You build three machines when you build a startup. Your ability to build these three machines is the Great Filter to your life in the business universe. This post is an effort to describe why some startups fail, why some are small, and why others grow big.
The Great Filter
The Great Filter refers to a concept that Robin Hanson came up with to explain why we don’t see any evidence of intelligent life in the Universe. One can get a better sense of different scenarios when one considers how many things need to be true for intelligence to emerge, and assigns probability to them. If it’s the case that the coincidences required for life to occur are exceptionally rare, then it’s very probable that we’re all alone. If it’s the case that life is very common, but the coincidences required for intelligence to emerge, then it’s very probable that that we’re all alone. If it’s the case that intelligence is pretty common, and it survives long enough to develop the technology to leave its gravity well, then it’s very probable that we aren’t alone, and even more terrifyingly, there’s some intelligence out there that’s killing off competing intelligence. Is the Great Filter behind us? Is it ahead of us? Or have we cleared all of them and we’re on our way towards the stars.
The Founder Great Filter
What are some of the Great Filters for building an intelligent startup in the business universe?
Angel.co lists over 4.4 million startups. It’s very likely that there are more. For every Google, there are millions of startups you have never heard of and will never hear of. Ever. Several hundred thousands of those startups grew into businesses which still add value to this day. Several thousand were acquired or enjoyed a major liquidity event.
The Three Machines
I learned of the idea of three machines from the host of an a16z podcast. A founder has to be able to build and run three machines. Their ability to build these machines represents at least three great filters.
There’s the machine that delivers value to customers. It’s the technology that creates value for some group of people who are willing to exchange money for it. What’s amazing about software is that once it’s built, and does something useful enough for people to pay for it, it can become very efficient run that machine for value.
Then there’s the machine that builds that machine. This is all the systems that go into supporting the social technology that builds the core machine. It’s the recruiting, the hire, the payroll, the furniture, the space, the policies, the culture and the activities that go into building that machine.
And there’s the machine that lets everybody know about the machine you’ve built. These are the systems that let the Total Addressable Market know about the value that you’re creating.
The Machine That Delivers Value
There’s a machine itself that delivers value. There are often pieces of the machine that are just required to be there for the system to deliver and capture value. These might include payments, authentication, and deployment. Then there are pieces of the machine that are at the core of the value delivery. These might include the application of learning, basic operations like CRUD, or sensory instrumentation.
The primary cost driver of the machine that delivers are the servers and infrastructure it’s required to run on. Software and hardware needs environments to run in. The costs of most of those environments has really come down over the past twenty years. And hopefully those costs will continue to come down.
The Machine That Builds The Machine
The Machine that builds the machine is made of people and the culture they create. It’s a practice in systems integration and in executing a creative idea. It’s staff recruiting. It’s customer support. It’s accounts payable. It’s the travel policy. It’s real estate.
It’s also how engineers collaborate with each other and others. It’s the relative prioritization of multivariate testing against the dead reckoning of an ego. It’s how people respond to adversity and how victory. It’s the relative priority of people, product, and profit. There are hundreds of decisions that go into building the machine that builds the machine.
The Machine That Lets Everybody Know About Your Machines
The machine that lets everybody know about your machines (all three of them) is vital. You may have a machine that delivers outstanding value, but it’ll never realize that value unless an addressable market, people who want that value, know about it.
Running this machine well, and efficiently, is vital. The Customer Acquisition Cost is a core factor in calculating Customer Life Time Value (LTV). LTV is a leading indicator of profit, valuation, and firm survival.
The Three Great Filters
Can the founder build the machine that builds the machine?
Building the machine that builds the machine involves a suite of skills. There’s a set of skills around the definition of value. What is value? To who? What represents an irresistible value, to who? How can that be known? What is the smallest possible machine that can deliver value? Who is needed to build that machine? Can it be built? Can the founder rally the resources necessary to build it? Can they convince others to join their conspiracy? Can the founder found a team? Can the founder team? Can they trust them to do things they’re not experts at?
Can the founder build the machine that delivers value?
Can the founder’s team build the machine that delivers value? Can they get through the technical challenges and setbacks? Can they learn? Can they learn to scale that machine if the market demands scale?
Can the founder build the machine that lets people know about the machine?
Can the founder let people know about the machine? Can they build a marketing team? Can they trust the marketing team? Can they build a sales team if they need to? Can they convince somebody to build them a sales team if they need to? Can they set up a disciplined deal desk? Can they found a team? Can they team?
If hard work was enough, there’d be more blinding successes. Hard work alone simply isn’t enough. The kinds of intelligence that’s required to build these three machines are varied and the standard is high. It isn’t enough to simply know how to do things, one has to learn, rapidly, how to do new, and often, unexpected things.