“The hard part is building the machine that builds the product.”
Successful entrepreneurs like Dennis Crowley and Mike Karnjanaprakorn at SkillShare have pointed that building a great product is only step one. The next, harder step is building the machine that builds the product which turns improvement into a repeatable process.
Mark Pincus, discussing his experience of growing Zynga, observed that products can be built “through the strength of your personality and lack of sleep,” but that doesn’t scale — and soon “you’ve got to find some way to keep everybody going in productive directions when you’re not in the room.”
To Ray Dalio, founder of Bridgewater Associates, the world’s largest hedge fund, what’s required is “higher level thinking” on the process of how to build product. It’s the ability to step back and take the perspective of someone who’s looking down at your company from an objective standpoint as if the company were a machine. The machine consists of people plus design, with goals as inputs and outcomes as output.
For Bridgewater and companies like Google, this means an obsession with tracking and data to peer into the black box that is the machine. Goals go in, outcomes come out — what happens in between? Without tracking and data, there are no log files to comb for debugging the discrepancies between outcomes and intended outcomes.
Dalio would say that it’s about the search for truth. I’ve talked to a number of startups that do a daily standup or scrum with the idea that it’s an iterative process that provides for improvement by rapidly addressing and resolving issues as they arise. It’s often a young company’s first stab at formalizing a process. But without a log file of what happens in a company, I’ve observed that issues inevitably get addressed reactively, on a case-by-case basis.
A record that shows how the machine works makes it possible for higher-level pattern matching to identify systemic issues plaguing the company — and this makes exponential progress, not just incremental progress, possible.