What makes a great data analytics team
It was a gloomy winter afternoon. Not having seen the sun for a while, the streets of London were lit with artificial lighting. The windows of the design agencies and artisan coffee shops north of Oxford Street contrasted with the grey sky, evoking a dream of a comfy chair by the fireside.
The huge conference hall was abuzz. People were finishing their drinks and settling into their seats when the host announced the start of the Team of the Year competition.
Data analyst team leader Shadi entered the stage with her notes and got on the stand. Hundreds of eyes watched her from the audience. Shadi wasn’t nervous. She knew how to make a great speech. The crowd went silent, and Shadi heard her team gathering around to support her. She knew they were going to win the Team of the Year prize.
While Shadi is delivering her awesome speech, I would like to invite you to take a look at how we work. I like my team. I sometimes hear that the data analytics team in our company has some special qualities. At first sight, it seems just like any ‘normal’ team that you would expect to find in a BI department. I have decided to analyse what our efficiency is based on and share it with you. I hope you will find it interesting. Let’s go!
Diversity
According to many studies, the diversity of team members is one of the most important factors that determine a team’s success. For example, McKinsey’s 2018 report proved the positive impact of diversity leadership on a company’s financial outcome. In 2017, BCG and the Technical University of Munich studied how diversity in industrial background, country of origin, career path and gender impacts the innovation performance of a team and found a positive correlation.
Our team is based in London. However, apart from people from the UK, we also have colleagues from Iran, China, Argentina, Russia and the USA. In terms of gender balance we are getting better, and at the moment 70% of our managers are female!
While there are no crazy exceptions (we haven’t hired any astronauts or orchestra conductors yet), we have quite a variety of professional backgrounds: we have colleagues who previously held product management, software engineers and data science positions.
‘All analysts are super approachable, and everyone has different expertise that can be used to help us all.’
And now let’s look at what makes our team members excited and happy. Having hobbies and personal interests is proven to be beneficial for productivity at work. Believe it or not, some people think that all data analysts only enjoy building machine learning models in their free time. The truth is, we are all different and that’s awesome. Among the data analysts in my team there are certified wine experts, field hockey players, bicycle lovers who have cycled hundreds of kilometres in the wild, screenwriters and vegetarian cooks. While many of the analysts have ‘traditional’ degrees in mathematics and statistics, some have quite different academic backgrounds, such as culture studies and psychology!
Continuous learning
Now, let’s talk about serious stuff. Technologies and skills. If you are reading this article by accident and don’t identify as a data analyst, you may want to skip this part. Or you may well want to read on and join our kingdom, which is always open for a seeking soul.
The main tools that we use today are Jupyter Notebooks, Python and SQL, which is quite a standard set for data analytics teams. Every one of our analysts is proficient in all of these, and sometimes they use R. Our databases of choice are Exasol and Hadoop. We have our in-house analytical infrastructure with a lot of features that are maintained by data engineering and BI portal development teams. However, the analysts in my team can also take an initiative and contribute to improving the tools and our own Python libraries. Such contributions not only help to make infrastructure better for everyone but also can be a good opportunity to advance coding skills.
‘Everyone is able to find a niche they are interested in and work in those areas.’
In our team, we adopt new approaches to analysis all the time. For example, we have recently used a lot of causal impact analyses, linguistic processing, image recognition, forecasting, geospatial clustering and much more.
One important part of our work is A/B testing, and we always promote the idea of proper hypothesis testing before making any change. Recently, we participated in developing a new A/B testing framework that allows us to make more accurate decisions in less time.
In order to continuously explore new ideas and improve our skills, we data analysts have our own training budgets and visit both world-renowned and niche professional conferences around the world.
Getting the processes right
We have a bunch of talented, highly skilled professionals. How can we make them work together so that they share a common vision and get the benefit of belonging to a team?
In our data analyst team, this is achieved through simple but efficient processes that we have established over time.
