Why many smaller bets might beat a single big one
I've been toying with the idea of creating my own business for a long while now. I have a lot of ideas and even more things that interest me that could be turned into businesses. What I lack, however, is time and perhaps focus to followup on each of these ideas. As soon as I write down the code to solve the problem, I reflect on what it would mean to get this kernel of an idea and turn it into something that is profitable or sustainable, after which I usually abandon the project because uncertainty is too high.
With many of these ideas implemented, code ready and all, it's indeed tough to select which project to allocate more time to, and work slowly towards profitability. This becomes especially relevant, when you understand that many of projects, no matter how good they are, are destined to fail due to a flurry of reasons.
It's estimated that around 80% to 90% of startups fail for various reasons. This means, one needs to prioritize time and money allocation properly, regardless of how cool the idea might be.
The natural question then a solo "founder" might face is: Should I work on one big project or on many smaller projects to have chances at monetary success?
If we take the example of "Levels.io", he started his ventures by deciding just to churn out ideas, and launched a "12 startups in 12 months" if I remember correctly, instead of launching 1 startup in 12 months, so the idea might not be that far fetched.
In this blog post, I'll try to explore potential answers to this question using statistics, powered by a new software I discovered called Squiggle.
The article is extremely simplified. The point is to showcase some of the technical capabilities of Squiggle as well as integrate it into a blog post.
You can of course adapt Squiggle to your use case and situation. I have a much more detailed model I use for financial planning to plan 12 months, 24 months etc, but that one is private. I also recommend you reach out and join the Squiggle Discord (at the bottom of the page). The community is extremely helpful and helped me with embedding the charts in this blog.
Please take the following with a grain of salt.
The odds of success for a single big project
First, let's consider the success rate of a single big project. According to this website these stand between 10% and 30%. Let's take that at face value and not dig any deeper.
To model this in Squiggle, we'll use the following:
// Assuming the big project has a success chance ranging from 0.1 to 0.3
bigProjectSuccess = 0.1 to 0.3
Squiggle provides visualization capabilities, embedded in this page to display the distribution, in this case, a log-normal distribution (takes me back to my Math years):
In simple terms, this means that on average, the success rate for a big project is around 18%, with a standard deviation of 6%, which is honestly very optimistic, but let's roll with it for now.
Combined odds of multiple smaller projects
Let's now do the same for multiple smaller projects. Smaller projects, require less effort and might have lesser chances of success. We'll go with these parameters:
// Assuming three smaller projects with different chances of success
smallProject1Success = 0.05 to 0.1
smallProject2Success = 0.05 to 0.3
smallProject3Success = 0.05 to 0.2
The numbers above give us the individual chances of success of three smaller projects. You can note that the chances of success for these smaller projects are smaller, we assume they can be done "at the same time".
To calculate the chances of success of at least one of those projects, we need to compute the chances of failure of all of them:
// Calculating the combined failure rate of all smaller projects
combinedFailureRate = (1 - smallProject1Success) * (1 - smallProject2Success) * (1 - smallProject3Success)
// At least one of the small projects is successful
smallProjectsSuccess = 1 - combinedFailureRate
The above is valid Squiggle code that we can write as is, and Squiggle will provide all the intermediary (not shown here) and the target plot of the probabilities for the smallProjectSuccess :
Looking at the graph, you can see that the combined success rate for at least one of the smaller projects succeeding represents a whopping 29%, with a standard deviation of 8%. This is significantly higher than the success rate for a single big project.
While this is extremely simplified and you shouldn't jump to conclusions for a flurry of reasons (compounding, feedback loops, etc), it might justify the first intuition of not putting "all the eggs in one bag" so to speak and proceed with smaller bets. This article explains it pretty well too.
It's of course also worth mentioning, that smaller projects might also have smaller returns compared to bigger projects, but this balances out with the fact that a successful small project might provide a confidence boost that a bigger, longer project might not yield right away (or before money runs out). On the other hand, fragmented attention might be a negative thing which would then speak in favor of one big project. It's all about assumptions and your personal situation.
You can try the above on your own, with your own assumptions, on the Squiggle playground. An interesting idea to explore would be to factor in founder experience (as a multiplicative factor), factor in the time available for working on the project etc. Squiggle enables you to make such smaller assumptions and multiply probabilities together quickly.
This approach is the reverse of taking data and trying to compute functions that map it best (classical ML), but rather starts from small rough assumptions and build up knowledge (with confidence intervals). Checkout more models here.
I just love it.
Conclusion
While this was a very simplified example, aimed at showcasing the power and flexibility of Squiggle, the data does seem to suggest that diversifying efforts across multiple smaller projects can significantly increase the chances of startup success.
It sure might be very tempting to work on a single, game-changing project, but the numbers advocate for a more diversified approach, especially for first-time founders or founders in countries with little access to venture capital to pad any runway. When outcomes are uncertain, and sometimes even random, quantity might just trump quality.
As with all ventures, it's essential to balance passion with pragmatism and let the numbers and the data guide the way.