Becoming Product Science
Creating a multi-skilled data team in a large organisation
At the start of the year we began trialling a new role bringing together two important aspects of our team: Product Management and Data Science.
The most common challenges we recognised as a team were how we were communicating both what we were doing and the process we followed, and how to realise the value of our output into something tangible for our business area.
Historically we had tackled this by hiring dedicated Data Product leads who navigated processes and were in charge of delivery and implementation. Increasingly, however, we found that bringing our Data Scientists into the ideation and strategy process allowed them to build out their skillset as creative problem solvers.
This got us thinking about the skills that were important to us (both within our team and the wider organisation) and the importance of creating an environment where individuals can continuously develop.
Our starting principles for Product Science were therefore the following:
1. Being multi skilled
2. Communicating stories from our data
3. Enabling ongoing opportunities to develop
1. Creating a multi skilled team
We weren’t alone in looking at our team roles as two distinct remits: product and science. In LinkedIn’s list of most promising jobs for 2019, Data Scientist is top with 4k+ job openings and 56% YoY growth, but the skills listed for the role are fairly standard: mining, analysis, python, machine learning. However when you look at the most promising soft skills of the year they are: Creativity, Persuasion, Collaboration, Adaptability, Time Management.
These soft skills are essential to being successful in a large organisation and yet we as an industry often fall into the trap of just assessing technical skills for Data Science candidates. This focus can also be misrepresentative of the environment an individual is being hired into — one that takes as much resilience and organisational savviness as technical expertise.
Taking inspiration externally also enabled us to think more broadly about what being ‘multi skilled’ means. Take positionless basketball as an example. Traditionally the team strategy has been based on predefined positions, with stereotypical traits defining each players role such as height or speed. Guided by analytics, this approach is changing. The game is becoming more offense orientated demanding players to become more versatile and flexible. This means that to remain competitive modern players have to adapt to a range of skills across all 5 historical positions.
Much like a basketball team, a Data Science team is expensive and hard to build. This can mean that you fall into the trap of wanting the team to do Data Science work. But when you engage your team in new and different ways this is where you’ll get a huge benefit.
2. Learning to storytell
In 2008 Google’s Chief Economist Dr. Varian said, “The ability to take data — to be able to understand it, to process it, to extract value from it, to visualize it, to communicate it — that’s going to be a hugely important skill in the next decade.” Well, we’re over a decade later and there hasn’t been a vast improvement in the communication. When you invest so much time in your methodology, understanding the data, the analysis and peer code reviews, the ability to tell your story can be easily dismissed as an afterthought or a waste of time.
Carmine Gallo, the author of Talk Like TED, argues stories are the biggest driver on being rated a successful TED talk. In his analysis of over 150 hours of the most successful talks, stories made up 65% of the content. And whilst stories themselves matter, so does how we communicate them. Humans have learnt how to communicate through visual language, from hieroglyphics to emojis, so the stronger your skillset in in visually communicating data, the more powerful your impact.
For us, storytelling means two things. The first is constant communication through the exploration and data profiling process. The transparency of the process followed and the packaging of insight along the way have become really important in keeping our stakeholders engaged (and understand what we were doing!).
The second is putting the business context, or ‘so what’, around every piece of insight. Constantly challenging ourselves to think about this from a commercial view point is critical to getting the right feedback (and critique) on the statistical approach.
3. Creating an environment where people develop
Going back to our sport analogy, what becomes apparent is that focusing on people, over positions, has made a demonstrable improvement to the success of the team. Freedom, happiness, flexibility, growing — this is what has driven success for the sport.
Product Science allows this flexibility of skills in the same way. Building a concept of being ‘multi-skilled’ allows individuals to test and build new skills that previously might have not been core to their role, and hence ‘off limits’ when time is tight.
Providing the time, and most importantly space, to experiment, to work on different projects and test strengths in new areas is key. Culturally, what becomes really important is understanding your team members aspirations — aligning your team’s future aspirations and dreams to skills they can develop now means you have a far higher chance of keeping them in your team for as long as possible.
The results?
This isn’t yet fully embedded across our team, but the results so far from following these principles have been really positive. Our Data Product leads have started to code, sending our sprint updates in Jupyter notebooks and our Data Scientists are joining workshops, gaining a deeper understanding of our dataset by viewing it with a business lens.
This isn’t without its challenges — both Data Product and Data Science roles already have a large remit associated with them individually and pulling these together can be overwhelming for just one person. However by taking a more flexible multi-skilled approach we’ve recognised that no one team member needs to be the unicorn…we’re stronger together and we’re working as one team better than ever.
