The ABCs of retention won't cut it for data scientists, which is why you must use creative tactics. Try these out-of-the-box retention strategies.
The acronym MTBOO (Mean Time Between Other Offers) was tossed around a few years ago, and it's a somewhat tongue-in-cheek derivative of the better-known MTBF (Mean Time Between Failures) that characterizes the overpaid darlings of the high-tech industry who change jobs as often as they change pants. About a decade ago, the MTBOO of a software engineer was about nine months. For a really good data scientist today, the MTBOO is probably at most nine months.
I often talk about the basics for keeping data scientists within your walls; however, I realize that you need to go a step farther. In this competitive environment for talented data scientists, leaders must go beyond the basics to retain their best people. Here's some food for thought.
1: Find out what really jazzes them
At the end of the day, it comes down to how you reward their behavior. You specify the behaviors that meet your goals, and when they exhibit those behaviors, you reward them. The bigger the reward, the greater the behavior is reinforced.
So, if you want your data scientists to be more agile, define exactly what agile looks like, and when they do that, reward them -- it's that simple. But reward them with what? Ah, that's the $64,000 question. The answer is it's up to them, because the value of any reward is in the eye of the beholder. You should have a dialog with each data scientist to uncover what really makes him or her excited.
2: Let them do what they want
Flexibility comes in various forms, including type of work, hours, and location. Most leaders and managers provide some flexibility for their employees, but think bigger.
Let the employee choose their own work, and let them perform it in the style that suits them (e.g., remote, off hours, etc.). It stands to reason that the employee will be most passionate about something they really want to work on, especially if they can do it where and when they want.
It's not anarchy if you have the right leadership. Explain your vision and goals, and then give the employee freedom to contribute the way they want.
3: Make them heroes
Everybody loves to be appreciated by his or her peers, and data scientists are no exception. You want to create a culture where data scientists are celebrated. Normative reinforcement like this is far more powerful than anything you can reward them with. It takes work, though, because pulling off a culture shift is not the easiest thing to do.
Work with a specialist who understands organizational change management to install a culture where your data scientists' behaviors are not only rewarded by you but also by extended peers (i.e., your group or organization). It's hard to walk out of a place that celebrates your work.
4: Walk the halls
The management technique "Walk The Halls" is one in which you physically walk around the company to observe what's happening (i.e., behavioral observation). Most managers walk the halls to spot problems, but I suggest you flip this around: walk the halls to find what's right.
Work with your supervisor to identify areas of excellence, and who's involved. And then, take some time to walk the halls and personally congratulate those people on a job well done. Engage with the data scientists on how they did it. People love talking about how they did something great.
5: Throw a party, again
Everybody loves a party, so don't be shy about sponsoring them. Some people think a huge annual or quarterly celebration is adequate, but that's not enough. The problem is the timing between their good deeds and when they get to celebrate. If the reward isn't immediate, it loses its power. It's much better to throw a lot of small parties that celebrate successes. When I was consulting for Silicon Graphics in the early nineties, there was a party going on somewhere in the organization just about every day.
Celebrate and hand out rewards to those that deserve it. The rewards don't need to be expensive, but they should be special.
6: Build a really cool science lab
What scientist doesn't love a lab? Let me explain something: The main reason why data scientists prefer to work from home is because they're more productive at home. My home office is custom-built for me, and it puts any VP's work office to shame; however, data scientists can't spend as much as leaders on a really cool data science lab with expensive gadgets and widgets.
When I was consulting at Chevron, they had a huge 3D visualization lab where geological engineers could view and analyze holographic terrain for potential oil production. How cool is that? You put something like that in place, and nobody's working from home -- guaranteed.
7: A big pile of money doesn't hurt
Money isn't the biggest retention factor, but it made the list because it can't be ignored. No, money can't buy you happiness, but it can sure help! I've had money and I've struggled without it, and I'm pretty sure that the R-squared on this is pretty high.
You need to pay your data scientists well and give them huge bonuses for exceeding expectations. What's the right amount? I wish I could tell you, but it depends on the situation. You can find out just by asking. Remember that bit about what gets them jazzed? That applies to money as well.
Retaining good data scientists is no joke -- they're hard to find and even harder to keep. The ABCs of retention won't cut it anymore, which is why you must use guerrilla tactics.
Take the time to find out what really jazzes each of your data scientists, and put resources behind some or all of these out-of-the-box strategies. To fight against a strong MTBOO current, build an other-offer-proof organization.
- Reward data scientists' good work with significant recognition
- Data scientist: Your mileage may vary
- Save on employee turnover costs by following these retention tips
- Make innovative hiring decisions and keep turnover low through big data insights
- How to reward your best employees
- Give thanks for your tech employees