At Home Depot, employees can’t even be bothered to check their systems to find out if they carry an item one aisle away on the shelf. (See SI’s recent series on Home Depot’s snafus that could be the beginning of their end in five parts: One, Two, Three, Four, and Five.) At Best Buy, if you want a manager, you have to get him or her yourself (Part One) and if you want your problem to be solved in a timely manner, forget it as it will take at least three customer service agents a minimum of a half hour to do so (Part II). Unless you want wi-fi porn on the big screen, you’re out of luck (as there is no best buy experience). And United, not content with breaking guitars, has decided to go all out and mange their switchover of their ticketing, web, upgraded, and related systems so poorly that United Airlines Should Be Ashamed (SpendMatters — read it!).

In a nutshell, customer service is going to hell across the board, and it’s doing so almost every single time a company places too much emphasis on technology to “automate” and “streamline” customer interactions. Organizations seem to have forgotten that unless you truly are one of The New Technology Elite, technology can never replace talent, systems can never deliver the service a real person can, and any attempt to implement such big data systems in one fell swoop will almost always result in an implosion that rivals the Big Crunch that some physicists predict will occur at the end of the universe (which immediately precedes the Big Bang, as we all know from the late Douglas Adams who told us all about what you see when you go to Milliways, The Restaurant at the End of the Universe).

We should take a lesson from this, as those who do not study the lessons of (very recent) history will be doomed to repeat them. So what should we learn? First, as Jason Busch points out in The Comments to his post on how United Airlines Should Be Ashamed, we should note that:

• Flex Capacity Matters
  When the technology inevitably fails, real people will be required to handle the situation.
• Hard-Wired Integration Outside the Cloud Will Always Be a Nightmare
  And this nightmare will become a reality when trying to add new nodes or change out the guts.
• Communications Matter!
  Upstream with suppliers and downstream with suppliers. Over-communicate and invest the time to keep everyone in the loop! Finger-pointing helps no one.

In addition, we should note that:

• Big Bang Upgrades Only Result in Big Crunch Failures
  We should have learned our lessons from the FoxMeyer and Nike ERP disasters which caused the bankruptcy of the first and hundreds of millions in losses for the second. If we didn’t, these recent fiascos should show us just what Big Bang upgrades do!
• Thorough Testing is a Must
  When you have to port hundreds of thousands or millions of data elements from one system to another, test thoroughly. Anything that can go wrong will, and anything you don’t test for will go wrong.
• A Rollback Option Must Be Available
  There’s a reason relational databases have rollback capabilities. In case of failure, you can always go back to a previous state. In the United instance, the old system should have been left online, and transactions duplicated across the new and old system for at least a few days, until it was clear the new system worked sufficiently well to run on its own. At the first sign of massive failure, the entire new system should have been taken offline and all transactions directed to the old system.
• The Disaster Recovery Plan Should Be Ready to Go
  And should be communicated up and down the chain well before any transition occurs. Be prepared for failure. You’ll minimize disruption and recover much faster if you do.

What else can we learn?