Optimize Your Recruiting and Retention Tactics

As you search for candidates to build your AI team, think of the process like dating: they’re in high demand, and you’ll have to work to attract the right match. You’ll also have to contend with big players like Google, who snap up data scientists as soon as they step out into the sunlight. Most companies make the wrong moves in this workforce dating game. Think of the following suggestions as a guide to attracting your perfect date.

Pay competitively & improve your work environment

You should not try to skimp on the salary. This is a fast way to fail—don’t cheap out. If you’re going to invest, invest. If you don’t have what it takes to invest, then don’t engage with data science.”

Lee Cohn
Senior Data Scientist, Nike
When an executive knows that all data scientists are just ‘wicked smart,’ you’re already set up to fail. So, that’s where I set business expectations—what [data scientists] can or cannot do. But if they walk in thinking that [I’m] going to walk on water, typically that’s a red flag to me.”

Bernard Ong
Principal Data Scientist, American Family Insurance Client Services

The average postdoc salary in the US is about $60K per year. A few years ago it wasn’t hard to convince someone who had a physics PhD to forgo the poor pay and academic headaches to be your data scientist for two or three times the salary. But those days are long gone. Data science salaries have been going bonkers for the last few years. A data scientist can often change jobs and increase their compensation by 30% or more. It’s just supply and demand.

To build your data science team, you have three choices:

  1. Pay under the market rate and hire unqualified people who have no other options.
  2. Pay whatever it takes to get people to work in a suboptimal data science environment.
  3. Pay competitively and create an environment where data scientists want to work.

We hope you’ll be smart and choose the third option. Competitive compensation is table stakes; you don’t have to beat all market salaries to build your team.


Don’t settle for unqualified people or try to solve your talent shortfall by throwing money at the problem.

Only about 20% of the data scientists we surveyed felt that it would be hard to find a new job that paid competitively. But more than 60% felt it would be hard to find a company whose leadership understood them and how to support them. To stand out from the crowd, you should not only pay competitively but also invest in your data science work environment.

Centralize your data science team and hire a strong leader

Data scientists want to work for leaders who have more experience than they do. This insight is critical for your organizational planning.

Many companies struggle to decide whether to centralize or distribute their data science talent. But realistically there’s only one option: centralize your data scientists and let them matrix into project teams.

Too many companies begin their AI journey by hiring a data scientist into a traditional software team. After a few months of trying to explain why the data science experimental process doesn’t align with the sprint release cycles, the data scientist quits. Unless you change your structure, you’ll have difficulty recruiting and retaining data science talent.

To make your job descriptions more effective, highlight your hiring manager’s understanding of the data science workforce and role. A rewritten job description can instantly lift the quality and quantity of candidates. For an example, see the “Resources” section at the end of this report.


Hire a strong data science leader, centralize your team, and use your team leader’s credentials to attract talent. These steps will instantly make you a more attractive employer.

Create a non-management career path

In terms of seniority . . . I observed that the higher you are, the less hands-on work you do. I want to go to the next level, without management responsibilities, while still being able to actually do hands-on work.”

Carlos Oliveira
Staff Data Scientist, Intuit

Fewer than half of the data scientists we surveyed have a strong interest in becoming a manager. Most were only somewhat interested in pursuing career growth through a management path.

Unfortunately many companies don’t offer an alternative to the management path. If management is the only available advancement option, many data scientists will leave so they can continue doing data science.

Offering promotion alternatives is a no-brainer. You can start doing it today.


Ask your data science team to build a career path based on your organizational structure. Then update your job descriptions to reflect the career path.

For a copy of the Prolego data science career path, see the “Resources” section at the end of this report.

Let’s Future Proof Your Business.