When it comes to crunching enormous amounts of data and making sense of it all, it can take finding a huge team of data scientists to get the job done and provide insights that are actionable for your business.
What Good Data Scientists Bring
Having good data scientists on your side, whether in your own organization or in an external team, can help you tackle some of the toughest questions that large companies face, such as:
- What techniques can we leverage to predict and understand what a customer will do next?
- How can we use big data solutions to improve how sales people pitch prospective clients and understand what company features they should highlight?
- How can we take findings of our data collection, aggregation, and visualization techniques and make them actionable for our marketing and customer engagement teams?
But finding good data scientists to work with is tough – and finding the rock stars is even harder. Here’s our guide to finding the best people to give your team the edge.
The Basics
There are standard things that every data scientist should be capable of, including:
- Being good at math
- Having a strong understanding of business concepts
- Being a good communicator
Finding The Rock Stars
But what turns that data scientist into a rock star, and gives them that x-factor that can drive your business forward? Beyond the standard list that you’d expect anyone to have, here are the other characteristics that I’ve noticed in finding standout data scientists:
1. Being empathetic to the customer’s point of view.
This means really understanding a customer’s business drivers and potential solutions, understanding the customer’s perspective, the pressure they are under and stepping up to be proactive. This makes you a better team member, business partner, and leader.
2. Thinking in a scalable, systems-orientated way.
It’s OK to hack things together every once in awhile, but a strong data scientist will approach problems in a systematic way and engineer solutions that are easily used by other team members and scalable across a variety of data sets and data problems.
3. Interpreting the data to actionable conclusions.
Great data scientists go beyond the raw data, to really interpret what is happening. If churn is the problem, then looking beyond vanity metrics to suss out what is causing churn by identifying hypotheses, testing those hypotheses and sharing with the team the root cause of a problem. Data scientists must use sound judgement to ensure they’re not overfitting the data; instead they must use their expertise and instincts to ensure there really is true causation vs. merely correlation.
4. Advocating for the change.
It’s not enough to share your conclusions and hope things get done. Great data scientists go beyond the presentation and the model, to advocate for change and make the change happen.
5. Always improving.
Great data scientists know it’s not enough to build a model and call your job done. They ensure that once they build their model on past data, it is tested in the real world, and then they improve that model. And A/B test their old model against their new model, continuously making improvements. For some of our clients a 1% change in accuracy means millions of dollar differences in outcome.
6. Being persistent with failure.
As a data scientist, your job is to help answer the toughest questions a business has (often times, coming up with the question itself). But a lot of the time, the data will be inconclusive. That’s OK. Great data scientists are persistent because if some projects don’t work out, others eventually will…and they’ll pay huge dividends.
Having a great data scientist at your disposal can have a huge impact on how your business functions. But at the end of the day, finding rock star data scientists is difficult work. So what makes more sense – to go out and hire those data scientists internally, or look to a 3rd party to help you? In our next post, we’ll look at the pros and cons of building your own team of data scientists vs. outsourcing the work, and in what situations each solution works best.
