The work of data science teams doesn’t always fit neatly into the daily workflows of other departments, so data teams tend to get siloed away. It’s a common problem that results in other groups, despite being in the same company, not interacting much with the data team or understanding what value it is capable of delivering.
Some people attribute this problem to data scientists being shy. After all, data scientists are numbers people, and numbers people are introverts and don’t like communicating with other people, right?
6 Tips to Improve Data Science Collaboration With Other Teams
- Don’t be a data gatekeeper.
- Get to know the needs of other teams.
- Get to know other departments on a personal level.
- Bring leadership on board.
- Teach other departments how data science can add value.
- Meet others where they are.
“That’s rubbish,” said Dina Mohammad-Laity, VP of data at Feeld, an online dating platform. “My team is very chatty.”
Whether data science teams feel connected to other parts of the company depends on the amount of work all parties put into the relationship between teams, which can also be influenced by company leaders and the ways data teams communicate their successes to the larger company.
If your data science team is feeling isolated, here are some techniques that can help.
Don’t Be a Data Gatekeeper
Teams that work in silos aren’t good for companies. Different departments like data, engineering, sales and product all need cross-collaboration to create better products for customers. But data science teams sometimes themselves put up counterproductive barriers that lead to the teams being siloed, according to Mohammad-Laity.
In practice, this can take the form of data scientists cordoning off the type of work that does or doesn’t lie within their jurisdiction.
“I hate this concept of ‘data people,’” she said. “It’s unnecessary gatekeeping … Everyone already knows you’re clever. You don’t have to put up barriers for other people.”
Data scientists should encourage others at the company who may not officially work on any data teams to learn data skills, Mohammad-Laity said. They can provide guidance that empowers members of other teams, like sales, to pull data about potential target customers and products themselves to inform their work.
When they help other departments learn how to gain insights from their data, data teams don’t seem quite as intimidating. Other departments may be more likely to seek out the data team for data advice and to talk more generally about other aspects of the business.
Get to Know the Needs of Other Teams
Data science teams often act as support teams for other departments within a company, providing insights about a product’s user base to marketing or about a product’s pain points to engineering. But effective data support for other departments is impossible if the data team is isolated.
A good way to bridge the distance is for data science teams to better understand the needs of other departments, said Saman Pourkermani, operations transformation lead at management consultancy Inspirant Group. That includes taking a closer look at who the data team serves.
“The data scientists can help them provide better services, come up with more innovative solution ideas, optimize how they deliver services and value to the rest of the organization, or even to external customers.”
“Try to understand who your customer is within the broader organization and what kind of value data scientists provide to those customers,” Pourkermani said. “The data scientists can help them provide better services, come up with more innovative solution ideas, optimize how they deliver services and value to the rest of the organization, or even to external customers.”
He emphasized the importance of asking questions about the type of work people in other departments do, for example, asking what the most time-consuming part of their work is. Sometimes data scientists might have the wrong impression about a department’s day-to-day work and the type of data insight that would be most beneficial to them — but having the customer drive the conversation about their needs leads to better results.
“Don’t make any assumptions around what it is that sales does,” Pourkermani said, as an example. “You’re going to learn a lot more going there with a mind to learn. … It goes back to understanding your customers, empathizing with them, trying to figure out what they care about — and then you can come up with much better solutions that you can offer to them.”
Get to Know Other Departments on a Personal Level
In addition to learning about different departments on a business level, Pourkermani said it’s also important to get to know the people that work in those departments on a personal level. Forming personal bonds with people in other departments is one of the best ways to foster cross-departmental collaboration and bring data science teams out of isolation.
Companies often have social or work events where employees from different departments are able to chat and mingle, and Pourkermani strongly encouraged data scientists to attend them rather than sit them out. Those are opportunities for teams to get to know one another and not necessarily talk about business, but that doesn’t mean they don’t have an effect on the business.
“Inevitably, work is going to come up,” Pourkermani said. “And as work comes up, listen to some of the challenges that they’re facing.”
If some of the challenges other teams bring up seem to align with what data science is able to help with, data scientists can step in and explain how their services work.
“Say, ‘You know what, we worked with a team that had similar challenges, and here’s how they went about it,’” he said.
Make Sure Leadership Is On Board
Ultimately, data science teams may continue to be isolated if those in leadership positions don’t understand and advocate for them, Pourkermani said. Leadership teams dictate the role and function of different teams. If leaders at a company understand how data can be used to help the organization, they will bake the importance of data teams into the broader structure of the organization.
If leaders aren’t familiar with how data science teams typically operate, teams can reach out and offer to walk through projects they have worked on and examples of ways teams have contributed to the success of the company. Leaders can also get good ideas of ways to leverage data in getting to their business goals that they may not have been aware of before.
Teach Others About What the Data Team Can Offer
Beyond just the leadership team, data scientists can also reach out to other departments to educate them on how the data team can help diverse departments gain insights and overcome challenges.
“Start with helping them be aware of the value that data scientists can provide to that specific function, like marketing,” Pourkermani said. “You need to understand their function, understand what they do, learn about their business and operations and use that to determine, ‘How can I add value and provide value and help them?’”
One of the best ways to show people from different departments how data science can help them is by giving them examples of similar situations where data has helped other teams. This can be as informal as bringing up successful past projects within the company or as planned out as presenting case studies.
“A good tactical way to do that, especially with new functions that pop up within organizations, is with roadshows,” Pourkermani said, referring to information fairs that introduce the data team’s offerings to different departments. “You get to build relationships, get introductions to the management of these departments and influential people in these departments.”
Another way to teach others about the data team is by including data success stories in the company newsletter. Some companies have their own internal weekly or monthly newsletters that share news from different departments, and that can be a good medium for getting the word about what the data team does out to a wider audience.
‘Meet People Where They Are’
Mohammad-Laity used to get frustrated when she worked with colleagues in other departments that had trouble understanding the data science concepts she tried to explain to them. Even when she felt she had simplified data concepts a lot, they weren’t always able to keep up with the technical concepts.
“I was expecting the stakeholders to meet me halfway,” she said. “And I look back on it now and I’m like, ‘That is such an unreasonable expectation.’”
“Sometimes the most reasonable thing is that they’ll move 1 percent and you move 99 percent, and you bring them with you, and that’s more exciting. If you bring them on a journey, you become better because you understand what they’re trying to do better.”
These days, Mohammad-Laity doesn’t look at the situation in the same way. She sees it as more of her job to bring the discussion to the level of whoever she is speaking to — some useful techniques that work for her include using different analogies to illustrate different data concepts. When she’s explaining complicated data concepts like Markov chains, Mohammad-Laity said she makes analogies to subjects the person knows well — how credit for soccer goals can be attributed to more than just the player who scored the goal, for example.
“My thing that I tried to do now is just to try and meet people where they are,” she said. “Sometimes the most reasonable thing is that they’ll move 1 percent and you move 99 percent, and you bring them with you, and that’s more exciting. If you bring them on a journey, you become better because you understand what they’re trying to do better.”
Kerry Halladay contributed reporting to this story.
