exist, the corporation could fix these problems almost as soon as they occurred instead of months later. To successfully fix the issues, General Motors must become a part of the global telematics market and increase its capability to receive live data. General Motors must also work on their internal issues and conflicts as well. General Motors must also think holistically when solving its problems. General Motors tunnel vision approach to warranty obviously is not working. Warranty elimination is a process that must be a corporate-wide issue and not a departmental or organizational issue. If General Motors is to survive and thrive in the coming decades, then they must minimize their mean time to failure discovery for the sake of warranty costs, which go directly to the bottom line, and for the sake of perceived quality.

Operate as an agile, responsive organization.  Give the Engineers the real-time information they need, then them loose and let make the great cars they are capable of making.