Month 6 at Amazon

Last month, I was offered the opportunity to take part in Amazon’s yearly planning process which meant writing an OP1 (Operating Plan 1) document covering our 2025 objectives. Although I have been involved in such exercises at previous companies, the bar at Amazon is high and the exercise spanned over many weeks involving numerous teams and reviews.

In this article, I want to relate my experience out of this exercise and how I am planning to attack future planning activities. In one of his past posts, Dave Anderson (ex GM and tech director at Amazon) offers a more in-depth day-by-day (and often amusing) look into the process, here I will rather summarize some of the lessons from this past month.

Gathering ideas (bottom-up)

Amazon is mostly run as a series of independent businesses with a high degree of freedom to trace out their own path and destiny. On purpose, most teams at Amazon run the OP1 process bottom-up where individual teams get consulted and can provide their ideas.

Such a process can go very wrong if people are not coached to put some guardrails in place:

• Is the idea big enough financially? As part of the Think Big principle, teams are expected to come up with ideas that make a true impact downstream
• Is it profitable? A certain investment might help either in cutting costs or provide new sources of income. For new income streams, higher profits will provide a better margin of safety
• What is the execution risk? Does the team have the skills and can it deal with the complexities involved in creating the new solution? If the unknowns are too big, it might be better to do some initial exploration rather than the full idea itself.

In my case, I used my notes of previous team discussions during my time at Amazon. Although I did not discuss with all teams in equal amounts, but probably avoided a lot of wasted effort described in the linked article mentioned above.

Bringing in top-down structure

The previous step yielded a large set of ideas (around 40) that I was able to slot in 7 major themes. When people started to review my initial drafts, most felt it was a too wide collection of ideas that lacked a top-down narrative.

Indeed, as described by authors like Dan Pink, if you want to be most persuasive, you should typically not have more than three main ideas to structure your arguments.

As a result, I rebuilt the document starting from three pillars suggested by my manager and the feedback on the result was more positive. This process also helped to start shaping the resource asks for each of the pillars and underlying ideas.

The need of functional alignment across businesses

Building independent businesses allows to make the right (local) decisions but it does not always align technical choices or bring consistent experience for customers.

To give one example, Spotify has been diversifying into new verticals such as podcasting. As explained on a recent podcast, it felt that a consistent end user experience should come first and therefore chose to slow itself down by aligning three horizontal functions:? backend, front end and recommendations. The purpose of such an approach for recommendations is to allow a customer’s taste in music to translate to better podcast recommendations. At the same time, this (horizontal) alignment requires more time and a slower pace.

Even at the smaller scale of our team, whilst we are in a single business, we still have different developer persona that we cater for. Often, functionality gets used differently by these different persona (e.g. internal developers vs. external ones).

When planning is not the answer

Sometimes, completely new opportunities pop up. Whereas planning is well suited for already established businesses, often as an individual, or as a company, you have to take the plunge.

Within Amazon, most of its major products and initiatives since 2004 have one very Amazonian thing in common—they were created through a process called Working Backwards. Its principal tool is a form of written narrative called the PR/FAQ, short for press release/frequently asked questions. Such a process allows to overcome the more incremental approach so typical of the OP1/OP2 approach.