Fish Food: Episode 639 - How to innovate like Amazon
Amazon's innovation system, the evolution of the marketing operating model, Deep Research, A2A, and snail mail newsletters
This week’s provocation: Amazon’s innovation formula
I spent a few days with Diageo’s European leadership team in Dublin recently and whilst there I saw a talk from Amazon’s supply chain lead Marcus Mallon who talked about how the company innovate. Amazon are of course known for their customer obsession and ‘working backwards’, their relentless experimentation and openness to fail, and for combining a long-term vision with a willingness to be misunderstood. Bezos once famously said:
‘I don't think that you can invent on behalf of customers unless you're willing to think long-term, because a lot of invention doesn't work. If you're going to invent, it means you're going to experiment, and if you're going to experiment, you're going to fail, and if you're going to fail, you have to think long term.’
But what stuck with me most about the talk was how they had systemised innovation through a reinforcing combination of four fundamentals which is sometimes represented as a formula: architecture and organisation amplified to the power of mechanisms and culture.
Each element plays a key role in enabling systemic innovation, so I’ve written up my notes and augmented them with a few other thoughts on how they bring this to life.
Architecture: Scalable, modular, and service-oriented
In many companies innovation dies in complexity. Yet Amazon’s architecture is intentionally designed to remove blockers and let teams move fast, scale safely, and iterate independently. There are several elements to this:
Service-Oriented Architecture (SOA): Early in its journey, Amazon broke down functional outputs into modular, API-accessible services. This enabled teams to build without stepping on each other’s toes.
Internal platforms: Teams leverage powerful shared tools (like AWS, ML toolkits), reducing friction and duplication.
Loose coupling, tight alignment: Teams can innovate freely within a bounded, interoperable system.
In fact the very architecture that enabled teams to innovate internally later became a product: AWS (that’s innovation that pays twice I guess).
Organisation: Structuring for autonomy, focus, and accountability
Accountability and decision rights are often blurry or distributed in large organisations. At Amazon, organisational structure and decision-making follows intent. It’s designed for speed, clarity, and empowered ownership.
Two-pizza teams: Small, autonomous, and multidisciplinary teams that can build end-to-end without top-down interference. Each team has a ‘fitness function', a defined business metric that the leadership agree with the team lead.
Single-threaded leaders: Each major initiative has one person fully focused on it, meaning that there are no contradictory agendas.
Clear ownership: Every product, system, and metric has a directly responsible individual. No ambiguity means no hiding.
Mechanisms: Turning bold ideas into repeatable practice
Great innovation cultures can’t scale without the right rituals and systems. The idea behind Amazon’s mechanisms is that they turn principles into practice and enable thousands of teams to experiment, launch, and iterate every day.
PR/FAQ Process: Every idea starts with a future press release and internal FAQ. This clarifies intent, forces customer-centricity, and prevents building for the wrong problem.
Bar raisers in hiring: They aim to ensure that every new hire elevates the standard, bringing intellectual rigour to idea generation and execution
Leadership principles: These are public, but also baked into decisions, reviews, and promotions.
Culture: Customer obsession married to long-term thinking
Amazon’s intent has long been to foster a culture which institutionalises customer obsession.
The working backwards methodology means that every product starts with a clearly defined customer need.
Failure isn’t feared, it’s budgeted for. Marcus used the example of how the Fire Phone flopped dramatically, but then paved the way for Alexa.
Amazon’s ‘Day 1’ philosophy fights entropy. The moment a company stops acting like a hungry startup, it starts dying.
These four pillars are not siloed but instead reinforce each other to create a self-sustaining innovation system and a self-reinforcing flywheel (I’m a fan of flywheels). Culture sets the mindset (customer obsession, long-term thinking). Mechanisms turn that mindset into consistent behaviours and action. Architecture provides the infrastructure to support rapid, safe experimentation and scalable execution. Organisation ensures the right people are empowered to move fast with focus.
It’s a fascinating example of how to build an innovation system within a large organisation so that innovation is expected not exceptional, distributed not centralised, and treated like infrastructure rather than theatre.
