For today’s businesses, the ability to convert data into meaningful insights and intelligent action represents a make-or-break factor. Effective data science teams can therefore have a massive impact on the business. These teams can enhance business performance at multiple levels, including boosting productivity, innovation, and revenue. Given this, in a very real sense, for the business to succeed, the data science team needs to succeed.
However, it is often hard for business leaders to collaborate with data science teams. Organizational silos, different backgrounds and skill sets, and limited transparency can all pose obstacles, and many of those challenges are only more difficult now that remote work is the norm. In this post, we will discuss the biggest issues that hinder collaboration between business leaders and their data science teams. Then, we’ll show you how these issues can be remediated.
Have the Right Organization Structure
Most data science teams are small and understaffed due to the fact that it is hard to find data science talent. Organizations can structure their data science teams in one of two ways: the centralized or the decentralized model. In the former, a single data science team serves the entire organization. In the latter, there are multiple data science teams serving different business functions or teams
Most organizations use the centralized model, in which one data science team serves all business teams. This works well in the beginning, but it’s difficult to scale as the organization grows and the workload increases. For large organizations, data science teams should become increasingly decentralized.
As a business leader in your organization, you should pay attention to the number of data scientists per team (or per the business functions that they serve). The higher the ratio of data scientists per team, the better. While this is the responsibility of chief data officers (CDOs), they’d be happy to have you champion their cause.
Involve the CDO from the Start
Before your project begins, you should set up a meeting with your CDO to discuss your requirements. If you have many routine requests, you may want to schedule a recurring monthly meeting with him or her.
Data science is a blend of the traditional business analyst and a developer with strong machine learning skills, and each project will require a balance of these skills. The CDO is the best person to advise you on which resources would be best-suited to your project.
Your CDO’s input can be incredibly valuable for your project—and your stakeholders. He or she might suggest minor changes or even a complete overhaul of your project’s objectives. It’s important to be open-minded.
Know the Difference Between a Business Analyst and a Data Scientist
Business analysts derive insights from historical data in order to inform business operations. For example, they can determine how your expenses per department have changed over the past four quarters.
Data scientists, on the other hand, are best suited for jobs that involve machine learning. For example, data scientists might be responsible for:
- Analyzing customer data to make better product recommendations
- Forecasting website traffic before a big marketing campaign
- Predicting the credit risk posed by new customers
- Using pattern recognition to spot the cause of an anomaly
- Visualizing target markets with pinpoint accuracy
- Estimating the impact of the pandemic on the organization’s performance
You’ll win over your data science teams if you approach them with problems that a business analyst cannot solve.
Take a Project Management Approach
Project management best practices can help you collaborate with your data science teams. Clarify project objectives and establish your expectations from the start. Use a project management tool to track the progress of complex, multi-step projects. This will keep you and your data science teams in sync. You should review the status of your project frequently, ideally every week on a set day of the week; say Friday for example.
The Importance of Good Timing
Since data science teams are almost always understaffed, the way in which you organize and send them requests will make a big difference. If you have a lot of work to get done, throwing it to the team all at once may only overwhelm them. Instead, break big projects down into smaller ones. Bigger projects can run into bottlenecks and get stalled; breaking them down will give your data science teams (and you!) more attainable milestones at every step of the way.
Expedite Requests for Resources or Data
Sometimes data science teams need IT assistance that must be approved by the finance or HR team. Or they may need a developer’s help to create a custom script or plugin for an application. In these cases, you should do your best to expedite approvals and give your data science teams the clearance and support they need to get their jobs done.
These approvals and one-off development projects often aren’t accounted for, and can fly under the radar, only to be recognized when it’s too late and the project is delayed. Here again, having accurate project trackers will help expose these bottlenecks as they occur, and you can even anticipate some of them right from the start.
Speak the Same Language
Key terms like orders, sales, and deals can be used interchangeably for the same data by different teams. This can be confusing. Some datasets can have very similar names, and pulling the wrong data will only lead to incorrect results.
In order to speak the same language, you and your data science teams need to see the same metrics and make sure that you agree on the key terms and data sources. This is especially important when you brief your teams.
Build Rapport with Data Science Teams
During this era of remote work, there’s more room for communication gaps to creep in. It is important to invest in building rapport with your data science teams from the start. If this means indulging in post-call banter about how they used R to crunch this season’s major-league baseball stats, then so be it!
In addition, you should avoid micromanaging your data scientists. Given the complexity of their tasks, data scientists may not have the bandwidth to listen to business leaders who try to tell them how to do their jobs.
Conclusion
By keeping these best practices in mind, you can enjoy a productive and happy working relationship with your data science teams. You can actually become their best client by learning how they function and adapting accordingly. Most importantly, helping your data science teams to succeed in their jobs will help you to succeed in yours.
