Jeff Bezos’s 2016 Letter to Shareholders: A Skeptical View of Proxies
In 2016, Amazon CEO Jeff Bezos published a 4-page letter to shareholders, packed with business insights. While the letter focuses on Amazon’s ideal culture (think: “How we do things around here”), its relevance goes beyond business. Bezos’ letter contains helpful ideas for any organization that’s striving towards a goal.
Bezos’s letter compares two types of organizational cultures (“Day 1” and “Day 2”), with nature’s biological cycle of growth and decline acting as the business metaphor.
Day 1 companies are healthy and full of life. Day 2 companies are plagued by stasis; they soon find themselves irrelevant, followed by a slow death.
Thus, it must always stay Day 1 at Amazon.
Day 2 is the Default Outcome
History tells the tale of countless companies that strut their hour on the stage and are heard from little more. For example, out Fortune 500 companies from 1955, detailing America’s largest businesses, only 50 made the list in 2020.
When businesses constantly face brutal market competition, if they don’t evolve, they won’t survive. That’s just capitalism.
Unfortunately for organizations, irrelevance and decline is the default outcome—given enough time. Yet, Jeff Bezos gives us a few prescriptions which might help stave off Day 2’s peril.
How does an Organization keep Day 1’s culture?
Bezos gives us four approaches to maintain Day 1’s vitality:
Customer Obsession
A skeptical view of proxies
High velocity decision-making
Each of Bezos’s four prescriptions can help organizations avoid stasis. Whether you’re part of a sports team, non-profit or a business, Bezos’s warning, to be careful with proxies is helpful. It serves as a reminder we should be careful relying too much on simple summaries for decision-making.
Maintain a Skeptical View of Proxies
“As companies get larger and more complex, there’s a tendency to manage to proxies. This comes in many shapes and sizes, and it’s dangerous, subtle, and very Day 2. ”
-Jeff Bezos, 2016 Amazon Letter to Shareholders,
Proxies: The Good
In an organization, proxies are numbers, figures, charts or processes that decision-makers often rely on to help guide future outcomes to favorable results. They’re typically easy to generate and use, which is part of their value. In an idealized world, there would be almost no daylight between the proxy and the broader reality they’re meant to represent.
A proxy can take many forms. For example, in the sports world, a baseball player’s batting average, runs batted in (RBIs), and stolen bases are all proxies for that players contributions to winning, in the same way a quarterback’s completion percentage is a proxy for their ability to put points on the board.
In the business world, the number of Amazon Prime subscribers could be a proxy for Amazon’s brand strength, or whether the customers are truly “obsessed”. For new Amazon products, beta-testing results from an initial study could be a proxy for how well the product will later be received by the public.
Proxies can even be routine processes or slogans everyone in an organization operates by, like emphasizing being the “lowest cost operator”, focusing on offering “Everyday low prices”, or optimizing for “customer obsession”. At best, these kinds of proxies act to serve the business, and not the other way around.
Ultimately, it’s up to decision-makers get to come up with which proxies to assign value to based on their own philosophies of what matters most. Proxies can also help align leadership and the rest of the team, so everyone can focus on the same goals.
But if we’re not careful, proxies can be taken too far. When decision-makers mistake proxies for gospel, forgetting to check occasionally what’s under the hood, trouble awaits.
Proxies: The Bad
As companies get larger and more complex, there’s a tendency to manage to proxies. This comes in many shapes and sizes, and it’s dangerous, subtle, and very Day 2.
-Jeff Bezos, 2016 Amazon Letter to Shareholders
Proxies are substitutes for more rigorous analysis that can make a complex story a bit too simple. When taken too far, proxies can cause an organizations’ decisions to become too disconnected from the underlying reality. The risk is that the organization shifts more focus to proxies than outcomes.
Decision-makers, who aren’t skeptical of proxies, will inevitably be mislead by their analysis. These tools are merely a starting point. You need to check under the hood.
The Bad: Proxies in Baseball
For example, let’s go back to baseball. Before the 21st century, Baseball conventional wisdom considered batting average, stolen bases, and RBIs as among the top proxies for determining a player’s value. These proxies were meant to help teams determine which players to pick and how much to pay them. Therefore they received a lot of attention.
But it turns out these metrics were somewhat misleading. Rigorous statistical analysis, notably by Bill James, revealed that a player’s slugging percentage or on base percentage were more reliable in predicting a player’s future contributions to a teams runs, and therefore wins. For decades, despite plenty of financial resources, teams had been emphasizing outdated metrics and building organizations on suboptimal proxies (this is the story Michael Lewis writes about in Moneyball). This was very Day 2.
In the end, the older proxies in baseball, remnants of an older game, gave way to the new—advanced data analysis in the form of Sabermetrics.
The Bad: Proxies in Business
In business, take Wal-Mart at the turn of the 21st century. Wal-Mart arguably maximized their proxy, through the process of making “Every day low prices” happen (in retail stores near you). They were wildly successful at achieving this end. And entering the 21st century, it was Wal-Mart that was best positioned for the coming e-commerce trend.
A common example is process as proxy. Good process serves you so you can serve customers. But if you’re not watchful, the process can become the thing. This can happen very easily in large organizations. The process becomes the proxy for the result you want. You stop looking at outcomes and just make sure you’re doing the process right. Gulp.
Jeff Bezos, 2016 Amazon Letter to Shareholders
But while Wal-Mart’s processes and culture emphasized getting “Every day low prices” a mere driving distance to millions of customers, Amazon’s processes and culture focused on a somewhat different end—the customer experience.
Further, Amazon’s obsession about the customer experience (Free 2 Day shipping, intuitive online shopping, helpful online product reviews) wasn’t anchored by a “you drive to me” business framework.
There was nothing intrinsically wrong with the “Every day low prices” retail strategy that came with parking lots—its been massively successful. However the model needed a facelift to catch up to the e-commerce reality and consumer trends.
Thus, Wal-Mart’s focus on getting “Every day low prices” in local retail—as a proxy—may have diverted attention (early on anyways) from the emerging e-commerce business model which customers now love.
Maintaining Skepticism of Proxies
Day in and day out, who can really leave no stone unturned?
When making decisions, proxies save time, which gives them value. With these desirable tools, a decision-maker is saved from suffering through tediously checking every little messy detail, one by one.
Thus, Jeff Bezos recognizes proxies have value in saving time when making decisions, but he advises we should maintain skepticism, lest they lead us down the wrong path. Proxies are just starting point for analysis, but if we’re not careful, they can become a thing.
For anyone looking to avoid being seduced by proxies, perhaps set up a forced daily, or weekly routine, that makes you occasionally spot check the smaller details or pieces of information that make up a proxy. Also, use several proxies at once to better triangulate the underlying story.
Part 3: Eagerly Embrace External Trends
Author’s Note
A proxy is a debasement of the underlying more complex reality, and our analysis of what the proxy means is a debasement of the proxy itself. So there are two causes of informational loss when utilizing proxies. Thus, proxies end up debasements of debasements!