If you do one thing this week…
I’m delighted to say that the next edition of Google Firestarters has just been published, featuring the wonderful Robin Charney, Partner at AAR, talking about the evolution of the marketing operating model. Robin is so insightful on why the ‘how’ of marketing (the operating model) is now just as important as the ‘what’ of marketing (strategy), and draws on a wealth of experience as well as research that the AAR have done into the ‘4Ps’ of the operating model: people, partners, process and platforms.
Each of these 4Ps is undergoing significant change, just as CMOs need to ensure that they are more intertwined and seamless than ever, and that marketing can effectively demonstrate commercial impact and drive real growth. Too many highlights to mention but I think her insights into the people side of change (skills, org design) and partnerships (including inhousing and external agency resources) are particularly interesting.
Links of the week
AI has totally transformed how I do research for my client work. Now Deep research is available on Google’s Gemini 2.5 Pro. I’ve not had chance to play around with it yet but have heard that it’s excellent. If you’re a Gemini Advanced subscriber select ‘Gemini 2.5 Pro (experimental)’ and click ‘Deep Research’ in the prompt bar. There was a useful (particularly for strategists) prompting template for Deep Research here. Meanwhile I’m continuing to find NotebookLM really useful for research - if you’d like to explore it more, there’s a basic guide here, and this 30 minute tutorial on how it will change how you learn is helpful, and there’s also this interesting Google Deepmind podcast about it with Hannah Fry and the author Steven Johnson
In the last couple of editions I’ve mentioned the transformational potential of MCP’s (the open standard for connecting AI agents to external data sources and tools). Now Google have just launched a new Agent2Agent Protocol (A2A) which will ‘allow AI agents to communicate with each other, securely exchange information, and coordinate actions on top of various enterprise platforms or applications’. These standards are going to accelerate the adoption of AI agents and the ability of enterprises to create agent ecosystems that coordinate and execute work efficiently. Again, potentially transformational. Nate Jones had a good explainer here.
OpenAI are rumoured to be interested in building a social network (really?)
Two interesting takes on how people are really using GenAI. First, research from Anthropic on how university students are using Claude. Secondly this HBR article which seemingly shows some surprising differences in thematic usage of GenAI from last year, with personal support use cases (companionship, therapy, purpose) coming high up on the list
‘When answers get cheap, good questions are the new scarcity’. I liked the premise of this piece by Sangeet Paul Chowdray about standing out in a world where everyone has good answers. It ties in nicely with things that I’ve been writing about recently, and coincidentally I’ve been listening to a podcast about the life of Picasso and had the same thought about how photography changed the course of art, and the parallels with how AI can change how we think
With everything going on right now I think I need more Jing: ‘Jing is the stillness needed to listen to yourself and the universe. It’s the pause before replying to a question. It’s watching steam plume from your coffee. It’s the moment of silence on a forest walk when even your breathing seems loud. Jing is the power of a mind unrushed and undistracted - a mind that doesn’t push but flows’.
And finally…
There’s still something lovely about getting a physical letter or postcard through your door, and so I liked the idea of Stamp Fans, a ‘snail mail publishing platforms for creators’. Russell Davies is testing one out: ‘8 pages a month of interesting things I've found and reflections on those things’.
Weeknotes
This week I was out in Dubai doing some work with PwC, and I also ran some sessions for media leaders on navigating and leading change. Dubai is like nowhere else and, alongside the work, I was lucky enough to take a few days off whilst out there. Next week I’ll be in the UK, writing up my next trends report for Econsultancy, and running an AI in Marketing session for a team in London.
Thanks for subscribing to and reading Only Dead Fish. It means a lot. This newsletter is 100% free to read so if you liked this episode please do like, share and pass it on.
If you’d like more from me my blog is over here and my personal site is here, and do get in touch if you’d like me to give a talk to your team or talk about working together.
My favourite quote captures what I try to do every day, and it’s from renowned Creative Director Paul Arden: ‘Do not covet your ideas. Give away all you know, and more will come back to you’.
And remember - only dead fish go with the flow.