Designing For "A" Customer Proves
Endemic
by Paul McGoldrick
In a recent Editorial I wrote about a semiconductor vendor that had, yet again, developed a product that was quite obviously directed at a single customer and how idiotic such behavior was/is. I was subsequently surprised by the correspondence that I received from readers which all basically said, "were you writing about my company?" followed by a tale of woe of one kind or another about something their company had done. All but one reader made it clear that their letters were not available for publication, for fear of reprisals.
The tales of woe were different but all were strangely similar. Most involved a project with an end-use of the product that the design teams were unsure of. Most involved rush decisions, a measure of panic about the competition also being briefed, and possibly getting there first. Most involved a single set of inputs from a single customer.
That some semiconductor companies could get snowed like this should not be surprising, given the basic lack of understanding of most management as to what their market is really about - beyond some paraphrased generalities that they can read at the bottom of their press releases - but these ridiculous projects could be avoided by the design teams insisting that they fully understand where the product, that they are going to spend the next 9 - 12 months of their working lives on, is going to be used - and how. Unfortunately, it is clear, there are a number of companies where wrath and anger are still component parts of the decision-making process and some engineering groups do not have a vocal enough management structure to respond.
In truth the customer is often an idiot who really does not understand what he wants.
Some years ago I was sitting at my desk in Southern California when I received a call from Area Code 703 (Virginia) from a man in "government" who insisted I call him back and then he would call me back so we could be sure of who both of us were. (Did you get that?) He asked me a number of technical questions about the features of one of our remote-control systems and my answers seemed to satisfy him. A few weeks later we received a government order for a system, at list prices, which we were happy to ship out.
Every 4 or 5 weeks after that we received an order for a replacement piece of hardware and eventually we were told that the system was monitoring the Top Gun sites off the San Diego coastline. Every time there was a direct hit we got a replacement hardware order!
The point of the story is not to show how incredibly inefficient governments can be, it is the fact that the procurement process was kept secret and was far from optimal for the taxpayer. If my 703 caller had explained what was wanted we could have put together a remote "target" box for them, with the functionality required, that would have cost them a couple of hundred dollars instead of the couple of thousand they spent every month or so.
I am not against vendors talking with customers, hardly that. But I am very much against customers only having a single point of contact in new product discussions, allowing the message to be distorted between origin and designers. If the response of readers is any judge - and I very much expect that it is - there is still an endemic problem out there which has to go away if a company is to succeed long term. One suspects that product decision procedures that don't allow questions to be asked along the way is one problem, and certainly announcements by CEOs to analysts that X-number of products will be released in any one quarter allow these single-customer products to slide right through.
Tell me more, people!