Web site disasters made easy

In 1997, I was working in the IT department at a midsize consumer products company in the San Francisco Bay Area. My job was mainly to keep the network up; the company had no Web presence. But as our competitors ate more and more of our lunch, it gradually dawned on management that we ought to be selling online. So I built a LAMP (Linux, Apache, and Perl/Python/PHP) sales portal that handled online ordering and a corporate Web site. It generated revenue from the outset.

Then, sad to say, the trouble started. Our parent company in Japan decided to consolidate its North American holdings. Within six months we had a new management team, and most of our IT department was outsourced. My little Web crew was pretty nervous, especially when we heard that a new corporate VP was flying in to help rescue our poor little LAMP system. We didn’t know it needed rescuing.

Apparently, someone in Japan wanted a new Web platform built on software I’ll call “NetTurkey.” When I pointed out to the new VP that this would introduce huge feature gaps compared with our current LAMP implementation, his solution was to dismantle my Web team and banish us to four different departments. I was kept on as the lone local expert, while a consultant team from “Acme Global Services” initiated what was supposed to be a three-month project to replace the LAMP platform.

Unfortunately, the Acme Global people were completely new to NetTurkey. Basically, they were cracking the manuals for the first time. And then someone decided to outsource the HTML pages to India. For some unknown reason, the Indian crew called for a Web site content freeze while they recoded static HTML pages. Needless to say, our sales and marketing executives ignored the freeze.

Clueless project managers came and went; I lost count after eight. At the end of month seven, Acme relaunched the Web site. But the NetTurkey software was still missing about 40 percent of the functionality of the LAMP platform. Customers hated it.

After a full year of work with very little to show for it, our Japanese parent merged with a competitor and decided to sell off all its North American businesses. A buyer was found, but the launch date for a fully functional online ordering site became a critical negotiating point. Desperate, our management team decided to launch a “fully redesigned” NetTurkey-based Web site — without testing.

One week before the North American businesses were to change hands, the site went live. It was a complete disaster with customers and sales staff alike. Within 48 hours, our soon-to-be-new CEO found his way back to the original Web team — which, by this time, was mainly me — asking whether we could roll back to the original LAMP platform.

I think they ended up spending close to $6 million on the “upgrade” — a complete write-off that was buried in the sale of the U.S. companies. Soon afterward, I moved on, but I still take a peculiar pride in the fact that nine years after its humble birth, that original LAMP site is still online, managing close to $200 million annually in e-commerce sales, I’m told.

Even so, I wouldn’t want to be the project manager who takes on the company’s next Web platform replacement.