Effective data team incentives, Motivators and demotivators

DTP #5

We spoke to Benjamin Gowan, Senior Data Scientist at RiskLens, on incentives for data scientists, what motivates and demotivates data teams, and the challenges data scientists have related to stakeholder communication. 

Quotations have been lightly edited for concision and readability.

Give us your thoughts on this issue:

What kinds of incentives motivate data scientists?

“As you know, we [data scientists] are learners, and oftentimes, self-taught in at least one of the three or four domains that we happen to be working in. Making space within the business context for on the clock learning instead of late at night after work, I think [would be] great internal motivator alignment and just making space and support for data scientists to learn while doing a bit more explicitly.” 

How much time would you say you spend learning versus actually implementing what you learn on the job? 

“It's hard to say. We do a pretty good job at my current role of providing space to learn. But even with that, I think there's still a tendency to say, oh, I need to learn about this thing that has to do with web tech and this other statistical process. I need to brush up on this other thing. And we push those into the evening during self-study. I think that's a pretty common tendency among data scientists. But yeah, I think in general just making the space to communicate that brushing up on a complex process at work is a part of work, I think that's a good message.”  

What do you think about compensation for learning? Is that something that companies are interested in doing ?

“The one I tend to lean towards is having an education and travel budget so that you can go to a conference, [etc.] I [may] have seen some things around certifications for roles and needing to get those. [That’s] probably be the closest to that I've come across [on learning based compensation]”

How difficult is it to find the resources data scientists need to up skill? 

“There are a lot of resources available. Lots and lots of resources. I think it's not so hard to find resources. It might be a little bit about just curating what's applicable to where you're at right now. What's high enough quality content and or sufficiently advanced both on the theory, modeling, statistical side, and then also on the deployment in tech side. So having both ends of those is pretty helpful."

What else is important when it comes to getting data teams motivated?

“I think, at least for me, it’s very motivating to see data science work in use downstream, so getting to see whether internal users or end users are using data predictions to improve their work or getting a lot of value from them. Getting to close that loop so you get to see the value [you create], I think is a good motivator.”

He touches on the flipside – Demotivators.

“Demotivators can be big. The two [major] things that demotivate might be, [for example] coming into a problem that has huge expectations for what data science can deliver with little to no data or specifics on what the actual business problem is. Big expectations with no data. It's a pretty common thing."

On the second big demotivator: 

“The other one is you end up having to do (maybe this ties into having a small team) everything like pulling the data, cleaning the data, modeling the data and then also have to communicate, deploy, manage, monitor, so it can feel like you're full stack. You know, horizontally and vertically. I think we have things that happen in a lot of roles, but in data science you end up [with] a little more breadth because you're communicating to stakeholders all the way at the engineer data level as well. [Not necessarily, but] I think there's an expectation that you [should] constantly move into another adjacent responsibilities that can be challenging.” 

So, roles aren't typically defined too clearly for data scientists. They need to do a lot of things.

“I think the industry is sort of kind of recognizing that and that's why there's been such a big boost in, you know, not just data science but machine learning engineers. It’s become a bit more specific like, ‘Oh hey, we need to kind of specify exactly what this role is doing a bit more or focus it in on a particular thing.’” 

He emphasizes that a lot of questions need to be answered upfront when building a data science effort:

“And then also this kind of shift [with] a lot of people going into just general data engineering. I think that's indicative of the industry recognizing the problem of ‘Oh hey, you can't really solve problems with data science if you haven't done a lot of upfront data work.’ I think a lot of that leftward shift in the data pipeline is trying to address that, yeah, data questions and the data provisioning process needs to start early on.”

There's this aspect of needing soft skills to communicate and coordinate with people. What do you think about those skills, in particular for data scientists currently in the industry?

“I think there's this funny inversion of communication you have to do. When you're in a role that's doing analysis, you're sort of breaking down a problem and you're building up to a conclusion, almost scientifically, ‘here's the data we're doing, exploration, discovery, figuring it out.’ And we build up arguments and methods until you get to a conclusion at the end. When you flip around to communicating with stakeholders and business side, you have to ignore like 95% of the work you've done and just lead with conclusions.” 

This shift when communicating isn’t easy for the type of thinking data scientists do. 

“Things like that are a very difficult transition to make. But it's incredibly helpful. The [term] we try to remind ourselves [of is] bottom line up front because it's very easy to get lost in the details and then never actually get across your point. It's almost uncomfortable as an analytical person because you feel like you have to build up to that.”

It seems like having marketing skills would be useful for data scientists trying to sell a solution.

“Exactly, yeah. And even if you don't have a strong opinion on it. Maybe, genuinely have, like, three options [you can share with stakeholders]. You don't [need to] have a strong opinion. You don't have to really sell anything in particular, but if you provide a bunch of details without providing the three options, you'll have failed to communicate clearly.”

Our conversation with Benjamin Gowan emphasizes the need for businesses to:

  • Provide their data scientists with space and opportunities for learning.

  • Be prepared with the specifics when commissioning data science projects.

  • Make sure data scientists are equipped to communicate with stakeholders.

See you next week.

Do you have a unique perspective on developing and managing data science and AI talent? We want to hear from you! Reach out to us by replying to this email.

Join the conversation

or to participate.