The perils of a writing culture — the Amazon hammer to decision making.
“We need a doc on this, as always”
I had the good fortune to work at Amazon for ~8 years (’12 thru ’20), and the thing that left the most impressionable mark was its’ culture. Not a place for everyone, but a daily masters’ class in creating and fostering highly-leveraged work driven by its leadership principles. But I am parting ways with its most prized cultural trait — one that I deeply evangelized— writing.
Those not familiar with Amazon’s writing culture, it is the only tool employees have to sell their ideas internally and then bring them to customers. Nothing is built unless there is a document that backs it. The specifics of a 6-pager, PR/FAQ or other formats used at Amazon is not as important as the “why” behind it. Jeff summarizes the why for his senior leadership team (s-team) in a note that dates back to 2004.
Take note of this date (2004). Jeff Bezos’s long-term view, and famous saying “those quarterly results were actually pretty much fully baked about 3 years ago”, would then suggest that this decision making tool led to more success? I’ll this examine this later.
The case for writing
A well-structured document sets context, develops a framework that naturally guides readers to the same recommendation as the author, and invites debate. Can’t slides do that? They sure can, as long a company develops culture that values details, minority dissent, and divergent thought. But slides do not scale.
Slides require a presenter to deliver the message. The nuances of delivery, and the challenges with retention and recall make slides less appealing as a scaling mechanism — very important that tools scale if you are an operator-first company. Writing scales better, even if it demands higher construction cost.
So what’s wrong with writing?
Perils of writing — and the impact on culture
Jeff famously once said “your margin is my opportunity”, and I’d argue that writing is an over extended tool that creates opportunities for others to beat you in “time to market”. Especially if you use the tool for decisions that are reversible, or when incubating new ideas whose vector is generally pointed in a good direction and talented people must shape its implementation.
This isn’t advocacy to remove structure from decision making, but instead to create a culture that is dynamic enough to apply the right tools to push innovation faster through company doors. As noted earlier, no decision at Amazon is made without a well-written, grammatically correct and neatly organized doc. That’s tax — which I happily paid and evangelized others to pay during my time there.
And the argument goes something like this: well-written documents lead to even more clarity and conviction, which help ship higher quality products, which create lasting customer value. And the writing flywheel spins faster. But now I argue that customers will forgive our spelling mistakes and typos. Khawaja Shams recently recommended I read Reid Hoffman’s If There Aren’t Any Typos In This Essay, We Launched Too Late!, and after my flippant reaction abated I realized that right tools must be picked for the job, relative to the stage of the project and the maturity of the team.
I will still advocate writing yearly plans, but can’t imagine using that same precision-demanding tool to incubate new ideas. I believe a healthy culture should actively harness the collective power of a diverse and talented team that is handed a problem, (perhaps a rough sketch to a solution), and asked to debate the most meaningful way to create customer delight.
Diversity (in thought, approach, and people)
A strong culture must value inputs. Teams represent the most critical input to the long-term success of a product.
When you write, there is a tendency to gravitate to an author vs. reviewer(s) mindset. Those familiar with the silent reading study hall at Amazon can probably attest to evaluation apprehension — feelings of anxiety creeping up as you see readers twitch and fervently scribble with a red pen all over your document. They mean well (these reviewers) but seldom do they emerge in these conversations as true contributors. We need to also measure individuals on a contributor metric, and the culture should incentivize this.
I am not advocating for social cohesion or loafing. I champion debate, minority dissent and divergent thought. Characteristics I feel are critical to the success of highly functioning teams in an organization. Charlan Jeanne Nemeth’s, research on the liberating role of conflict — is a fascinating read about how conflict produces outsized results, leading to a culture where you can achieve “low levels of relationship conflict but increasing levels of process conflict”.
Nemeth also argues that evaluating outcomes on the basis of a single meeting is an inaccuracy. And a writing culture values selling over iterating because its hard enough to make decisions in an organization, so meetings must achieve an outcome. I think teams, especially those incubating and experimenting, should be offered rope and encouraged to get comfortable with high levels of ambiguity for some periods of time. Because innovation doesn’t follow a liner path.
