It has been a funny old week. As we face what seems to be an unending lockdown with an uncertain path forward and Liverpool sink even lower in the Premier League, I have been scouring the digital airwaves for news that can brighten our outlook. Two different recent stories struck me as offering insight into the promise and perils of our current predicament.
The announcement this week of latest Amazon results has reminded us that Amazon has “not let a good crisis go to waste”. Net sales increased 44% to $125.6 billion in the fourth quarter of 2020. Furthermore, its highly profitable Amazon Web Services (AWS) segment grew significantly. With the Covid pandemic forcing people to stay inside, Amazon’s e-commerce sales have exploded way ahead of expectations. It is estimated that Amazon has added more than 427,000 employees in the past 10 months to grow its workforce to over 1.2 million.
The second goes back to the earliest days when Amazon was forming its digital platform strategy. According to oft-repeated legend, an essential source of Amazon’s success is a key decision made by Jeff Bezos in 2002 that established the foundation for Amazon’s decentralized web-based architecture. The mandate declared that all services would be made accessible via Application Programming Interfaces (APIs) to ensure transparency, openness, and component reuse.
Why does this matter? The main advantage this has offered Amazon is a seemingly infinite scalability based on an ever-expanding ecosystem. It you are a small team of programmers wanting to build the next “killer app”, it is much easer to build on top of existing services than start from scratch. And Amazon has created a large pool of services in AWS that programmer’s access through well-defined public APIs. Every time they are used, Amazon’s ecosystem grows, and they earn both direct (via pay-per-use) and indirect (by expanding Amazon’s service footprint) revenues.
To a large degree, this mythical Bezos mandate from 2002 can be said to have heralded a new era in software engineering that has led to the serverless architectures and API economy that dominate software architecture and delivery today.
All this leads directly to the second key insight of the week, and a key question for all organizations driving hard towards digital transformation. Why are so many software-driven technology projects still being delivered late, spend way beyond their budget, and operate ineffectively when they arrive?
As someone with a history and a small role in improving software design and delivery practices, when I see the success of large web-based architectures such as those delivered by AWS it is hard not to feel a little pride. At last we have made significant progress in delivering largescale software-intensive systems, we may think. But remember that pride often comes before a fall.
So with dismay and embarrassment I read the announcement about yet another failed rollout of a critical software-based system. With an estimated $44M spent on the US Vaccine Administration Management System (VAMS) you would hope that an appointments and bookings application could be effectively created and delivered. But it seems not. Can we look at the story of VAMS as just another failed government project? Or is this a sign of the immaturity of software-driven digital transformation?
Reading through some of the details, it is clear that there are many different things going on here. Here are a few of the challenges that are highlighted in the reports:
Contracting. Letting a non-compete contract to Deloitte “despite having no experience in the field”. There is still much to learn in who gets awarded government contracts and why, despite efforts to improve transparency and fairness such as the Uk’s G-Cloud.
Requirements management. Pressure on teams to deliver quickly meant that they failed to establish a stable set of user needs, and lacked insight into their priority and importance. Without a way to manage the scope of the project, it is impossible to deliver solutions that solve a well defined problem.
User experience. Ineffective usability of the software led to numerous complaints about the right way to interact with the features offered. Working with a wide variety of users, many with limited digital technology experience, the usability is critical.
Adaptability. In the fast-changing context of the pandemic, the software failed to adapt to the needs and experiences of users. No account was taken for evolving the system in response to lessons learned in use.
Stability and Resilience. Lack of consistency in the software reduced effectiveness and undermined confidence. Public-facing systems require resilience across the variety of usage models, and without this they quickly fall in to disrepute.
Integration. By not working across the most common web browsers and technology components, users were forced into multiple software installs that created unfamiliar operating environments. Simplicity, familiarity, and convenience are critical for new software to be adopted.
It is no surprise, then, that this catalog of problems has led to a poor result for VAMS. Improvements are being made, but it will be an uphill battle to build a successful solution from this basis.
Perhaps like many things we have seen over the past few months, these two scenarios demonstrate that progress in difficult times is inevitably a case of “3 steps forward, 2 steps back”. For now, I’ll do well to remind myself to keep my focus on the fundamentals of digital transformation.
Digital Economy Tidbits
How Microsoft Became an Intelligence Driven Organization. Link.
It is helpful to see some of the context for Microsoft’s strategy toward an Intelligence Driven Organization (IDO). They describe 4 key axes in this model based around digital feedback loops, and see digital transformation to be framed by organizing your approach around these axes:
