It might be the sexiest job of the 21st century1, but hiring for data science jobs is rarely a smooth process. Despite the leaps and bounds with which the field has grown, companies still struggle to draw the right talent in a market where demand outstrips supply.
When it comes to data science hiring, the missteps begin with the job description. As an ever-growing field of organizations try to catch up to the great data revolution, a growing gulf is emerging between the fantasies of job descriptions and the realities of data science jobs on offer. This often translates to promising candidates lost in the mix, frustrated employees contributing to high turnover, and companies staring at financial waste from ineffective hiring processes. One survey of 1,400 executives found that 36% of executives felt that, aside from performance issues, the top reason for bad hiring is a poor skill match2.
Here are a few fundamental dos and don’ts to avoid such costly consequences:
1. Don’t use data scientist as an umbrella term
With headlines popping up every day on why every company needs to hire a data scientist, some might eagerly get on board without a clear sense of what they want. Unlike technology professions with a longer lineage, definitions and distinctions in the data science job market aren’t always cut and dry. Often the term “data scientist” gets thrown around for a variety of job roles and responsibilities, including data analyst, data engineer, machine learning engineer, statistician and business analyst - to name a few. . In other words, just usage of the term “data scientist” as an umbrella one for anyone who deals with data or analytics might by itself lead to a significant mismatch between the skills sought and skills sets offered down the road.
2. Be clear about the business context/state of data science at your organization
Building up an organization’s data capabilities from the ground up is not everyone’s cup of tea in terms of interest or skills. Some senior data scientists might relish the idea of helping a company develop its data strategy and take responsibility for growing data teams. But others want to join data teams with a clear data strategy, executive buy-in, and all the necessary tools and resources so that they can wade right into the analytical problems that they’ve been hired to solve.
3. Sell the job, but be clear about the basic role responsibilities
While salaries and perks sweeten the deal, one of the biggest attractions of a data science job is the potential it offers in terms of exciting problems, new tools and technologies and so on. Emphasizing what’s unusual and interesting about the kind of problems data scientists will be working on at an organization goes a big way in selling the job. Emphasizing the value given to data teams’ work and its importance for the overall goals of the organization can also add value.
But it’s also vital that the job description does not sound too good to be true. If a large part of the new hire’s job is going to involve cleaning up poor quality data or building dashboards for various roles and functions, it’s important to highlight these responsibilities too. Too rosy a job description can leave candidates disillusioned with the rest of the hiring process.
4. Use the right terminology, not just the latest buzzwords
Candidates can judge between good and bad data science jobs by looking at whether job descriptions correctly and appropriately use data science terminology or overflow with repetitive jargon and buzzwords3. Appropriate terminology and specific detail give candidates the sense that a company knows exactly what it wants to do with its data and recognizes the value that data scientists will bring in. Buzzword-stuffed job descriptions, on the other hand, automatically suffer a deficiency of credibility.
5. Don’t overdo the required qualifications
Ask five stakeholders what the required qualifications for a data scientist are and you’re likely to get five different lists of requisite skills and knowledge. A job description that tries to tick all those boxes, especially for lower level data science vacancies, will most likely chase off many great candidates. This is especially the case for women candidates, who are unlikely to apply unless they feel they match 100% of the requirements4.
It’s much more useful to specify a list of fundamental must-haves absolutely required to perform the job advertised, and to provide a second list of desired additions that add further value. One useful way of splitting the lists is to consider what basics a new hire should come into the job with and what they can learn on the job.
When defining qualifications, it’s also vital that the job description isn’t laying out expectations that can’t reasonably be fulfilled. For instance, the job title “data scientist” has likely only existed for a little over a decade5. So, to expect candidates to have held the job title for several years is patently unreasonable. Besides, given that demand far outstrips supply in the data science job market, high expectations cannot be met without a commensurately attractive job offer.
- https://www.hiringlab.org/2019/01/17/data-scientist-job-outlook/ [return]
- https://www.linkedin.com/pulse/hire-bad-kent-cameron/ [return]
- https://www.kdnuggets.com/2019/04/recognize-good-data-scientist-job-from-bad.html [return]
- https://hbr.org/2014/08/why-women-dont-apply-for-jobs-unless-theyre-100-qualified [return]
- https://hbr.org/2012/10/data-scientist-the-sexiest-job-of-the-21st-century [return]