Data science is quickly becoming one of the most in-demand skills in the world, for a simple reason: A good data scientist can deliver an organization massive value.
Finding Value in Data
The big data hysteria has gone global, and companies are now collecting insane amounts of data. Everything we do now is recorded, measured, and stored in the hope of extracting useful insights about customer behavior.
But, companies are now quickly realizing that just collecting and storing data is nowhere near enough. It’s not the organizations with the most data that have the advantage, it’s the ones that can draw the best insights from that data. That’s a very important distinction. Check out our post on business process automation with data science here.
For that reason, data scientists can be the difference between a company gaining huge value from their data streams, and gaining nothing. In a world driven by data, this could mean the life or death of an organization.
Hiring the right data scientist is crucial.
What is a Good Data Scientist?
Data science can be thought of as the job of “turning raw data into understanding, insight, and knowledge” (Wickham & Grolemund, 2016). Data analysts, statisticians, and quantitative analysts are similar roles, with emphasis on different skills.
This job requires a unique combination of skills. Drew Conway’s famous Venn diagram of these skills is a nice visual way of understanding them.
You’ll be competing against many great companies for the best candidates with all of these qualities. Here are 7 tips to help you identify the right data scientists, and successfully bring them to your organization.
1. Design a great recruitment system
“System “is the key word here. Companies are finding that, on average, it can take hundreds of applications to find a suitable candidate for a data science position. It’s going to take a lot more thought than just putting up a job listing and doing a few interviews to get it right.
You’ll need a system that can:
- Attract the right candidates to apply
- Weed out the wrong candidates efficiently
- Identify the most talented applicants
- Convince your top choices to join your team
An effective recruitment system can be thought of as like a funnel. Hundreds of applicants enter the top of the funnel, and each step of the process eliminates the wrong candidates while keeping as many good ones as possible.
A good place to start is with technical knowledge. There are many more skills required to be a great data scientist, but mathematical and statistical prowess is non-negotiable. This could take the form of a do-at-home test for your applicants. The test must accurately test the skills your hire will need as part of your team.
The next step of your recruitment funnel is where you’ll differentiate between the good and the great applicants. Technical aptitude is needed, but most importantly, they need to be able to make data work for your business. You’ll need some creative methods to reveal those skills.
This doesn’t need to be a boring traditional interview. It could take the form of a day of problem-solving with your team. Invite everyone that passed the take-home test. After a full day of problem-solving and interacting, you should have a much better idea of who to pick. Data science is extremely collaborative, so your whole team should be involved in the decision-making process.
Lastly, you have to offer the job. This step has to be done well when recruiting data scientists. Remember, you are competing for the best. You should take care with this step to make your offer as appealing as possible, and doesn’t come across as arrogant.
2. Cast a Wide Recruitment Net
Once you’ve got your recruitment funnel thought out, you need to start getting people into it. Good Candidates from obvious places, like good tech schools, will be inundated with offers for interviews. These are great sources of talent, but not the only ones.
Data science is a small world. One good way to meet potential hires you wouldn’t otherwise find is through networking. Linking up with individuals other companies active in this field will be invaluable. Anyone in your network will likely introduce you in a very positive way.
If you’re having problems finding people will all these skills, don’t worry. If a candidate doesn’t have all of the technical skills required, they still might be a good fit. If someone has a talent for communication and analytical thinking skills, teaching them specific technical skills like R or Python won’t be a problem.
3. Use objective methods to avoid bias
Unconcious bias is a problem every recruiter has to deal with. Strangely, even cutting-edge recruitment algorithms designed to avoid this seem to share some of these prejudices. With the competitiveness of data science, it’s just exaggerated. If left unchecked, this can damage your chances of successful recruitment.
Some common biases include the conformity bias, beauty bias, as the halo effect. To deal with them, you’ll need objective methods of evaluating candidates in your process. They can be tests you come up with ahead of time that measure skills you are looking for, where the results won’t be affected by your emotions.
4. Don’t Shoot Yourself in the Foot
There are many classic interview mistakes that are repeated by a recruiter after recruiter. These can produce false negatives (accidentally turning down a good candidate) and false positives (offering a job to the wrong candidate). Both are bad for your business.
These could be things like:
- Asking interviewees to solve “toy” problems that don’t actually demonstrate real-world ability
- Strange interview questions that have no purpose (we’ve all heard these, e.g. “If you were a fruit, which one would you be, and why?”)
- Coming across as arrogant, bias, or rude
- Demanding free work
- Getting too personal with questions
You might cringe at some of these, or think they are obvious, but these types of things happen in interviews every day. The job of data science recruiters is already difficult – don’t scare away awesome applicants by doing something silly.
5. Get the right kind of data scientist
What many companies don’t realize, is there are two distinct kinds of data scientists.
- Data scientists that deliver to humans
- Data scientists that deliver to machines
The two are very different jobs and require different skills.
The first kind will analyze a business’s data, and come up with ideas and insights to present to business decision makers. The results must be presented in a way that non-techy can understand. This kind of data scientists will look deep into data sets to find insights, and present them as stories, graphs, and charts.
The second type of data scientist might analyze the same dataset but has very different goals. These will be machine-readable reports that can be fed into the companies other business systems to produce automated responses. Things like:
- Product recommendations
- Advertisement placing or targeting
- Buying or selling stocks
This type of data scientist needs razor sharp statistical and programming skills, in order to build models that can make predictions and decisions in real-time.
6. Make Your Interview Process Sell
Remember, good data scientists are in high demand. Data science is so competitive that you have to find the right candidate AND convince them your firm is the best to join. Your whole recruitment process needs to double as an advertisement for how great your company is.
The tests, interviews, and other parts of your process should mirror how candidates would actually work at your firm. (If that’s not a good advertisement, then maybe your firm needs a culture shift!)
7. Make Sure You Actually Need a Full-time Data Scientist
It’s difficult to hire data scientists – that’s why I’m writing this article. So, you need to ask yourself: Do you really need one full-time?
What you are really looking for is a system that takes in your business data, and returns to you valuable, actionable insights. Depending on your type of business, there’s a good chance you might not need an in-house data scientist.
The data science skills shortage has lead to some great companies offering analytics services and data scientists for hire. They can build the systems you are looking for and help you maintain them, without you recruiting data scientists to your team at all. It could be worth checking out.
These days, a great data scientist (or team of them) can mean the difference between a company thriving, and a company fading into insignificance. Making the most of your data isn’t a choice anymore, it’s a necessity. Follow these tips make sure the process of finding one isn’t too painful!