Also tools used for cautious decision making can’t be the right ones for quick iteration. Experimentation must be valued over precision in the early days or when the risk is markedly low. This is why I encourage people that I work with to not take pride in their ideas, but instead in their execution. Doing so fosters a culture where people are incentivized to help others quickly get to the start line.
Story Telling
Writing also somehow gives permission to mix resolution i.e. story telling at the high-level and implementation details at the low-level. Customers making decisions about your product don’t read the full length of a press release, but instead look for short form messages and product placement cues on how your offering creates value for them. And I find cultivating the art of story telling to be an underrated cultural trait.
You must connect to customers in all forms of media, and in their language. But if your form of expression is simplified down to writing then you surely have missed out on maximizing the talent diversity of your team. Writing creates for incredible operational leverage but it doesn’t celebrate story telling in authentic ways. Shouldn’t we let designers express their ideas with visuals or workflows; developers with demos and prototypes? Why should we rob ourselves of the privilege of someone’s unique way of story telling?
Circa 2012, I had the fortune to meet the person who was credited with the idea of Amazon Prime (launched in 2005 — take note of date again). The $139/year subscription service from Amazon that allows customers to get free shipping on millions of times, video streaming and more. I asked the person about what led to the launch of what now is a moat for Amazon. And he proudly described a short note that he sent to Jeff about this loyalty service. I suppose back in the early days Jeff’s inbox was more accessible. He then went on describing meetings he had with Jeff’s at his home, even on weekends, to flush the idea out further.
He made no mention of lengthy executive reviews prepping for a Jeff meeting, no meticulous combing of prose, no silent reading time — just a talented team iterating on an idea that had its’ vector pointed in a good general direction. I don’t know the story about AWS (Amazon Web Services), but do know that Amazon also didn’t originate from its writing culture either.
Did Amazon see measurable impact from its writing culture?
Disclaimer: Correlation isn’t causation, and take high-level metrics with a grain of salt
For simplicity, lets look at Amazon’s gross revenue pre/post the initiation of its writing culture. The following two charts show Amazon’s gross revenue from 1997 thru 2007, and from 2008 thru 2018 respectively. As noted earlier, Amazon’s writing culture went into effect in late 2004, and because Jeff knew that results were pretty much fully baked three years in advance, decisions made in late 2004 would show impact in late 2007. So I examined correlation for two equal periods to see if Amazon’s writing culture had any impact on its top line revenue.
Compounded revenue growth rate from 97' thru 07 ‘ was 58%, and 28% from 08’ thru 18'. Revenue multiplied 98x from 97' thru 07', and 12x from 08' thru 18'. The polynomial regression line in these periods suggests, however, that there wasn’t a real material change in the trajectory of the business. So for the sake of this post, I opine that effects of writing on revenue is inconclusive.
Nuances don’t work with a top-down message.
If the impact on writing is hard to quantify or rather inconclusive, then why not experiment with more loose forms of debate and idea generation. Well, as someone who has the privilege to support a large staff, I know first hand that a nuanced message from a leader deviates quickly as the message trickles down. It is important to create operational leverage by sticking to a simple message, and just repeat, repeat and repeat.
“A friend of mine once paraphrased David Gergen, saying on the subject of repetition, “If you want to get your point across, especially to a broader audience, you need to repeat yourself so often, you get sick of hearing yourself say it. And only then will people begin to internalize what you’re saying.” — Jeff Weiner, CEO of Linkedin
So it seems complicated to create a nuanced framework that demands writing in some instances, and allows for the use of a less precise tool to communicate ideas in other instances. These nuances may not stick!?
But I argue leadership is about judgement, the development of frameworks and transferring of these skills to others so that you can create a culture that is dynamic and adaptable to situations. I look at developers that surround me daily, and I see them apply exceptional judgement when developing locally (breaking things quickly) vs pushing changes to production (making sure nothing breaks). They use different tools for the job at hand. We need a similar ethos for innovating quickly, because I believe that survival depends on creating ways that help our teams move faster.
The irony of me writing a long-form document to convince you that this format isn’t suitable as a canonical mechanism doesn’t escape me— but I thought it was the right tool for the job. Also this post is not intended to dunk on an extremely high performing culture, its just a minority dissenting view that I hope my former (and current) colleagues can appreciate.
