was left in SF without my lifelines: a SIM and a credit card. Was it worth it to continue with my plans for the day? I identi fi ed the risk as low and called an Uber–and then told the whole story to the unfortunate but very supportive driver.
after arriving at the conference, I bumped into the CEO of one of the fastest-growing commercial Open Source products out there, Supabase. We ended up chatting, and we discussed several possible paths we could take together. Overall, it was one of the best networking days.
a product development consulting company for developer tools startups CEO @ AnyCable 5 2x startup founder Research analyst @ J.P. Morgan Investment Bank C++, Ruby on Rails engineer BEFORE
CTO An internal startup that Vladimir Dementyev and I are building. AnyCable allows you to build real-time features–chats, noti fi cations, real-time updates, and collaboration features within Ruby and Rails apps. It scales ef fi ciently and provides delivery and order guarantees, while you keep all data on your servers and save on SaaS costs. The product is an open-source Go server replacing ActionCable. And the paid Pro version is also on-premise and includes features bene fi ting larger applications.
you that risk is the probability of an unwanted outcome. And that risk, uncertainty, probability are models we use when we don’t have all the information about what’s happening in reality. But if you ask me, I will say that risk is not just a model, but risk and probability are true, inherent characteristics of reality. Reality is probabilistic and non-linear. But I will not bug you with that today! I just want to say something important you should know about risk…
take risks all the time. But the rules changed: the past economy was something like a drag race, a straight and simple path to the top. Floor the gas and maximize the speed. Last year this race abruptly turned onto a Formula 1 circuit, with sharp turns and tricky bumps requiring much more precision. You must learn to read the road signs to remain in the driver's seat.
is unique, but let’s look at the most common types of risks and use AnyCable as an example. Market – we cannot build and distribute something people want and pay for. Do people want to use the on-premise services instead of PaaS like Pusher and Twilio? And will they pay for the Pro given that there is open-source version. Liquidity – we run out of cash (even if we have the product people want). Can the money we get from AnyCable sustain our salaries and key expenses at least? Execution / growth – can we scale team, product, ramp up sales and marketing. Will people buy AnyCable when it’s not Vladimir Dementyev personally helping you use it? Technology – can we run into technological limitations? Also, can new technology make our product redundant? What if we want to build AnyCable for Laravel, but it is not technically possible? Or, Falcon or new compilers speed up Ruby to such a level that AnyCable is not needed anymore. Vendor / Platform: we are banned on GitHub and Stripe, that we rely on. Macro – US economy goes into a recession. Legal – ActionCable sues AnyCable for trademark violation because of similarity.
year key risks have grown: - Liquidity is a larger risk because money is much more dif fi cult to raise - Macro risks escalated because of global recession, war, post-covid logistics and production problems - Platform risks play more often - Finally, AI brings new technology risks Now we have recognized the key risks and got a ballpark idea of their magnitude. Is there a more precise way to evaluate risks?
come from? Simple model for SaaS revenue: – new users generate average check, existing users have a certain pro fi le of retention – you spend money on getting new users
purchasing the product) in N months after the fi rst purchase. Normally, you have to start with a model predicting retention. Later, you adjust the model to the fact based on the cohorts you already have.
user acquisition, payroll, legal and accounting services, Cloud and SaaS cost, etc. Make sure to differentiate between fi xed and variable cost. Variable cost should be linked to the number of active users or other variable.
support team size is 0.3 specialists per 1,000 active users (rounded) – cloud and PaaS cost is $0.9 and $1 per active user – user acquisition cost is $30 per new user
as cash at the beginning of the month plus the operating pro fi t (revenue minus expenses) and plus the net change in “accounts receivable”, i.e. money that clients and partners owe us. This is how we get the 4 key parameters forecasts: revenue, expenses, operating pro fi t, and cash at the end of the period.
certain risk can affect our vitals. Take a market risk, for example, the risk that our new marketing strategy fails. We notice it and start looking for a new strategy, and we need to build something new to support it, so in total, it erases 6 months of growth in new users.
Normal people. Afraid of risk. Too slow. Entrepreneurs. Too risky, too fast. You. Driving with precision, like a pro. Evaluate your decisions on the information available ex-ante. Avoid “resulting”.
less: -Cut unessential expenses -Bring your whole team on board -Economize by moving services from SaaS to on-premise, like AnyCable! Improve your fi scal policies SOLUTIONS
Liquidity Execution / Growth Technology Vendor/Platform Legal Macroeconomics … LOW TIME TO MARKET EFFICIENT FULL - STACK TEAM RAILS SCALES MATURE ECOSYSTEM NO VENDOR LOCK