data leader
unless you are technical
you cannot be a
This week, we get Up-Close with Piers Stobbs, VP Science at Deliveroo and former Chief Data Officer at Cazoo. In this interview, Piers discusses how the economic volatility has reinforced pragmatism in his approach to data and how to truly become a data driven company.
What is your career highlight to date and how has that influenced you as a leader?
I quite like these questions because they force me to look back on all the different things I have done in my life. The most life-changing experience I had was quite early on in my career, back in my early twenties. At that time, I worked as a wireline oil engineer: basically lowering measurement tools down oil wells to understand their performance. It seems crazy to me now when I think about it - a 23 or 24 year old running international teams out in the remote Saudi Arabian desert. However, that was a long time ago, and my abilities with shaped explosive charges have definitely waned over the years.
Joking aside, I was very fortunate to have a fascinating three years in my previous role with Cazoo. It was an unbelievable opportunity, building a modern data team and data stack from scratch, and being told to aim for hyper-scale. It enabled us to be bold and creative and try new cutting edge approaches. I thoroughly enjoyed the process and learned a tremendous amount during those three years, with the company growing to over 4000 people, a billion in revenue and going public on the NYSE.
And I’m very excited about my new role at Deliveroo- I’ve only been here six months or so, but we have an amazing team, an excellent tech-stack and some really interesting problems to get stuck into, particularly in the optimisation space.
“It seems crazy to me now when I think about it - a 23 or 24 year old running international teams out in the remote Saudi Arabian desert”
In your opinion, what is what are the top 3 attributes of a great data leader?
I am an ardent believer that you cannot be a data leader unless you are technical, and I know that this perspective may be somewhat controversial. However, I genuinely believe that there is so much nuance in how you derive value and deliver results through data, that understanding what is possible and what is not, technically and practically, plays a crucial role. This is why I believe being strong on the technical side is essential for a data leader.
Moreover, I think that curiosity and inquisitiveness are virtues I look for in all my teams. I love seeing that sort of engagement with a tricky problem because I believe it is essential.
As a data leader, communication skills are also key - the ability to translate complex and nuanced analytical results into simple and easy-to-understand messages for a wide-range of audiences. It is also critical to know how to translate business problems into data problems, and vice versa.
These are probably the three things that I highly value, and they are certainly qualities that I believe make a significant difference in the field of data leadership.
Do you think you need different data insight skill sets for different stages of company growth from start-up to scale-up to listed?
I believe there is a broad array of skills that you need to have at any time. Some of those things we talked about before, but the relative importance definitely shifts with company maturity and size. The reason earlier stage nimble companies can be exciting is that it is about being a bit more of a generalist, cutting through the complexities and being pragmatic to get things done.
As the scale of a company grows, and as the age of a company grows, there is much more of a need, like in any role, to take people on a journey, to influence others, and to build a coalition to make larger scale changes or improvements. So, fundamentally, it is less about small-scale influence and much more about larger scale influence across a wide organization, which is definitely a different skill set.
“The reason earlier stage nimble companies can be exciting is that it is about being a bit more of a generalist, cutting through the complexities and being pragmatic to get things done.”
If you think of creating the “North Star” data insight team, how do you create that and what's the optimal outcome?
I think there is a certain amount of upfront effort required to get buy-in from the business leaders and the senior leadership team on the importance of being data-driven. Everyone talks about wanting to be data-driven and says that they value analytics, but I believe it means different things to different people. Therefore, trying to obtain that buy-in early on involves understanding what data and insight means to them. Ultimately, being data-driven comes down to making better decisions, but for one group that may translate into looking at dashboards every day and trying to gain insights; whereas for another group that may mean they are interested in receiving reports to aid in decision-making – it can vary.
As a result, there is a lot of formulating upfront, and then securing that buy-in on what the data team is going to focus on. Personally, what I love about building good teams is that feeling of trust and almost collegiality, creating a sense of family where we're all in this together. Since the skill set in this field is so wide, naturally we all have a pretty diverse team, and that's something I strive for. Everyone contributes in different ways and the sum total is hopefully bigger than the sum of its parts. Developing a culture where everyone's opinion is valid, and where we are all focused on quality and delivery, is an interesting balance that I always strive for.
How has the current economic and market contacts impacted your strategy?
I think it's an interesting question and I have definitely experienced the impact of changing economic conditions first hand. When Cazoo started in 2019, it was all about huge valuations, raising lots of money, and growing as quickly as we possibly could. Then Covid hit and that changed the playing field - that was a significant learning experience for all of us. In addition, I used to work in the office almost exclusively, but I have since discovered that you can actually function and be very productive remotely. So, whilst the optimal hybrid balance is still a work in progress, the key takeaway is that you don't necessarily have to be in the same room to get things done. This learning has greatly benefited the teams I've worked with.
Moreover, the economic crisis and changes in the economy have reinforced the importance of pragmatism, from my perspective. It is crucial to focus on what is most important and how to make a difference as quickly as possible. This requires ruthless prioritization, which I find to be one of the hardest challenges for a leader. There are so many things that could be worked on in the data space, and it is essential to be critical in deciding what to prioritize and when to move on to the next thing.
Balancing the need for quick solutions with the understanding that some issues may require more extended efforts is vital. Sometimes the 80% solution might be sufficient for the short term, and moving on to the next project may yield better results. These attributes are what I believe strong leaders in this time should possess.
