I never really intended to start blogging about my work kitchen, but it keeps providing me with anecdotes. Today, I was thinking about the milk. There are six fridges in the office, and typically you’ll find that they each have eight semi-skimmed jugs of milk. There’s also a single jug of skimmed milk. Sometimes there’s none. So, someone somewhere has made a decision as to the appropriate amount of skimmed and semi-skimmed milk to provide. Now we enter the realms of pure speculation, but bear with me.
How do you determine what the mix should be? Well, the obvious googley answer is: you measure consumption and use that as feedback. I’m pretty sure that’s what someone at the supplier is doing. The thing is, there’s an absolutely tiny amount of skimmed milk being supplied, far less than you’d expect given a glance at the supermarket. So what’s wrong?
Ask anyone who drinks skimmed milk at the office and they can tell you: the skimmed milk tastes bad. I mean, really bad, so bad you think it’s off. Typically, the first time you try it you pour it down the sink. The second time you realize it always tastes like that. Every so often you see someone new do the exact same thing.
Selling Ice Cream In Winter
Famously, if a government department wants to kill something, they’ll run a study under circumstances that make it look incredibly unpopular. But even when someone isn’t deliberately trying to manipulate the figures, unexpected factors occur. The good news is, for user interfaces, there’s an easy answer: learn a bit about usability testing. With internal systems the problem is harder. Worst of all is with performance questions. Just because you’ve reduced one number, doesn’t mean you haven’t increased another. Ultimately, you need to understand your numbers holistically if you’ve got any hope of doing something other than performance whack-a-mole.