The Three Skills You Need to Instill in Your Data Science Team
Data science and AI have enormous potential to make your organization run more efficiently and profitably. They can be used to optimize processes, identify emerging risk and fraud attempts, spot emerging commercial opportunities, and serve your customers better. You know that. Your team knows that. But does the rest of your organization know?
Top data scientists are highly sought-after, but data scientists who deeply understand the needs and priorities of the wider business are like gold dust. Sure, your team members might be wizards with Python and R. They may know how to create innovative models using existing libraries. But to be truly indispensable to the company, they also need to be masters of specialization, communication, and driving business value.
The data science field is evolving fast. The rise of the IoT means Big Data sources are proliferating. New regulations and standards change how you can legally capture and use data from users in different parts of the world. Dynamic, real-time data flows require a different approach to handling historical data. Your approach to a machine learning problem will be very different depending on whether you’re trying to identify a fraudulent transaction or build a personalized recommendation engine.
What’s more, every industry, sector, and business model is different. Knowing what data science and AI can do for the organization isn’t just about machine learning skills. It’s also about having the domain knowledge to put that technical prowess in context and to spot new use cases based on that understanding. A data scientist who also knows the ins and outs of a particular area of the business (be that marketing, operations, sales, or another critical function) will be a major asset.
That means it will be much more helpful for your team to specialize in the areas of data science and machine learning that are most relevant to your organization’s goals than to demonstrate that they know how to use many different scripting languages, for example. The value of your team isn’t that you can build many different types of machine learning models; it’s that you know exactly what kind of model will answer your company’s most business-critical questions and tackle their most challenging questions.
People respond far better to a memorable story than they do to dry stats and figures. Try to present your proposals in a way that not only offers a compelling narrative about the business successes your project will bring but that makes them the hero of the story. The key is to make heads of departments understand that your team isn’t simply offering a tool or resource but a strategy that they can use to make themselves indispensable to the company, too.
For this to work, of course, your team also needs to be adept at talking to your colleagues from across the organization to get a sense of what they need to know in order to succeed, what domain knowledge they can bring to the design of your data science team’s models, and what data sources or external datasets would really fill in the organization’s internal knowledge gaps.
The important thing here is collaboration — and that starts with communication. Encourage your team to translate the language of data science and AI into business-speak that their colleagues can understand. Use data visualizations and demo your models to make things clear.
The fact is, if your colleagues don’t understand what you do or how it will help them, they probably won’t tell you — they’ll just avoid coming to you, which means you can’t be useful to them. Using impenetrable jargon and data science language is ultimately self-defeating. Get your team used to explaining their work in more accessible terms.
Driving business impact
In a business context, data science is all about driving value, delivering solid business benefits, and boosting your company’s ROI. Ultimately, this is what makes you valuable to the wider organization.
One of the biggest problems with machine learning projects is that they often involve long lead times to prototype and test, with no guarantee that they’ll make it into operation. Experimentation may lead to significant business benefits, but it may also lead nowhere, and this risks undermining ROI and losing the confidence of your company’s decision-makers.
This is why it’s so important to keep business impact in mind right from the start, looking at the big picture. How will you scale your machine learning models to maximize ROI? How could you deploy AI right across the funnel? Where are the quick wins for optimization?
And, how might you streamline the development process so that less is wasted if a model doesn’t work out? Could you automate some of the preprocessing? Would a machine learning platform that incorporates augmented data discovery and some AutoML help? Where can you find valuable training data that you can feed straight into the model without spending days cleaning and harmonizing it first?
Your data science team doesn’t operate in a bubble. It’s part of a wider organization that needs to drive ROI and business impact to survive and thrive. That means you can’t afford to just stay in your lane, focusing on machine learning and AI. You need to understand how to relate what you do to the wider organization and to make your business colleagues partners in what you do.
Every team in the data science field needs the three skills we’ve talked about here. If you don’t have a specialist on your team, you’ll find it more difficult to bridge the gap between business and science languages. Without a communicator, you won’t get the buy-in and resources you need to grow, find your place in the organization, and deliver results. Without focusing on real business needs and ROI, you might come up with some nifty ideas, but you won’t deliver real benefit to the company.
By developing these skills, you won’t only make your data science team more valuable, you’ll also get better at explaining why you are so valuable. Ultimately, it’s this that will put you and your team at the center of the organization.