“there's a lot of excitement around the progress that has happened in the AI space recently, particularly in Generative AI, where you're synthesizing new outputs from existing materials – this is truly groundbreaking”
And what would your advice be for data leaders leading through economic challenges for the first time?
It can be tough because during these times you’re obviously going to be much more resource-constrained, and you’re going to be under much more scrutiny as costs become, not surprisingly, the critical factor. I think engagement with stakeholders is really important. It always is important, but particularly in these sorts of times making sure it's clear what everyone is working on, how you've come to those prioritizations, and what you hope to deliver is crucial. Continuously updating these elements is essential, even if it's not something that comes naturally to me or some others in the space. Finding ways to work around and through issues and seeking guidance from others is important. The goal should be to avoid surprises as much as possible; the more you can proactively anticipate, the better.
I also think it's important to take pride in different things. In good times, people may get excited about research projects or trying something completely new, yet in these times, it can be about celebrating different victories. For example, rolling out an automation that saves people a bunch of time and resources is a great win. Recognizing and celebrating these successes, even if they are of a different nature, is crucial.
“The goal should be to avoid surprises as much as possible; the more you can proactively anticipate, the better.”
What trends have you started to see develop in the last 12 or so months that are going to have a big impact going forward?
I always find this type of question tricky. I have spent lots of time with forecast models and I know well how hard it is to predict what's coming next week, let alone next year.
That said, there's a lot of excitement around the progress that has happened in the AI space recently, particularly in Generative AI, where you're synthesizing new outputs from existing materials – this is truly groundbreaking. There's lots of discussion and thought about how this is going to change what we do in the data space and how we do it. I do think it's the classic case of perhaps overestimating the short-term impact while underestimating the long-term impact. It will have an impact in certain places, and I think the data function, for instance, will likely evolve pretty dramatically over time.
For example, there's already a lot of discussion about how some AI driven coding aids (such as copilot) are benefiting engineers and data scientists. I believe the way we do our work will definitely change. However, from a business perspective, it may still be a bit early to define the discrete applications of AI and where they will have the biggest impact. The way we interface with data is likely to change with more text style interactions with data sets, which will certainly expand the reach of data and analytics to more people. How we make sure the new interfaces produce accurate and well explained outputs is likely to be the biggest challenge.
The concept of MLOps – making the process of getting machine learning models live as efficiently as possible - has been around for a while but it still far from a solved problem. It is already driving quite a lot of incremental value just below the surface and will continue to do so.
Do you have a notion of what the future of data will look like in 2025 and beyond?
Again, I think it is very hard to say, but as I alluded to, I believe there will be some fundamental changes. One thing I do know from my career is that it's a forever evolving technical space, and you must have curiosity and a desire to learn new things because that's always happening. I increasingly wonder and actually hope that some of the more automatable tasks do get automated effectively. For example, smart solutions around ingesting data, building pipelines and producing warehouse tables could be more automated over time. This might enable smaller teams to handle these tasks or make the process more efficient.
I believe there is still a lot of scope for how best to solve problems and utilize data to make decisions. Whether it involves assiting humans in decision making or computers making decisions in a machine learning automated system, there are various approaches to consider. The move to the cloud and improvements in computing make these services more accessible. However, it's also crucial to consider how you apply off-the-shelf solutions. While it may seem simple, tailoring standardized packages or approaches to your specific problem and data is essential. This can still provide a competitive advantage, and I believe few companies do it well.
How do you stay up to date in your function and as a leader?
I am fortunate enough to be on the Data Science and AI Section at the Royal Statistical Society. As part of that role, I put together a monthly newsletter, which I really enjoy. It forces me to read a bunch of things over the course of the month and then synthesize them into some vaguely coherent thoughts. Staying up to speed on generative AI, research, pragmatic approaches for implementation, and other relevant topics is essential, and I find great satisfaction in doing so. Additionally, I subscribe to various newsletters myself. I tend to save articles that I find interesting and then dedicate a day to putting everything together. I’m obviously biased but I would definitely recommend checking out the RSS Data Science Newsletter, specifically the data science section.
Apart from that, I remember considering myself to be pretty objective and logical until I read "Thinking, Fast and Slow" by Daniel Kahneman. It opened my eyes to all the biases I have and made me realize the significance of understanding human behaviour, including our own biases. I believe these behavioural aspects are incredibly useful in our work - we can become overly focussed on the objectivity of what we do, and I think there's a real nuance around how you get things implemented, as well as the impact on human behaviour that can sometimes be underappreciated.
Personally, I’m an introvert and I have at times struggled with the idea of being a leader as it’s not something that I have felt comes all that naturally to me. However, after reading a book called “Quiet”, by Susan Cain, that highlights the power of introverts, I realised how they can be successful leaders by leveraging their unique strengths. It was an eye-opener for me and allowed me to be true to myself as a leader, and I realised that certain personality traits don’t stop you from being a successful leader. This has really encouraged me to create a workplace where people can be themselves; building such an environment fosters closer ties and creates a better work atmosphere. As you can tell, I find behavioural economics to be a fascinating space, and I keep a vague eye on it to see how it impacts our work.
From a technical perspective, there are many newsletters available, and I highly recommend finding ones that interest you to stay informed and avoid falling behind. It requires dedication to keep an eye on developments in the field.
