Wednesday, August 01, 2007

So Monday night I made a global change to some server settings. According to process, I burned incense, made half a dozen posts on the official support forum (EMPLOYING TEH CAPS LOCK & MISSSSPELLING OPSHUN) and waved my lucky chicken foot over the hardware.
My manager asked if I'd heard anything from the users yesterday and I told him that in ten years of doing this I'd gotten only one "Hey, everything is going great, thanks" email from a user and that their silence in this case was an indication of success.
He decided to call them (any user who has complained about the issue in the past six weeks) to find out what they thought about the change.
To their credit (and I hate extending users credit more than I hate actually learning their names), most seemed . . . happy.
My manager continued to make these calls for most of the morning. Secretly, I suspected he kind of got a rush out of talking to users who weren't angry about some issue and his ability to set the topic to something which was working well and then hang up after asking his question was a big motivating factor.
Late in the afternoon, I was summoned into his office so that he could announce this success in front of our VP of IT.
The VP wanted charts and graphs proving that performance was better and said that user testimony was valueless as a data capture point.
"What if I have three pieces of toast before logging in and everything is great? And then I have pancakes for breakfast another day and everything is working wonderfully? But then, after a night of drinking and staying out I wake up with a headache and log in and decide it is working horribly? User perception depends too much on outside factors we can't reliably quantify."
There was nothing for me to do but offer to write a "best practices" document making pancakes our officially supported architecture.
I don't think he found that as amusing as I did, but I don't have any way to track that other than by his reaction so my assumption may be invalid.


Ted said...

It's all about controlling the data!
Break the system for a few days, then miraculously fix it, THEN take the user survey.

Garrick said...

I think it is perfectly acceptable to be good by comparison.