Reading Time: 5 minutes

by: Zank, CEO of Bennett Data Science


It can be really tricky to hire your first data scientist or analytics professional. You might ask yourself: What are the most important qualities to look for? How do I test his/her math skills? How important are communication skills compared to math skills? Are we really ready for someone who’s dedicated to data? What value should I expect to get out of this person? Who will manage her/him?

It’s not easy! There’s a lot to think about, and these few questions are only the beginning! I’ll go over the framework I use to hire and hopefully help you avoid some the pitfalls I commonly see during the technical hiring process.

To do this, I’m breaking the post up into several posts that I’ll release over the coming weeks.

Part 1: Do you really need a data scientist?


Hiring a data scientist is a big step: they’re expensive, still very hard to find, and can be a challenge to hire, manage and mentor. So, if you’re on the fence, ask yourself if it’s really the right time to make the hire.

I’m a big fan of visioning. Visioning can be thought of as imagining the future situation as though it were reality. I often spend 10 minutes writing, in as much detail as possible, exactly how things will change in the new situation. Say I’m going to buy a new car, I might spend 10 minutes writing about what it will be like to own the car and my feelings and emotions around: how much it will cost (gas/electrical, registration, insurance), where  to park it so pine trees don’t drip sap onto it, whether my friends (or bikes) will fit in it, and lots of other things.

In doing that, I’m able to get a good feeling of what it will be like once I’ve made the purchase. I wouldn’t feel good about a car that cost me a lot more in insurance that wouldn’t hold the necessary cargo, and I would be able to figure all that out by spending some time visioning. I believe this also works well for assessing the need for a new hire.

Here are some places to start:

  • What is your base requirement for hiring a data scientist?
  • How does this hire strategically lift your business?

Sometimes, we all chase after the “shiny new thing.” And data science is certainly shiny! But how necessary is it to have a data scientist on staff?

In this post, I want to spend a few paragraphs helping you understand when the time is right, or perhaps when you’re not quite ready, to hire a data scientist.

Why do you feel that you need a super brainiac math nerd (I say those words with tons of affection!) walking through your hallways? Is it to “find something interesting in your data” or something more pressing like, “our web store needs a recommender system”? Then imagine for a moment that you already have this person/wizard on your staff.

Here are some really good answers. If you can answer yes to these, you should probably close this window and start penning a requisition right away:

    • Our team uses simple heuristics (like most-popular product) to sell more widgets, but we’ve been collecting lots of data and we’re sure we’re leaving 30% on the table, we just don’t know where.
  • We’ve been collecting transactional data on our users and products for 6 months and we don’t have the first idea about how to analyze it to calculate things like lifetime value or propensity to churn.

Here are some examples where perhaps, it’s a bit early or not even necessary to have a data scientist on staff:

    • We’re about to start collecting data for the 1,000 customers we have, and we really need to know which columns are going to be the most important (I’ve heard this many times!)
    • The CEO read an article last week that said that AI will grow by $14BB next year! If we don’t start using data science now, we’ll be left out.
    • We’ve got this big pile of data. There HAS to be something in there somewhere! (This may be the most common, and also the hardest for a data scientist to come into)

Sometimes a data scientist is not necessary at all

Maybe what you really need is someone who can run reports and keep a team/company abreast of ongoing changes. This is more of a BI (business intelligence) role. In this case, you can save yourself a lot of money! The point here is, be really sure you need a data scientist before hiring one. If you don’t, you may end up in a position where you’re paying for an underutilized resource and that person will be doing unfulfilling work. If this is the case, it’s a painful situation for you and your new hire.

Another way to frame this is to think about what your new data scientist will be doing during their first few weeks. If the result of those first few weeks is a data product that really drives product and/or revenue, you’re probably on the right track. Stop and ask yourself, how does this person drive product or help push revenue forward?

What’s the point?

Admittedly, this sounds a bit blunt, but at some point early in most conversations with new companies, I’ll ask this question. What I’m really getting at is, what’s the objective of the proposed work? Has it been well defined? Does it produce a big revenue or product impact? Those are the areas where data science can really help. But defining the objectives is tricky, and could easily expand into its own post.

Here’s a simple example:

Say I have an app distribution platform, and I want to recommend apps to my user base. Great! I hire a data scientist who’s proficient in recommender systems and six weeks later, she delivers code for this awesome high-tech recommender. How do I know we’ve succeeded in building an effective recommender that will serve my customers? What’s most important to the business? To end users? In other words, What’s the point? So, I take a stab at it and say the point is to maximize engagement. But how is engagement defined? Most in-app purchases? Most time spent with app open? Greatest number of app opens? It turns out, the objective (the point) is not so easy to nail down.

Defined objectives

If you can define your objectives and clearly articulate them before beginning the hiring process, you’ll have much more success. You will feel that you now have a standard against which to measure candidates, and your candidates will have a good understanding of what the job entails. If you don’t know “the point”, your candidates won’t either and it can lead to a lot of wasted time and money.

This is why it’s hard to see requisitions that list 20 or more required technologies alongside a myriad of data science skills, as though taken from an intro to data science syllabus. Finding the right candidate begins with honing in on what your company needs and finding a smart hire who understands those needs, can say them back to you, and who will grow and thrive in the role.

In summary, here are the three main steps to determine if you need a data scientist:

    1. What’s your vision? Make sure you really need a data scientist
    2. Understand the objective you’re hiring for.
    3. If you feel strongly enough about your need to work on that objective, it’s time to write a requisition!

Go through those steps in as much detail as possible, and hopefully you’ll know if you’ll really benefit from hiring a data scientist.

Zank Bennett is CEO of Bennett Data Science, a group that works with companies from early-stage startups to the Fortune 500. BDS specializes in working with large volumes of data to solve complex business problems, finding novel ways for companies to grow their products and revenue using data, and maximizing the effectiveness of existing data science personnel.

Signup for our Newsletter