Working Backwards: Dave Limp on Amazon’s Six Page Memo

Amazon’s SVP describes how the Devices group moves forward on a product

At The Verge, my old colleague/boss Nilay Patel has a new Decoder interview with Dave Limp, SVP of the devices group at Amazon. It’s available as a podcast or lovingly transcribed, and is well worth listening to / reading — I’ll be referring to the transcript throughout here.

As others have pointed out, this interview is about as candid as Limp has been about how he and the Devices group work within Amazon. But what caught my attention were Limp’s continual references to a well-known part of Amazon office culture: the six-page memo.

Jeff Bezos instituted the no slide decks / narrative memo requirement in an email to Amazon’s S Team back in 2004, and described it in an interview with Charlie Rose about a decade ago:

BEZOS: We have study hall at the beginning of our meetings… The traditional kind of corporate meeting starts with a presentation. Somebody gets up in front of the room and presents with a powerpoint presentation, some type of slide show.  In our view you get very little information, you get bullet points.  This is easy for the presenter, but difficult for the audience.  And so instead, all of our meetings are structured around a 6 page narrative memo…

When you have to write your ideas out in complete sentences, complete paragraphs it forces a deeper clarity… Time doesn’t come from nowhere. This way you know everyone has the time. The author gets the nice warm feeling of seeing their hard work being read… If you have a traditional PowerPoint presentation, executives interrupt.  If you read the whole 6 page memo, on page two you have a question but on on page four that question is answered… And so that is what we do, we just sit and read.

What began as a requirement for top executives became an Amazon institution. Instead of slide decks, narrative memos of slighly different types and genres became a core part of Amazon’s office culture. Junior executives write six-page memos; Bezos’s annual letter to shareholders is typically a six-page memo. It’s also a genre that’s taught (sometimes not entirely correctly) in writing programs, and one that’s been dissected online many, many, many times.

The two-page (or one-page) mock press release format is also a writing genre common to Amazon, although it’s less often discussed than the six-page memo. What’s interesting to me about Limp’s invocation of the six-page memo in this Decoder interview is that 1) he describes how the two writing forms are, at least for his group, a combined genre and especially 2) how the memo has actually become a roadmap for product development. In short, the Devices group not infrequently begins with the press release, expands into the six-page narrative memo, and uses that document as a tool to decide which new products to develop and release.

LIMP: The invention process is very different. It has a different decision-making framework, we use a thing we call the “working backwards process,” where we don’t use PowerPoint, or Keynote, that much inside of our business. We write narratives. And the narratives are six pages long, and a new product, any new product inside of Amazon, the first page of that product, that narrative is a press release, as if you were launching the product tomorrow. And then the next five pages are frequently asked questions; how is this going to be differentiated? How would it be priced? What invention do you have to solve to be able to do this, etc., etc. And there’s a bunch of those questions that go in.

And those docs come, and we review them, but that process is much messier. It is very rare that I would see a working backwards document like that for any new product — could be the original Echo or the original Kindle — that we would approve the first time we saw it. Normally it takes many iterations of that. And what’s critical about that is that because we really want that to be as good as we possibly can make it, because as soon as we agree on that document, the decision is made. That project is green-lighted. The next step is to find a single threaded leader to run that project. Somebody that wakes up full-time, every day, their job is to make that product happen, because you never want to have somebody inventing part-time, that’s a very important thing. And so, that’s why that process tends to be more iterative.

Later, Limp describes how this type of “working backwards” memo has (to some degree) penetrated Amazon acquisitions like Ring, Blink, or Eero, including the decision to combine elements of the Ring and Eero together in the new Ring Alarm Pro:

The way you phrased it made it sound like there’s some dictatorship. What happened was exactly what I described in the invention process, which is, we wrote down a working backwards document basically about what the next version of the Ring Base Station would look like. And one of the things that came out of that was this idea, I remember having the conversation, this idea that you might want your internet to have backup. I live in Washington, we have wind, I have a generator because my power goes down pretty often. And so, I have power backup, but the internet is equally as important to me as my power, as we’ve gone through the past decade. I don’t know if it’s a human right, but it sure feels really important.

The idea of being able to back up the internet was, in the case of that product, to have LTE built into it so it can flip over from wired connection to cellular. And once you have to flip over, then you kind of go, “Well, how do you do that seamlessly?” Well, it made all the sense in the world that it would be on the back of a router, and we have a company, Eero, building world-class mesh routers. And so, it really started with the customer problem first, and led to a “better together” scenario. And I think that’s the better way to do it than just say, “Eero must work with Ring. Ring must work with Blink.” In my experience, those work in the short term but in the long term the organization will rebel against it because it is often not right for customers.

Later, Limp uses the six-page memo as both a metaphor and an example of how different technological paradigms don’t dispace so much as augment each other:

If I have to write one of those aforementioned six-page documents, that QWERTY keyboard, as old and as obsolete as it should be, is still the best way to write a six-page document. Now, did smartphones move us to another level? Yes. In the same way, ambient computing, Alexa and Echo and this ambient intelligence, is not going to supplant the computer nor the smartphone. It’s not going to. It’s doing something different, but I would argue that it’s already here and here to stay. It’s at scale, people are using it, they love it, they are adopting it.

There is so much more that has been said and could be said about Amazon’s internal (and occasionally external) use of the six page memo, the discipline of silent reading during meetings, and the important elements of timed discussion and iteration as a process after the memos are written. I may return to this topic again as soon as later this week.

But for me, what’s revealing in this interview is the use of the memo as a structuring document so early in the product development process, as early as invention and identifying a person responsible for product development. This is not just “this is why we should name our tablet the Kindle Fire and how we’re going to situate it in the market” — this is “here is why Amazon should be interested in having a tablet at all, how it could be used, and what services we need to package and/or develop to make it viable.” That is a big load for a six-page memo to carry, especially for a product (or even a product team) that doesn’t yet exist.