Data Scientists. Post 3 of 3
How to keep one happy
So, we’ve discussed what a data scientist is and things to consider when deciding if you need one. Now you’ve got one, how do you keep them happy? How do you give them what they need so that they can give you what you need?
Content is the key to contentment
In the previous posts, I wrote about the nature of work that a data scientist is suited to. The content needs to have variety, require novel approaches (not just the application of an out of the box solution) and involve interesting data that requires taming. Not every problem you task your data scientists with solving needs to necessarily have all of these components, but if they are consistently being presented with problems that don’t meet any of these criteria then it is highly unlikely that they will be stimulated enough to stay long-term.
Toolkit
Data Scientists tend to be passionate about technology and relish the opportunity to work with cutting edge techniques. Therefore, if you hire a data scientist and give them excel, access and SQL server to play with you are unlikely to retain them for long!
In the first post of this series I described data scientists as the Swiss army knife of team members. However, they can only perform in this way if they are given access to a range of different blades. In technology terms this is likely to mean providing access to a range of development environments and platforms, preferably including open source. In fact, I would argue that facilitating the process of obtaining new open source technologies with IT (easier said than done in most large organisations) is a key responsibility for anyone managing data scientists. In addition, I strongly advise taking the advice of your data scientists on the tools that you should be using. Good data scientists have their finger on the tech pulse and should be able to talk with authority on the latest developments and how they could add value to your business processes.
Macro management
Perhaps it is a slight generalisation to state that data scientists don’t want to be micro-managed, but certainly in my personal experience this has held true. Data scientists want to be given an interesting problem to solve, and then left the hell alone to solve it. Looking over their shoulder, micro-managing their deliverables or challenging their technical approach (when you don’t understand it yourself) are all sure fire ways to frustrate data scientists. Of course, like all team members, data scientists need to be held accountable to certain timeframes and deliverables, but more than others they require the freedom to innovate without restrictive management practices.
The focus of management should be on clearly framing the business outcomes so that the data scientist knows what the problem is and how they will know when they have solved it (not just handing them some data and telling them to ‘find some insight’); and facilitating the removal of road blocks to delivery such as technology and unnecessarily restrictive governance.
Support from the top
Without senior support and buy-in, data science (like many business activities) is unlikely to flourish. However, given the (over) hype associated with data science, this can become a tricky balancing act. Executives who have been fed the message of data science being the answer to all of their problems may expect the impossible and, when it inevitably doesn’t materialise, lose faith in data science as a tool for good within their organisation.
There is also a danger of executives putting data science into the bucket of ‘things I don’t understand and therefore don’t trust’ and as a result not using the output from their data scientists to drive business decisions. This is an easy way to demotivate and disenfranchise your team. Data scientists need clear line of sight from the work that they are delivering to tangible business outcomes and value. There is nothing worse than slaving over a piece of analysis only to have it languish in the inbox of your manager or, worse, be used to drive a big decision that you never hear about.
Money money money
Last but not least, you need to reward your data scientists appropriately. Qualified, experienced data scientists who also have business acumen and interpersonal skills are rare and (predictably) expensive. In the UK market there is extremely high demand which means if you are not giving your data scientists what they need (financially and in all the other respects discussed in this post) then someone else will. And such is the competitive nature of the market that they might not even bother to interview before making an offer.
That said, there does seem to be a trend of very junior data scientists, or candidates who have decided to rebrand themselves as data scientists, expecting unrealistically high salaries for their level of qualification and experience. It is therefore important to consider candidates and their worth on an individual basis, and not just assume that anyone who calls themselves a data scientist is worth top dollar.
