A growing number of companies are setting up their data science teams to get maximum results from their data. But it’s not that simple to set up an effective data science team. There are many different types of people who work in data science. Today I would like to share some insights that I have gained in this field and hope that they can help you set up the right data science team for you.
Why Do We Need Data Science?
Data science is about the use cases and where you need to start. You’ll need a team to help you improve your targeted media campaigns, and you’ll also need a team to help you optimize your core business algorithms. They’re very different animals.
While it might be tempting to hire for everything, we know that hiring for everything usually ends up hiring for nothing, and that’s not a good way to start your team. It’s impossible to set up a data science function without first defining the problem. It will be helpful to seek help from outside experts to support you in defining the problem and its potential solution.
Will Organizations Be Able to Act on It?
It’s common for data science teams to fail because they can’t integrate with the organization and senior stakeholders often underestimate the level of change required to make it work. A key factor determining how much change management is required here is the size and maturity of the organization. There is often initial inertia or resistance when companies introduce new products or services.
Data Science is based on the fundamental principle of test and learn, and it requires the organization to have a culture that is willing to change and learn from the data. It’s as important to think about how you can bring change to the organization as the actual change you will bring. So, while building your data science team, make sure to do the same amount of work on your organization to change culture as you would any other part of your business.
How Do I Set up The Team Structure?
Now the question is how to build data science team? When trying to grow your business from nothing to something, creating an independent team that only focuses on transformation and innovation is tempting. But if this approach doesn’t work, don’t do it. Innovation cannot happen in a siloed environment.
Each functional team is responsible for setting its standards and efficiency gains, but they should probably hire a data scientist. The debate on whether data science should be done as a centrally coordinated function or a distributed one is old, going back at least to the first book on data science. However, I think the trend toward dual reporting in a matrix structure will be the inevitable solution.
The Data Science Lead reports directly to the Data Science Lead and the Functional Team Lead but is responsible for working closely with all functional teams. I don’t think it’s a silver bullet, and managing that structure will make the process more complex. But the benefit should outweigh the costs if done well.
Don’t Underestimate the Value of Leadership.
Good hire for a data scientist should be someone who’s there to help build the team and has a passion for data analysis. I’ve seen lots of companies hiring a technical data scientist with the hope of moving up the corporate ladder when they want to grow as a team leader. I don’t think that approach works because there is a distinction between technical and team leaders.
If the team sponsor has a clear understanding of what data science should look like and have the capacity to lead, you may be able to get away with hiring a team of technical resources. You’ll be surprised how much of a benefit it can have when someone understands the value of leadership within a technical context and can dedicate all of her efforts to leading the data science team. When selecting a leader, you should not have an afterthought, and your first hire needs to be able to lead if you want the team to grow.
Problem-solving skills over expertise with tools:
As more and more new tools come out daily, expertise in using each tool is not nearly as valuable as it used to be. Data scientists should always look for answers to questions, not focus on the tools. That is what data science should be all about.
To be a successful data scientist, you must understand the problem and come up with the right methodology. You can assess this by having the candidate walk you through a problem he’s solved in the past and look for which aspects he emphasized.
