I can’t remember where I heard this, but it’s been on my mind a lot lately so I wanted to mention it.

When you’re designing something that’s going to be used by a lot of people, it’s vital you talk with some of your customers and get them to test your prototype to be sure you’re building something they actually like and find usable.  

Sounds like a no-brainer, right? Well, it ought to be accepted wisdom in any design-related job, but it isn’t universally so. I’ve worked a few places where talking to actual customers was considered a complete waste of time because customers “don’t know what they want”. (Not Microsoft, so much… most groups I worked in were very conscientious about trying to build products customers liked.)

In software, there is a useful accompanying rule of thumb: if you’re designing reusable code such as a library or an application programming interface, you need at least three different customers using your software successfully before you can be sure you’re anywhere near the right ballpark. If you don’t have at least three, you are just guessing at what a reasonable customer interface and user experience should be. And you will be wrong… you’ll miss something important, or design something the wrong way, or (my most common weakness) overbuild the product because you think they actually want all the bells and whistles that you do. In fact, you should save your energy, stop coding, and wait until you have three customers before moving ahead.

Of course, this isn’t to say that three is a sufficient number for getting a design right. For some products you need millions! But three is certainly necessary.