Data Scientists are just awesome and we are sure you’d like to hire a team of them to pimp up your business. But how do you hire your first Data Scientist? And how should you put together a team? Stay tuned and learn some important insights from the three DN19 speakers.
As Data Science is proving to be a real game-changer for businesses, companies have been jumping to install a Data Science team. But as with anything you try for the first time, much can go wrong. According to a study by Gartner, as many as 59% of IT organizations aren’t prepared for the necessary changes that come along.
So, how do you build a Data team like a boss? We asked three leading Data experts about the do’s and don’ts.
“Clearly outline the journey”
Sophie Brüggemann, Data Analyst at Spinnin’ Records (Warner Music)
As a Data Analyst, Sophie Brüggemann joined the sub-label ‘Spinnin’ Records’ of Warner Music, tasked to optimize strategies in music marketing and consumption analytics. She thinks it’s important that a company should be fully aware of what they want before they go and hire Data Scientists.
“Before a company hires a Data Scientist, they should ask themselves three questions,” says Brüggemann. “What data do we have? How do we store and combine the data into one database? And, what do we want to do with this Data?”
Be specific about what problem you want to solve and what kind of Data Scientist you need to hire. “Data Science encompasses many disciplines, its an umbrella term,” she says. Therefore, it’s important to get into detail about the expertise an applicant has. Most times when a company hires a Data Scientist for the first time, they actually need a Data Engineer.
“If you hire a Data Scientist for the first time, you can’t just expect him or her to organize and pipeline the data infrastructure,” she tells. “That is not what I would expect.”
She advises making it clear during a job interview what is expected of the Data Scientist. When Brüggemann was hired at Warner Music, she was well informed about her role in facilitating Data Science integration into the whole supply chain. “They clearly outlined the journey for me and put it forward as a challenge,” she tells.
Then it’s important to create the right data-minded environment at the office. The main problem is often when Data Scientists feel isolated. “It takes a lot of personal integrity to go for what the data says if you go against someone’s decision,” she says. “It doesn’t make you popular.”
That’s why it’s important to regular join business meetings when you hire, so they can connect the dots.
“If leaders of other departments understand our processes, they ask the right questions and can set project outlines,” she says.
“Simple is the best”
Lior Barak, Managing Partner at Tale about Data
Lior Barak, who was prior project manager at Zalando and Lovoo, agrees a company shouldn’t hire a Data Scientist before they have their data structured and their goals thought through. “Meaning, they know what data sources they use and have an alignment between the different teams regarding the KPI’s they are being measured upon,” he tells.
Then come the high expectations of companies. Barak thinks ideally, a company has developed some basic models before the Data Scientist arrives, who can then aim to improve these models by 10 till 15 percent at a time:
“Too big movements will cause people to lose trust in the models and lead to a failed project,” he says. “People don’t always trust machines, especially when the models are too complicated to understand.”
That’s why Barak advocates starting off simple.
“To build a rocketship to the moon, you first need to learn how to fly an airplane,” he says. Over the past years, he saw lots of companies trying to build complex models from day one and most of them failed. He thinks companies shouldn’t expect complex models too fast and also shouldn’t let a Data Scientist work on one for months: “It’s better to start with a small one-engine airplane.”
When Barak hired Data Scientists in the past, he always gave them a simple problem to solve. He found that the more experience a Data Scientist had, the more complex solution they came up with. “The main key is to understand how to solve simple problems,” he tells. “Simple problems should have simple solutions and every Data Scientist should learn to think pragmatically about problems.”
“Data is the new oil”
Zinayida Kensche, Data Scientist at Dropnostix
Zinayida Kensche points out that there isn’t always enough awareness about the value of data in companies. “Have you heard that data is the new oil?” Kensche says. “If not, print these words out and pin them above your screen, to read them every day.” Companies should be aware that Data Scientists don’t bring more work, but instead, open up opportunities.
She thinks companies can trust more on the judgment of Data Scientists, who often likes to make changes in a system or process. Companies should consider the benefits and then sell the insights well to all involved departments. In order to advocate and give insights into these changes, Data Scientists shouldn’t work in isolation, but in various departments: “Give her the possibility to dive into important fields of your company,” she tells.
Kensche stresses that it is important to trust and listen to one another.
“Sometimes managers have to accept that they don’t have enough expertise to decide which technologies or approaches are possible and in which time frames issues can be solved,” she tells. On the other hand, Data Scientists should listen carefully to managers and the sales team: “to avoid too complex and time-consuming solutions.”
Most of all, diversity will be a true asset to your future team. A mathematician, physicist or computer scientist, a “geeky domain expert” and natural scientist can be helpful to your Data Science team. It is also important to have at least one senior in the team, which can be hard to find. “You could think of borrowing one before others in the team will be trained enough,” she says.
What is an environment a Data team can flourish in? “I believe I can write a paper answering this question, but I’ll keep it short,” she tells. “Trust, collaboration and time to learn and try new things.”
Take these insights into consideration and we think you are on the way to create a fabulous Data Science team. Do you have any do’s and don’t or experience you want to share? Leave your thoughts in the comments!
Sophie Brüggemann, Lior Barak and Zinayida Kensche are speakers at DN19 on 25 and 26 November. If you like this article, we think you should meet them in the wild. Come see them and tons of other Data experts on stage and join us by getting a ticket here.
