Downtime and outages are two things that — if planned for — can make or break a customer’s respect for your company. I was on the Blockbuster site yesterday, and behold the message I got — a generic Weblogic server error. Did this give me confidence in Blockbuster’s ability to manage my account and provide the service I’m paying for? Hardly.

These system messages can be customized. Here’s what I should have seen if the server had a hickup:

Service Temporarily Unavailable

’This web site is down for routine maintenance. We apologize for any inconvenience. Please try again later.’

Or better yet:

Try Again in a Bit

’Due to an overwhelming surge in the popularity of our site, we’re down for just a bit. We’re adding better capacity to meet your needs. Your needs are valuable to us — just try again in a few minutes.’

Come on Blockbuster. You can do better.

On a related note, production software inevitably needs maintenance. One way to determine when and how it can be done is through analysis of prior user traffic.

Define a scheduled maintenance policy so that everyone knows when changes to a production system can be deployed.

I often send out an email to my customers a week in advance of any systems changes. The day of the change I post a message to the web site alerting users to the planned outage later in the day.

Give users adequate advance notice of scheduled downtime. Doing this sets expectations and avoids embarrassing issues of surprise software installations.