Unrealistic Expectations
In a post about customer service and tech suport, Mark over at Enterprise Search makes the point that “zero problems is not a reasonable expectation when you’re living on the crest of a technology wave”. If fact, it might be a definition of the crest of the tech wave.
At one time, new products had extremely intensive testing and trials with small groups to figure out where the problems were. And this was after very extensive design and prototyping. Detroit is the only area that this sort of process still goes on in its full glory. Which is one reason that Detroit’s having problems now.
Another reason Detroit’s having problems is that the oil crunch in the 1970s wasn’t taken well to heart. The learning only seemed to be short-term with no long-term retention. Or, no way to respond to changes quickly. This many be populist of me, but the CEOs and other management are directly to blame for the loss of jobs and failure of their companies to do well now. Yet, they’re getting the buyouts.
Back to the subject though: While simulation, prototyping, and testing have all been made faster it isn’t enough. The increasing speed of products appearing in the market is forcing companies to be more and more flexible and agile. This almost guarantees that products are pushed out to consumers without the same level of testing as earlier. In the 1980s Hewlett-Packard could figure on a software product having a liftspan of at least 5 years, hardware was longer with at least 7-8 years. Today, product cycles in the consumer markets are 6-9 months. Commercial-to-commercial or government markets are longer, sometimes much longer. However, they have speeded up too over their historical cycles.
Low-cost manufacturers can’t afford good customer service. That costs time and people. They’re only alternative is to see what kind of feedback they’re getting and make a new product without that problem. They’re using customers as their testers. The customers in turn pay less as that cost risk of a badly working product is essentially factored into the product price.
Other manufacturers charge more and have better support. Since the support costs would skyrocket without adequate testing, these manufacturers have a strong incentive to test their products more effectively. Still, their fallback is to make new products without any problems found by customers.
Software has exacerbated this. How long has Google had products in beta? Hardware systems are often designed to be modified by changing their firmware. Upgradable products are one result. Customers have come to expect at least some give and take in products they buy and will pick where on the product maturity curve they are comfortable. An “early adopter” in one area is a “tried and true” product user in another area.
In the end, the line between manufacturer and consumer is less distinct than it used to be. More information flows back and forth there. Customer support is one traditional vehicle for that, but isn’t the only solution.
November 12th, 2008 at 11:26
Good stuff. An issue I have with software QA is the over-reliance some organizations place on automated regression testing. YES, this is a good tool, especially for nightly builds, to insure the coders haven’t broken something, but it’s only half the story.
For more complex software, especially software tools and “platforms”, employees throughout the company and key customers should be urged to build sample applications. Trying to build something is the number one way to find new bugs, missing features and usability issues. And, after all this “testing”, a company has a large library of sample code, which is often more helpful than some of the formal doc, assuming they every get around to writing it. And don’t let editorial review bottleneck these demos - have a “quality” stamp for how polished each demo is - if it’s never even had spellcheck run on it, just label it as such and still put it online. The only real review requirement is to do a quick check for true intellectual property issues, obscenities and hardcoded passwords. Let the users flag the demos that need the most editing. Show me a company with 100+ sample apps and I’ll show you a reasonably well vetted tool.