Staff Reports

When you start a company doing something new, you know nothing. Not only don’t you know anything, but you are also intoxicated with the possibilities of a new venture, and your thoughts are even more irrationally exuberant than usual. You are wrapped up in your fantasy world where you believe everyone thinks the same way you do, interacts with software the same way you do, and wants the same things you do.

Once you have stopped your brain from spinning wildly, you can start making progress. Progress begins with measurement, and there are three obvious ways to go about this: talking, software analytics, and user observation.

The first thing you do is talk to people about your idea. This is the easy part, and it usually comes before you have funding or anything like that, maybe before you are even committed to starting the company. You need to find people who will listen to you prattle… anyone will do for this, including your mother and close friends. You have lots of ideas floating around, so you say them aloud, people comment, and you begin to refine your idea and the narrative around your company. You start with whoever will listen, but as you begin to understand what different sorts of people will tell you, you seek out stakeholders to talk to – people who might be actual users, investors, and employees of your company. You talk, you refine your ideas, but most importantly you listen.

After you get all the talking done, the real data collection begins… the other two ways to measure data are the hard part, but that’s where you start to find real answers. This means that you need to have people use your product, and 1) measure their usage with analytics (i.e. Google Analytics, Pinch Analytics, custom Python scripts), and 2) watch them use it and talk to them about what they did.

But whoa there! Where’s the product? How do we measure usage of the product if we haven’t built it yet? The answer to that question is the primary point of this blog post – you need to build a product, as simply as possible, and as quickly as possible, as soon as you get done with the talking. The point in keeping it simple is not because it’s easier to build or that people will like it more… the point is to keep it simple enough to measure clearly.

There are a lot of ways to waste time at a start-up. You can putz around with logos, PR, brochures, blogging, and while all of these things have their place, it is not at the beginning of a start-up. The only thing you do at the beginning is measure and iterate. The cycle looks like this:

• 1) Founders talks to friends and potential stakeholders
• 2) Founders builds steaming pile of dog feces in 1-7 days
• 3) Founders measures interaction with software
• 4) Founders changes software
• 5) Repeat steps 3-4 until millions of people praise your software and you are rich

Everyone who reads Hacker News thinks they already know all of this. I am just describing Steve Blank’s customer development model, the lean start-up, and all of those other terms that we throw around. But I often think people miss the point of why we do it this way. It’s not just that simple products are easy to use and get right, it’s that they allow you to measure. If you start out with complicated software and begin measuring then, there’s a good chance you’ll never find the right metric and never make any progress. People spend a lot of time talking about the minimum viable product and how that’s the right thing to launch. But before you even start thinking about a public launch day, you need to have a really minimum product. It shouldn’t just be embarrassing… it shouldn’t work properly or do much of anything.

Let me give you an example. Let’s say you want to build a Facebook role-playing game. You have a great idea for a role-playing game like Mafia Wars, but with a financial theme where all players are corrupt business owners and stock traders. Where do you start?

I would say that the more complicated your game is to start, the worse off you’ll be. For this sort of thing, that means you probably want to build a 30 second-game experience with zero art, zero sounds, and zero Facebook viral elements. It should just be like a 30-second Zork experience. For those of you that don’t know, Zork was a pre-cursor to World of Warcraft where game play would proceed roughly as follows:

• Terminal spits out: You are attacked by a ghost. It haunts you for 1 point of health.
• You type: I attack ghost with sword.

The reason to keep your game short and build almost zero game play initially is so that you can start measuring. You have a noise-free baseline from which you can iterate. Now that you have a “game”, you can add in a picture of the player. Does that affect retention? You can add a sound for swinging your sword. Does that move the needle? You can add a viral message to notify your friends when you kill the ghost. Did that help?

If you had started your game with some sounds, some art, and some viral messaging, and it started out reasonably well, that may seem good, but do you have any idea which of those elements is the one getting you traffic and retention? Probably not. It is much preferable to start with something that doesn’t work at all and find some little smidgeon of goodness, than to measure some mediocre game hoping for a signal to emerge.

Simplicity is something I thought I understood when I wrote my first website, but I now realize I almost totally missed the point. All of my software projects to date have started off too complex, but that won’t happen again. I won’t muddle through mixed signals and waste countless hours building things without reason. I’ll start simple and measure, measure, measure, and I suggest you do the same.

The true strength in simplicity is clarity of measurement, which gets you to insights that no one else has, and in the end, those unique insights are what start-ups are really about.