Systems thinking, incentives, and picking metrics

Recently, I read "Thinking in Systems" by Donella H. Meadows after hearing several folks recommend the book in different forums. I’m a loose believer in The Lindy Effect, essentially a saying that "things that have been around have been around for a reason," and the fact that several people recommended a book from 2008 (not that old at all, but also old enough to survive a few thought leadership hype+crash cycles).

Possibly because I’ve been fortunate enough to work in roles where I build and operate systems for most of my career (multimillion-person developer communities, ultra-high-scale infrastructure), I devoured the book and found myself nodding along as I thought through Meadow’s stances. Specifically, the core assumption is that healthy systems are not static — they are alive and ever-changing. Operating within a system introduces fractional changes and microscopic variabilities, which can be detrimental if not actively addressed or can slowly start a step-function evolutionary improvement if adequately nurtured. To over-control is just as harmful as under-controlling, and the reality is that this over/under window can gradually widen or narrow over time. Moreover, Meadows addresses the brutal reality of "sunk cost fallacies" within systems that have become crystalized within their own bureaucracies — sometimes, a system has become so bad that it needs to be completely torn down and rebuilt to drive the corrective actions necessary.

The book also discusses a long-held rule in the modern business world: "Incentives Matter." Charlie Munger explains this in another way: "Show me the incentive, and I'll show you the outcome." Individual players within a system respond to their direct incentives, and often only to their direct incentives, even if they are counter to the long-term, sustainable success of the broader ecosystem, nation, or company.

Some players and organizations can operate with outside, higher, or secondary systems in mind. However, these teams and people are the rare, selfless, brilliant saints we meet all too infrequently. For the rest of us mortals, we need to find ways to ensure that the systemic success within our immediate control ties to the greater success of "the whole," whatever that "whole" may be.

In an episode of Lenny’s Podcast from earlier this year, Doordash’s VP of Analytics stressed the importance of picking the right "proxy metrics" — metrics that are directly within a team’s control, but also can support the greater mission of the organization. That way, a team can fully control their destiny while knowing they support the higher strategy. An example that comes to mind is if a company’s strategic goal is to drive product adoption, a specific team in charge of landing pages may focus on just getting interested customers to sign up and kick the tires with said product. The landing page team may never control pricing, enterprise contract negotiation, or even if the product works properly. Other teams do that, and quite frankly, the landing page team has to ensure that senior management is making sure that those teams do their part of the process correctly. The landing page team fretting over contracts, product quality, or other components outside their control is a waste of effort — it’s arguably net negative work as this fretting distracts them from their core responsibility (driving signups and inquiries).

So, do yourself a favor. Read the book, and let the ideas wash over your brain and reshape how you think about the systems you directly operate, the systems you build (even the tiny ones matter), and the broader systems that surround us without us thinking about them.