Warren Buffett said “Risk comes from not knowing what you’re doing.” We’ve all experienced that feeling of immense ignorance as leaders of software businesses. One area of constant anxiety is not really knowing how adjusting or changing one component of our business will impact the whole. Software businesses contain an incredible number of moving parts and variables. How can we isolate and study them? For most of us, it’s a combination of trial and error and learning from others’ mistakes when they’re brave enough to share them. Matt has spent the last 10 years starting and running his own software company (Riskpulse), learning many hard lessons along the way.
This led him to wonder: in addition to stumbling forward and searching for mentors, could he write software to simulate the future of his business? What lessons could it teach him? What could he learn if his computer could run 1,000 different startups in minutes, not years? What light could a simulation like this shed on the underlying laws (physics) of software businesses? About people?
Get on the BoS Conference Mailing List
Be first to see new videos, join free speaker hangouts, stay up to date on BoS Conference programmes and a regular dose of thought provoking ideas.
Unsubscribe anytime. We will never sell your email address.
Matt Wensing: Hi. I’m Matt Wensing and in the next 7 1/2 minutes I’m going to share what I learned by building a startup simulator. First a little background on me. In 2008 I started a weather website called stormpulse.com with a friend and through a ridiculous chain of events it ended up in 10,000 businesses and became profitable, and even appeared in the Obama era White House which is how I’m living in the only multiverse where I got to bro hug the 44th president of United States. Sorry Mark, politics.
So with the success I said Hey let’s go to Sand Hill Road and raise money. That’s what an entrepreneur would do if you feel successful. And I did that and it turned out that things got a little weird just like with Charlie in the factory. They didn’t like our sales cycles and the reason was they were lengthening. So we had six to nine month sales cycles and this V.C. said what if they get even longer? And that was a fair objection but it brought up a lot of questions which was how much money do you actually need? What are you going to do with that money? What happens if this trend continues now? A quick detour to the data that led him to think this way. Here’s a chart of our maximum contract values over the last 10 years. This is where we are today. But the trick is this Y axis is actually logarithmic so this bar here is actually forty eight thousand times taller than this bar – which means that you have a nonlinear and very scarily pace of change business. Or if you prefer the modern art view, this is us now and this was this tiny little dot was us back then. Which means that I need tools to help me figure out how this business is going to grow.
The tools that we have today are mostly made from an era of manufacturing as I like to call it it’s, hey next month’s going can be about the same as this month, maybe a little bit of change – its linear production. And as we all know by running software companies software businesses especially startups are non-linear and they’re complex meaning if you change one small thing about the system (as Jeff Goldblum taught us all) chaos can ensue and you really don’t have control you think you do. So you pull a lever and that thing over there moves and you say I didn’t even know that was what was going to happen. So as a founder I thought to myself OK I can’t do this in Excel… I can do it in Excel but it feels dishonest because I can torture the data right. I can make it say whatever it needs to say. So where do I need to go where’s my canvas? And I thought you know what I really need is a working model that’s outside of my head. It’s there it represents everything I know about my business but it can keep me honest right. It can tell me what’s going to happen and maybe can even surprise me. So I turn to simulation.
Now simulation is a little bit different – you might think about pattern recognition. This is where most of us are when we’re in excel and we extrapolate. And if we’re fancy we might even do some kind of exponential extrapolation. But it’s still just saying I’ve got metrics and I’m going to do a forecast. The key takeaway here is that in something like a simulation, which you can do in Python if you want to which is what I did, the output can surprise you because you don’t actually control it – what you control is what you control as a founder which is your business model. You’re trying to take uncertainty out of your business model and then all of your metrics are output from there. Now it might seem weird to think about simulation in terms of business but in other domains this is perfectly normal, like Meteorology which I happen to know a few things about. This VC, I I’m sure he’s a great guy but I’m to say Leo you’re wrong. Meteorologists are not guessing. If they were guessing they’d say the hurricane is going to go over here. That’s a linear extrapolation. This is a statistical model a little bit more fancy. Modern meteorology is actually running simulations than interpreting them which is how you get accurate forecasts. So what if we had a simulation for startup-land – what would that look like?
There’s a growing body of evidence that startups actually do have some kind of physics underlying them. The units of growth of the team sizes that we have, how quickly does headcount scale, change probabilities – a growing body of evidence that there are these governing equations if you will that just like with the atmosphere govern how things grow especially with companies and networks, and we’re all familiar with the two pizza box rule, right? And so the question becomes is there enough science here for me to have fun this summer and build a startup simulator? I decided to try that out and built something a little bit like Sim City if you will so we’re all familiar… well I hope you’re familiar with this. If you’re not you missed out on your childhood! Governing equations. So they need more industrial. You know human sales reps have finite capacity, distribution channels eventually saturate. These are not formulas like CAC/LTV/MRR all that stuff. These are entropy – conservation of mass, the laws that we have to obey if we’re going to succeed and the things we talk about when we come to conferences like this. These are hard lessons 1 through 10 we should learn them.
So what if we put that into a simulator like this? And you know what we can do is since it’s a simulation and it’s not excel we can be more honest here. We can say look I don’t know what my contract values are really min/max I might want to try out something new maybe my sales cycles are going to lengthen maybe they’re not I can leave things blank that I don’t know. And I can pull this just for many common metrics that I may have and if you’re too early to have metrics you can just punch in guesses right and the simulation will figure it out for you. So I said All right. Why not? Let’s run five simulations based on this input data these parameters and see what happens.
Well it turns out that this company five simulations reach break-even around 14/18 months from now. But then what in the heck happens here. I’m guessing that if you did this in Excel you would not have a dive in net income at month 20. But what turned out is that startup is doing so well that they had to hire another customer success person and engineer to take care of all the customers that you are getting right. So if you’re planning on a lot of profit you might be surprised. All right. Surprise number two. I went back and I asked myself ok what if our sales cycles did increase to 15 months. Well turned out it’s not a pretty picture. Good news is we ended up more like this. Things went well our sales cycles actually shortened but in this world where the sales cycles actually kicked in and about 15 months you’re burning a hundred twenty thousand dollars a month in three years from now and you probably aren’t able to raise money and people are quitting and it’s just a terrible place to work. So small changes can have a big effect right. Jurassic Park.
Next thing I ask is so why don’t we run this on two thousand startups right. Let’s do eight thousand years of business time and I learned that reality is the number one cause of death of unicorns globally. What I did is I introduced a flaw into the sales teams of the test group where they would just forget to follow up on things every once in a while. Right. This is human nature right. So turns out that without perfect humans your valuation is in trouble. Your horn gets sawn off. The rest of us are just struggling with unit economics down at the bottom somewhere. Then I said OK what else can I learn from this data? And it turns out that there’s one thing that is constant. Whether you’re a unicorn or you’re a small company is that the number of first impressions you get to make in a target market is constant. Right because there’s only so many people at target market which means that you as a founder, your number one job is to scale trust. And you might do that through design you might do that through referrals. You might do that through telling a story about how you’re going to save the world and move us all to Mars. But how do you do that. You get to make a first impression and scaling trust is your brand. The set of things you choose to focus on is your brand now.
I could go through and list off all the cool fun things that I took away from this little experiment. But the number one thing I wanted to share with you guys is this seems to corroborate and validate what I think is one of the central themes of BoS right. It’s that in this simulation, the more honest we can be about what we don’t know, the better we can be as business leaders because we can make more humble and thoughtful decisions. Right. And it gives me hope for the future. I thank you guys for listening and I hope you enjoyed.