If only someone had told me
Following the protracted excitement of the recent US elections, I am reminded of an Oval Office tradition (started in 1989 by Ronald Reagan), in which the outgoing President leaves a letter for the incoming Commander in Chief. These missives, passed from one POTUS to the next, have ranged from light-hearted jokes to profound insights. Both practical and inspirational.
Beyond my curiosity about what (if anything) Mr Trump may choose to impart to president-elect Joe Biden, it struck me that I wish that such a tradition existed in the (typically) square office of an incoming chief data officer. And wouldn’t this ‘been-there-done-that’ advice be even more pertinent for the new data lead of a start-up?
Having held the role of CDO — see notes below — for a tech-start up myself, this is the letter I’d write and the 3 tips I wish I’d known about data execution when I’d started the role:
Being a CDO in a technology start-up is a thrilling experience. You are directly influencing the product, working with smart people, using the latest technology and solving stimulating problems.
But it’s not for the faint hearted. Start-ups are not all space hoppers and sneakers! The environment is invariably messy and the challenges stressful, sometimes resulting in misaligned messages, MIA colleagues and equally wayward deadlines.
The famous axiom ‘the only constant is change’ couldn’t be truer of the start-up world. CDO is a unique office, without a clear blueprint for success. I’m sure you’re asking yourself how can a CDO make sure the data strategy is being executed decisively and swiftly?
Well, let me offer 3 practical reflections of what has worked for me over my past 10 years of experience:
1. Communicate extensively with the whole business 2. Limit yourself to three strategic priorities 3. Be deliberate in adopting new methods and technologies
Let me walk you through these in more detail.
Communicate extensively with the whole business
Approach every interaction with ambition to learn more about the business. A previous company I worked for measured workplace behaviours with a view to lifting business performance over time. I had zero expertise in behavioural sciences and large-scale organisational change. So, I actively chose to spend a lot of time talking with my colleagues to deeply understand the nature of the business, and thus be able to support them and our strategic agenda through data and analytics.
During these interactions I also made every effort to explain what data and analytics could and — crucially — could not do for the company. Plus, in acknowledging that not everyone was an expert in analytics, I limited myself to just a few core messages, explained them as clearly as possible, in different ways, and with many, many repetitions. I’m going to repeat that: many, many repetitions.
Beyond making the most out of each interaction (most of which were informal) I also introduced a more formalised process of post-project presentations. At the end of each ‘sprint’ the team openly shared successes and failures from our projects. The outcome was twofold. Firstly, the team learned to present their projects in five minutes to colleagues with various backgrounds. And secondly, our colleagues got excited and inspired by data, which generated a lot of beautiful ideas and generally made the life of the data team easier. While initially rusty, these team sessions evolved quickly and beyond my expectations. Yes, all of this was a lot of hard work. But it was definitely worth every effort. Even if one can’t communicate with the entire business, do it at least with the key people — like the leadership team, dev team, product owners or research team.
Limit yourself to three strategic priorities
You want to achieve so many things! And management probably wants you to accomplish even more. But resist the temptation to chase too many rabbits. And don’t let the critics discourage you or push you off course. Cut through to what truly matters.
In my first year as a start-up CDO, my top three strategic priorities were broadly:
Evolving our machine-learning models and other analytics to the next level;
Upgrading the data engineering and MLOps; and
Crystallising and tightening the relationship between analytics, data and the product.
Having just three priorities enabled me clearly (and repeatedly!) communicate the data team’s ambitions and ensure that the whole company was familiar with our aims. It also made the prioritisation of individual projects far easier and the data team knew exactly how their projects fit into the overall picture.
Be deliberate in adopting new methods and technologies
We data scientists, machine learning engineers and data engineers have curious minds. And we all love to play with the newest toys (see ‘The Buzz Lightyear effect’). However, it is crucial to strike the right balance between following every trend and ignoring industry progress altogether.
Remember, every decision to try and deploy a new method or technology needs to be made by weighing up all aspects of such a move. Ask yourself: What would be the costs? What would be the impact to the existing solution? What would be the benefits? How likely would we achieve them? How long would the implementation ‘dip’ take?
Alas, there’s no a silver bullet to this problem. But I believe as the chief data scientist it’s good to stir the status quo if it’s too quiet and play a devil’s advocate when new technologies and methods are suggested by others.
There are many challenges on the way, but diligently following these three tips certainly made life easier for me, my team and the business as a whole. I certainly hope they hold true for you too.
Note: I’m using terms CDO, chief data scientist interchangeably for anyone responsible for the data strategy in the organisation. It could also be a VP of analytics or one of the co-founders for that matter.
As ever, I’m indefinitely grateful to Chelsea Wilkinson for patiently shaping my thoughts into a publishable format and pointing me to the POTUS tradition.