Units. All analysts are assigned to ‘units’. A unit is a cross-functional team that is dedicated to a particular area of the business and has ownership of its KPIs. So, for example, there is an ‘algorithms’ unit which is responsible for the matchmaking system and a ‘CRM’ unit that is responsible for communications, etc. It’s a very flexible structure that can be easily adjusted to the current business objectives and priorities. New units appear when there is a new project, and old units can be dissolved if they are no longer needed. The rotation of analysts between the units makes it a ‘change-resistant’ system where most of the analysts have knowledge of several business areas. Every unit has its own backlog, which is prioritised by the analysts and product managers.
Reviews. Another important thing that we have is a code review. Each piece of analysis gets reviewed by another analyst before being shared with the target audience. The reviewer goes through all aspects of the analysis — the code, the visualisation, the conclusions — and produces a list of notes. The most obvious purpose of the review is to make sure that everything is correct and there are no mistakes in the code. However, there is another very important goal — to suggest a second opinion about the overall analytical approach and conclusions. Because all our analysts have a unique background, having your analysis reviewed by a colleague can help you see things from a different perspective and open a new line of thought.
Weekly catch-ups. We keep quite a simple approach to meetings. Once a week, analysts from all units get together and share the updates and findings from the week’s work. It is a rewarding time because it gives everyone another opportunity to share the findings with other analysts and get feedback from different points of view. So far, it has been enough to enable us keep up to date and avoid the overlapping of projects.
Psychological safety
Do you want people in your team to take initiative, freely share ideas, be honest with feedback and easily admit mistakes? Google’s Aristotle research analysed what makes an effective team and found that the most important feature determining a team’s success is psychological safety. Psychological safety is defined as ‘a shared belief held by members of a team that the team is safe for interpersonal risk-taking’. It’s an ability to trust colleagues, ask questions and speak up without being seen as disruptive, incompetent or negative.
In our team, this is something that has been a high priority. In my opinion, it is the main ingredient in the team’s efficiency, so if you take away only one point from this article, please make it this one.
In order to determine a team’s level of psychological safety, you can ask the team members to rate several statements on a scale from 0 to 100 and then calculate the average:
- Working with members of this team, my unique skills and talents are valued and utilized.
- No one on this team would deliberately act in a way that undermines my efforts.
- It is safe to take a risk on this team.
- Members of this team are able to bring up and talk about problems and tough issues.
- It is difficult to ask other members of this team for help.
- People on this team sometimes reject others for being different.
- If I make a mistake on this team, it can be held against me.
If you are interested in building such a culture, I recommend taking a look at this IDEO course.
‘I can’t imagine many, if any, circumstances where I would feel unable to share my thoughts. It has a bit of a family feel to it, where we all kind of look out for each other.’
One of the things that we do every week to promote a healthy psychological environment is to share feelings. These are special moments that usually happen during weekly catch-ups where each analyst can tell the rest of the team how he or she feels about their week and other team members can show empathy or support if needed.
Another thing that we do to celebrate diversity and inclusion is talking about the differences in the team. We all are different and unique, and what works perfectly for one person can be very uncomfortable for another. Acknowledging the team members habits and personality traits can lead to better understanding and relationship, so I highly recommend this exercise.
Other cool practices that help in creating good culture include 1:1 sessions between the analysts and managers, team building events and trainings dedicated to inclusion, diversity and biases recognition.
Appropriate leadership
One of the essential parts of a great team is great leadership. Different kinds of management styles can be applied in order to achieve the best results in a particular situation based on team size, degree of communication, the personality of team members and other factors.
In our team, we value proactiveness and engagement in the decision-making process, which requires analysts to be as empowered as possible. As the team consists of highly skilled people who are capable of taking ownership of their goals and projects, the most appropriate management style is democratic. Team members managed by such a leader feel empowered to express their opinions and ideas, which improves motivation and commitment to results. Research shows that this participative management style leads to higher job satisfaction and productivity.
So, to conclude, I should say there is no magic in our team’s efficiency — only talented and diverse people, opportunities to learn, just enough of the processes to maintain the effective work, psychological safety and strong leadership!
By the way, speaking about leaders, what about Shadi, you may ask. How did she get on with her speech?
Worry not, dear reader. There she is, Shadi, on a stage, smiling, with a golden cup in her hands, surrounded by happy team mates, myself included. After company-wide voting, they have won the title of MagicLab Team of the Year.
And now, shall we turn the page and set out for new adventures? :)
