Random Sampling

PG had this talk and essay about founder mode - I came across a similar version of this when leading an Engineering team / function. I was calling it “randomly sampling”, but the truth it’s not completely random. When leading functions or teams, we can quickly get stretched pretty thin. My objective was to “work myself out of job”, or to lead from the back while maintain control of the direction. In more obvious terms, it was to hire great people, give them autonomy to do their work (opposite of founder mode), but maintain control of direction by measuring inputs, outputs and “randomly” sampling the work.

With 40+ hours in a week, we can’t be in everything and doing everything, we do need to delegate. If you delegate and blindly trust the team to deliver, often with smart people, the outcome is positive. For me, it was setting up systems to know when to look, where to look and randomly looking to confirm that the system was working.

At Convictional, this was controlling inputs with Shape Up through the betting table. We would have the team shape projects. Head of Product, head of eng (me), Principle Eng, and Founder would pick bets = controlling inputs. The team had autonomy to shaping the bet as they saw fit (let smart people do their work). There was a feedback loop before the betting. After betting, they would work on the bets that were selected (let smart people do their work). Then they shipped to production. I had control over outputs in three ways. I did the project management, I some did code reviews, and I QA/used the product. For project management, I could control outputs in terms of resource allocations, hitting deadlines, and managing the team from a people perspective. For code reviews, understanding technical decisions. For QA/used the product, seeing how the shaped project turned into solving a problem for the customer.

I did do all project management. But I randomly sampled the QA of the product and code reviews. It wasn’t totally random if I spent more time on specific peoples code reviews then others based on seniority or coaching. The same was true with watching customer interactions for specific new features.

Ultimately, this is how I manage. Some people micro-manage and are in everything all at the same time. In my opinion, team growth suffers, you become the bottleneck and it creates an unhealthy team dynamic. There are some people who manage and blindly trust. That works until you have a underperformer with really good excuses and it goes unnoticed. That also leads to an unhealthy team because people aren’t pushed in good ways. My happy medium is to sample the work, control the inputs and outputs.