A Probabilistic Approach in Decision-Making
The most critical aspect of running a company or product is deciding on capital allocation, regardless of monetary, time, or human capital.
In the frenzy of pursuit for disruptions, it’s common to over-invest in projects that seem promising initially because the specific market runs hot. While sometimes running experiments in the trending market is necessary to sustain organic growth, delivering shareholder value - long-term company value appreciation is the ultimate goal.
A probabilistic approach
Sharing a few notes combining both my personal experience and the learnings from The Outsiders: Eight Unconventional CEOs and Their Radically Rational Blueprint for Success:
1/ Evaluations should be dead simple enough to be done on a napkin, instead of financial models under the veneer of complexities - In short, if the evaluation is done with financial models, there’s a great chance that it goes against the final result. Any slight changes can drastically drift the reality off the simulations. Especially in the post-COVID era, changes ripple through the whole system unexpectedly and quickly invalidate previous assumptions, e.g. interest rates, inflation, or supply chain disruptions. Assessments built upon a few observed principles that withstand the test of time are more likely to bear fruit despite the economic/political changes.
2/ Assign probabilities rationally to a handful of variables - Not all assumptions weigh equally. Probabilities assigned based on historical evidence would easily make initiatives comparable when referenced along with project investment size.
3/ The success of initiatives should rely on as few premises as possible - Occam’s razor evangelizes hypotheses that come with the least assumptions and are more likely to be valid. The inversion would mean assessing areas of potential investment returns based only on a handful of premises. Imagine there’re two projects to choose one from and assign a team to work on:
a. One requires securing several industry partnerships to bear fruit. The success is realized only when ALL or the majority of the target institutions join, assuming there are two in total. This applies to products working on a protocol level, e.g. SWIFT network, blockchain protocol, or Visa). Assume the probability of each joining is identical - 20% and is independent, then the total probability of joint occurrence is 20% * 20% = 4%.
b. The other one requires buy-ins from the product’s customers, facing any customers with similar needs. If the success is defined by at least one customer paying, enough to cover basic costs. Let’s assume only two customers exist in the targeted market. The adoptions of the product by the customers are not mutually exclusive, and each has a 20% chance to pay for the product independently, then the probability of the adoption by customer A or B is
P(customer A adoption) + P(customer B adoption) - P(customer A and B both adopt) = 20% + 20 % - (20% * 20%) = 40% - 4% = 36%.
The above would partially explain why the 2nd type of products or services accounts for most businesses while the 1st type is usually under cooperative entities or associations, such as FIDO Alliance, SWIFT, etc.
Essentially, each premise added to realize an outcome is a tax to be paid and dilutes the probability of the desired result.
What to work on > # of hours worked
Any major initiative - new product line or business line requires coordination and communications with stakeholders, engineers, marketing, and customer support to execute. Picking the wrong battle means $ and time wasted on payroll, effort, internal credibility, and the revenue/usage foregone from the alternatives. It’s much easier to build upon the existing foundation, thinking about what to optimize for new users and retain the existing ones instead of engineering a turnaround.
Iterate - cut fat to stay lean
Despite the best effort, not all goes as intended. Products with depleting runway and ongoing costs - product design hurdles, server, or staffing costs could become a cash drain. Closing down the underperforming products would free up cash and staffing for the alternatives that present a great chance of thriving; this resembles the self-cleansing mechanism in the market-cap weighted index like S&P 500, in which companies positioned for growth or profits will replace the declining ones. In the long term, we can expect to build a product portfolio that comes battle-tested and with proven revenue streams.