gavinshearer.com

« Interviewing | Main | Go East, Young Man »

August 12, 2004

Lies, Damn Lies, And...

Back in November of 2003, I went down to Apple for an informational interview. I was curious about doing my summer internship in Cupertino, but cultural fit is super-important to me. The point of the interview was to get the scoop - Up Close And Personal - before I really threw myself in to the full-court press of working for Steve.

As luck would have it, the UW career center set me up, and I wonld up chatting with a few UW alums on Veterans Day (fun fact: Apple's cafeteria offers free coffee to employees ... and free apples!). We had a wide-ranging, nearly two-hour talk about MBAs, jobs in tech, the Bay area, and what Apple is like. At one point I asked Chris (a Finance guy) what classes he would take if he had his MBA to do over again. His advice?

"Statistics. Take as much as you possibly can."

After being at Microsoft for just a few weeks, I now know exactly what he was talking about.

I've written before about how Microsoft is a metrics-obsessed company. Part and parcel with a commitment to metrics comes an ability to interpret - not just collect - data. This philosophy has massive currency inside Microsoft, and it's manifesting itself in very interesting ways.

Let's take a simple example. If you've ever watched IE crash on you (and, c'mon ... we all know it happens), you've also seen that XP will pop up a dialog saying, "Hey! IE died! Can I send Microsoft an error report?" Clicking "send" will package up a bunch of environmental information inside your PC and shoot it off to Microsoft's engineering department.

The pack-it-up-and-shoot-it-off software is called Watson. Watson data streams into Microsoft from all over the planet, literally every second of the day and night. Our engineers use statistical procedures to examine the crashes, determining which are similar to one another (clustering), which are happening with stark regularity (frequency), and which need immediate attention (customer relevance). Using these statistical detection routines, the engineers can now find obscure bugs, fix them, and then roll out the fixes using Windows Update in a very, very short cycle. The product gets better, faster, and the customers get happier. Everyone wins.

The Watson idea - "get the facts, and use them to make better decisions" - is a core part of what we do in Planning. While I'm not a statistician, I have to understand standard deviations, significance tests, sample sizes and the rest to be effective in my job. As all companies become increasingly data-driven (especially true in marketing, which is undergoing something of a revolution right now), your ability to be an informed consumer of statistics becomes critical. As an MBA, you need to be able to look at a set of data and understand what it's telling you. You don't need to be a "numbers person," but you do need to be unafraid of the numbers. Numbers oftentimes can tell us things that our intuitive senses cannot.

So if you want to take something really useful, I second Chris. Take stats. (And thanks again, Dr. Pilcher.)

(And, because I know you're going to ask, the Apple story has two funny postscripts. First, I'm clearly at Microsoft .. but let's just say that I do my work while listening to my iPod every day. And second, Chris is now here, too. He's working on Longhorn.)

Posted by Gavin Shearer at August 12, 2004 9:20 PM. Posted to MSFT.

Comments

Post a comment

« Interviewing | Main | Go East, Young Man »