I love Jeff Bezos’ annual letters to shareholders. I read them all, even published an article with the key highlights from 1997.
Lately, I have been thinking about something Jeff Bezos mentioned in his 2005 edition, where he talks about two types of decisions in the day-to-day job. One can frame the decision-making process in the chart below:
Bezos elaborates on the following in his letter:
In short, he asserts that when decisions are not costly and reversible (what he calls “type 2 decisions“), you should make them quickly, with simple processes, and delegate as often as possible. And corollary, you must focus on taking the right “Type 1 decisions“, that are consequential and (nearly) irreversible.
To me, there is a major issue with this framework. What’s underappreciated is the fact that all these type 2 decisions are slowly accumulating over time and have a strong effect on organizational cultures. This is what economists call path dependency, where self-reinforcing mechanisms are at play; that explains why sometimes suboptimal standards end up dominating better alternatives (think of your keyboard layout).
I talked previously about how important adaptation was, especially in times of major changes (random illustration: a pandemic), and the role of culture in that matter.
Decisions might seem trivial, but we should not ignore that they can spread by virtue of signaling (“this is the type of decisions we make here“).
This process can lower the level requirement at a given time (“I’m not sure I’ve put enough work behind it, but XX has taken a similar decision on such a light basis“) and even lasts and strengthens overtime. One of the worst areas where it can happen is hiring — with a bar to entry progressively lowering (and processes that are not preventing it).
It can also slowly but surely alter the design of an organization.
Let’s dig into that by taking an illustration from a previous edition, talking about client-centricity for financial products.
Starting a fund means securing capital to be able to invest. Yet, not only the first few commitments aren’t easy to get (it’s a leap of faith to invest when you are the first believers), but also most institutions do not want to exceed a “control ratio“ (eg: more than 30% of the fund). This means that even the first believers may require to find other believers because they cannot represent a too high proportion of the fund you deploy. So the natural temptation for a new team launching a fund is to fall into the trap of customizing the product too strongly towards one side — capital.
The model may be optimal just for the first few steps, but not for the long run. Yet without constant effort, it will just solidify and won’t be able to be adapted overtime (too many resources allocated, habits formed, promises made, etc.). This is path dependency.
It’s possible to avoid falling into that trap. But ensuring an always-dynamic equilibrium isn’t easy and it requires a lot of intentionality.
This dynamic equilibrium requires to be very aware of the feedback mechanisms that could transform harmless, one-time decisions into a long-lasting equilibrium and a colossal induced resistance.
It requires interrupting the slow escalation when it feels wrong (all these little details that make a situation unacceptable, but which are each tolerable if they are isolated exceptions).
It requires designing elements that are balancing, if not conflicting, each other. One of the usual ways to counter this is to have competition between teams or individuals. This can also be done in a more collaborative way, by assigning different people responsibilities that are complementary but in tension, exactly like in a board meeting, where a CFO and a CEO can have very opposite views on certain investments. You can also monitor certain effects by pairing indicators, as advised by Andy Grove the former CEO of Intel, so that together both effect and counter-effect are measured. “Thus, in the inventory example, you need to monitor both inventory levels and the incidence of shortages. A rise in the latter will obviously lead you to do things to keep inventories from becoming too low.“ (High Output Management, A. Grove)
It also requires constant check-ins and synchronization, to make sure the decisions are made the right way and the culture grows healthy.
All this is not to say that we should get overwhelmed by any decision and its potential impact. It would make life and work unbearable. Yet, we should be very cautious when defining something as trivial. Some decisions must be made quickly. But these so-called Type 2 decisions must exist in an environment that prevents the bad ones from spreading.
Path dependency could be a curse if not handled. Happily, you have all the cards in hand to build your own, desirable road.