What I learnt at Amazon that I wish I knew in government

This month I left Amazon Web Services (AWS). I joined six years ago as a civil servant from the European Commission, not because I no longer believed in the institution, but because I believe that good policymaking should seek diverse skills, perspectives and expertise. I was planning to learn a trade that could make policymaking more robust, resilient, and long-lasting.

So what did I learn about operational excellence from Amazon?

1. Governments should work backwards from citizens. 

Amazon’s business model and innovation methodology rest on one fundamental principle: customer obsession. The ideation of any product, service or feature starts with the same simple questions: “Who is the customer? What is the customer’s essential problem or opportunity? How do you know that? How does your solution address the customer’s need? What does the customer experience look like?” In other words: who are we building a service for and why?

If the working backwards process starts with users’ needs, the policymaking process often starts with ideas or plans. Governments tend to focus on what to do, based on data from surveys, demographic studies or economic indexes. But mobilising the right knowledge to make good decisions is trickier. The delivery of policies or products is made messier by behavioural biases and complex dynamics that are invisible at the design stage. For instance, citizens might be reluctant to insulate their lofts despite the government showing clear economic savings they would accrue. The reason? People needed a service to clear the loft. When the government subsidised it, loft insulation increased fivefold. Starting with who, the citizen, and ending with how that citizen will experience a policy or a service, creates a common denominator from policy design through delivery.

Several governments around the world have created innovation units based on design thinking and behavioural insight. However, these are exceptions, discrete experiments confined in small departments that risk being dismantled when political mandates change. Asking the questions that put citizens at the centre of policymaking should be governments’ basic instinct. Each policy brief should have in mind the name of someone whose essential needs are being addressed.

2. Governments should give ownership of policy projects to those with the right skills, not the right title. 

The most difficult part of my transition from the European Commission to Amazon was cultural. At AWS, my first ‘doc read’ – the live reading of a 6-page narrative that articulates your business proposal to stakeholders – lasted less than five minutes. “Do you know what you are doing? We didn’t hire you to ask for permission, you need to be the CEO of your business” – my then boss told me. That way, he ignited a permanent change in my mental model.

In the civil service, there is a tendency to ask for permission to a hierarchy that is not always the best judge of a specific domain. Proposals can lose their value as they move up through the bureaucratic chain, since those closer to the subject are excluded from the final decision. At Amazon, even as a junior employee, I had the right and the responsibility to show backbone and disagree with senior leadership on my projects.

Governments could benefit from introducing the concept of a 'two-pizza team', namely, a team large enough to be fed by two American-sized pizzas and which contains all the skills necessary to deliver on a project. As an Italian, I never understood the concept of an American pizza, but as an Amazonian I experienced the benefits of these self-contained, multi-disciplinary teams. Their consistent, single-threaded ownership - where one individual is ultimately responsible for a given project - creates accountability and agency. These traits are especially important in government, where personnel is permanent and you have few tools to monitor and motivate people. Single-threaded ownership also maintains connection between the product and the user, and increases the chances of using the right team for the right job.

The additional benefit of this working method is speed. One of the most painful experiences of my career is called ‘Inter-Service Consultation’. In such a consultation, representatives from all the Directorate-Generals of the European Commission are called to provide feedback on a draft legislation. That means that twenty-seven people are locked in a room for at least an hour, obliged to comment on policies that they may have nothing to do with. This process gets repeated at various levels of hierarchy, wasting countless hours, creating coordination costs, and diluting the content of the policy just to appease colleagues. The role of hierarchy should be to make policies effective, robust, and durable, not to rubber stamp at every level of the organigram. The boards that review policies should be agreed in advance based on minimum necessary consensus to create the best possible policy.

3. Governments should be free to acquire the right talent and tools for the right job.

At Amazon, a candidate gets hired if that person raises the bar: the interview committee needs to determine not only whether the interviewee meets the requirements for the job, but whether he or she will raise the level of the team overall. Not only does this mechanism assess competence and cultural fit, but it ensures that a new hire helps the overall business improve its capabilities. Hiring in the governments I know is more often based on testing specific knowledge that is not functional. A friend who recently took a national civil service exam told me that one of the questions was: When was Queen Elizabeth crowned? Disclaimer: this test did not take place in the UK.

Hiring systems such as the one above are unhelpful in at least two ways. One, while general knowledge of the British monarchy is admirable, I would argue that experts in Artificial Intelligence would be more useful to address the challenges of government. The example I report is extreme, but the point is that governments still recruit following a notional system based on an obsolete vision of bureaucracy. This is why lawyers and economists abound, but there is a shortage of engineers, computer scientists or designers.

Two: hiring in bulk through national competitions does not give governments the agility to adapt to changing needs. The case in point is the regulation of AI. The lack of AI experts among civil servants makes the government reactive, vulnerable to asymmetry of information and at risk of regulatory capture. Can we afford to wait another too-many-years to hire AI experts that can help governments level the playing field with tech companies? It is true that most governments are highly unlikely to compete with private sector salaries. What is also true is that a growing number of people would prefer a more meaningful job, granted a certain salary threshold. Governments’ comparative advantage is purpose, but they need to offer more innovative and flexible work environments to leverage it. 

Governments suffer from the same penalty with the procurement of technology: they lock themselves into contracts for five or six years that buy them expensive technology systems that decay fast, only because they are used to account for capital expenses. These rules are then systematically broken during emergencies such as COVID-19, where vaccines needed to be procured at record speed. But the changes rarely outlast the crisis. 

Finally, hiring is just the start, governments should develop the best. I remember leaving the civil service in part because I felt like my competencies were not growing, but being consumed. Of course, I was learning to get stuff done and I was privileged to work in a team that was tasked with producing foresight, but I was working in a system that did not incentivise learning, update knowledge, or encourage field trips or internships to understand how a given technology works. At Amazon I visited fulfilment centres, did customer service for a day, and was required to learn about AI because it matters for the future. Developing the best in government should not rest on individuals’ curiosity, it should be a duty. 

I liked to joke to my colleagues and customers at Amazon that I am a recovering civil servant. The truth is: I have never recovered. And that is by choice. In a world where headlines focus on everything that will change driven by technology, we tend to overlook the important things that will not. Governments are and will remain a necessary actor to make the world “better led, served, and governed” – as my School’s motto recites.

Realising that vision requires a government better managed. I have experienced first-hand that large, sometimes bureaucratic, organisations can be a model of operational excellence. It is unfair to civil servants and to the citizens they serve to assume that governments are destined to be slow and ineffective. 

It is in everyone’s interest to make governments a place of excellence. I will now take these lessons and offer them as my modest contribution to build a technology-ready public service: capable of building solid policy pipes beyond perfect architectural designs.

Leonardo Quattrucci has over a decade of experience in building, leading, and scaling big-picture programs across government and tech, globally. Today, he advises start-ups, business, and public sector institutions on building strategies and operations that leverage emerging technologies to create public value. His work has been recognised by the BMW Foundation, Forbes Magazine, and the World Economic Forum. Find him on LinkedIn or X (formerly Twitter).

Photo by Adrian Sulyok on Unsplash.