How to improve data scientist job satisfaction

Understand what you want to achieve

It's important to have a clear vision of what your organization wants to achieve with data science because you have to sell that vision to candidates.

"[Some companies] haven't thought through at a strategic level and at a tactical level, what is the role of the data scientist in our firm," said Michael Walker, founder and president of the Data Science Association. "They say we have all this data, there has to be a way to monetize it to make credit decisions, to sell more product, but they really have no clue in terms of how to use the data scientist and how to use data to achieve a specific goal."

Avoid bait and switch scenarios

Companies don't intentionally misrepresent the data scientist role, but they don't always understand what a data scientist actually does. As a result, candidates may be told they'll be doing advanced, technically challenging work, but the job could be more mundane than that.

"I think some companies fall into the trap of selling data scientists," said Steven Mills, partner and associate director of machine learning and AI for global management consulting firm Boston Consulting Group's (BCG) federal division. "Disillusionment is one reason for massive exits. They're not getting to use a diversity of techniques or they're not approaching many different kinds of problems."

Provide intellectual challenges and growth opportunities

Boredom is the number one reason data scientists leave their jobs, according to Walker.

"[A] lot of data scientists are bored. They're not doing what they were trained to do, and they'll leave," Walker said. "If you don't give them intellectual challenges and make them see how they're adding value to the firm, they're not going to feel fulfilled."

Data scientists, like other employees, want to advance their careers. If they're not being challenged, finding another job will be easy.

"If you're just starting a data science team, you're sort of selling them on coming in and helping you figure out what you're even going to do in the space," BCG's Mills said. "If you've got a more mature team, you're going to sell them more on a career path, upward mobility and a really solid chance to learn."

Support continuous learning

Data scientists need to continuously update their skills in order to be effective. Organizations should include educational support in their budgets.

Continuous learning opportunities can take several forms, including allowing them to work on different kinds of projects, sending them to conferences or enabling them to take online courses.

Mills said many data scientists want to feel connected to a higher purpose.

"I've just consistently found that giving people that kind of outlet, even if it's as simple as letting them host a meet up in my facility or do a hackathon will allow them connect to a community," Mills said.

Market your company appropriately

Some large organizations have all-star teams that attract top data scientists. Other companies have little or no visibility in the data science community, which makes it harder for them to compete for talent.

"If you don't regularly interact with the data science community, you have to do a bit of intentional outreach to give them visibility into what you're doing, why you're doing it and get them excited to come work for you," Mills said.

Jim Johnson, senior vice president at global staffing firm Robert Half, said it's a mistake to assume everyone knows your company is a great place to work. You must communicate that to candidates and make sure your existing talent is engaged in developing the message.

Don't expect too much from one data scientist

Some companies seek "unicorns" who are mathematical and statistical geniuses, ninja coders and business domain experts who understand everything even remotely related to data. However, these unrealistic expectations aren't helpful.

"A lot of organizations hire a data scientist as a lone wolf, some genius [who is] expected to do extraordinary things without putting together a team, without thinking about what the goals are, without including other stakeholders in the firm," Data Science Association's Walker said. "[T]hey need to create a data science team with different people who have different expertise levels and they need to be integrated into the company."

Finding a single person who has mastered machine learning, software development and infrastructure is tough, which is why a data science team includes other roles such as data engineer and data architect.

Maintain an ongoing dialog

One of the best ways to monitor data scientist job satisfaction is to maintain an ongoing dialogue with them.

If you're surprised when somebody leaves, you've failed as a leader because you weren't talking to them ... about whether they were happy doing what they were doing or what their issues were.

"Just sit and talk to them and really listen to them. Make them feel that they're having input on the direction of the team and their own careers. That alone will give you a lot of warning signs," BCG's Mills said. "If you're surprised when somebody leaves, you've failed as a leader because you weren't talking to them, you really weren't getting the story from them about whether they were happy doing what they were doing or what their issues were."

Robert Half's Johnson also stressed the need for solid communication.

"I think the most important thing is to make sure that the door is always open," Johnson said. "Don't just say, 'Here's what's happening.' Say, 'How are you feeling about that? What can I do to help you feel engaged, to help you feel like we're committed to your development?'"

Ensure organizational support

Organizations can be their own worst enemies when it comes to creating a successful data science function. Missteps include isolating data scientists, consistently dismissing their findings or demanding they contort data into saying what business leaders want it to say.

"It's important for data scientists to be really engaged and have them work outside of their core data science team," said Brian Shepherd, executive recruiter at Burtch Works. "When we see candidates coming in and applying for any of these jobs that we have posted, it's usually because there is that lack of involvement."

Johnson said that acknowledgment and recognition are also important.

"When somebody is coming after you, whether it's a recruiter or someone else, they're telling you how great your background is and how valuable you are. [Meanwhile, your employer] is piling more work on your desk and asking you to work overtime," Johnson said. "While that work may well need to be done, don't forget to pat them on the back. Recognize them and help them continue to develop and enhance their skill set."

Pay appropriately

According to a 2019 Burtch Works report, U.S. data scientists with advanced degrees earn a median salary of $90,000 to $250,000 based on several factors, including level of education, location, years of experience and level of management responsibility.

"Make sure you know what the averages are locally, what people are being offered, what your current team is being paid," Johnson said. "Also [consider] what else you can offer. Is there potential for flex work schedules, remote work or education or tuition reimbursement? While you don't have to include that in every offer, it's important your organization is clear on where the flexibility is because you may need to offer those things if you can't hit a dollar figure."

Walker said that even data scientists making $250,000 to $300,000 may accept $50,000 less to feel great about what they're doing.

"If I go to work and feel like I'm doing something great, I'm using my mind, I'm adding value, I work with great people, I feel good -- that's priceless," Walker said.