Are You Ready For a Data Scientist?
Are you wondering if now is the time to get a data scientist on board to help investigate the value of data assets, maybe even your big data? My advice: don’t do it if you are not ready. The demand for data scientists is currently much higher than the supply which naturally leads to difficulties finding a good hire. The return on the decision is critical; a data scientist is not a minor investment.
So how do you know if you are ready? Here are a few pointers:
Are You Ready For Change?
You don’t hire a data scientist and put him or her in an isolated room with a powerful computer and unlimited access to data and then expect miracles to happen. You need to be ready to deal with the results from the data science and ask the difficult questions. A data scientist will conduct experiments on the data and business. And they want their results to be taken seriously.
Take as a warning the company that brought in a small data science team to more efficiently manage inventory. The team chose a demand-driven forecasting approach. But when it came time to implement, the company balked. It wanted the talents of a data scientist to break away from its dependence on spreadsheets – but it wouldn’t trust the forecasting approach. The data scientists left the company.
There are a number of things a data scientist might recommend – whether it is automating the process of markdowns or segmenting customers to receive very different kinds of offers. If you are not ready for change in how you do business and make decisions, you are just not ready for data science.
Do You Have a Plan To Make The Change?
If you’ve decided to segment the offers a customer receives when they call in to upgrade or change their service, and you don’t explain what you are doing and why to your call center staff, it could cause confusion. Say a call center rep gets a call from a customer that wants to lower their monthly bill. You’ve set up a system where analytics provides the offer the rep can make. For some customers it could be quite a deal (people who likely to have long-term value to your company) and for other customers the deal might not be so good. If the call center reps don’t understand why a screen says one thing for one customer, and an entirely different thing for another customer, they could be saying all sorts of confusing things to your customer like “I know there are better deals out there, let me put you on hold”.
It’s not just front line employees that need to understand. Your C-suite needs an education in what is being done and why. A Chief Marketing Officer might be hesitant to offer different types of discounts to different customers unless they understand how the analytics are driving those suggestions.
One way to avoid pushback is to roll out programs slowly and measure the results. A major transportation company did just that when data scientists came up with a different way of loading trucks and organizing routes. It didn’t use the data science approach everywhere at one time. By rolling it out over time, it helps get buy-in from those who prefer an intuitive approach.
Are You Aware It’s a Team Endeavor?
A data scientist is not an isolated magic unicorn. Although a good hire is one with computer expertise, math smarts, domain knowledge and communication skills, this person will work best as part of a team that compliments the skills they bring. For that reason it is imperative that you avoid hiring into this position someone that is condescending or too introverted. The data scientist’s team can be virtual or directly assigned. The hub and spoke option is one way to organize data scientist talent. In this model, the scientist (or scientists) work from a central location that doesn’t seek to control all the analytic activities but enables them. This team is helping other teams by providing self-serve analytics. Another approach is to embed data scientists into business units to work alongside their business counterparts that are thirsting for answers that the data scientists can help deliver.
Along with determining who they will work with is deciding who they should report to. The CEO or CFO are likely choices. What is not a good idea is assigning a supervisor who is skeptical of the value of data-driven insight.
Are You Providing a Learning Environment?
Data scientists are curious by nature. They love problem solving and tackling the difficult. Data scientists I’ve interviewed mention the need to spend part of their day brainstorming and they also need flexibility, describe their job as creative. Yet they want to demonstrate measurable results. You need to provide an environment that embraces learning and willingness to ask questions – and the freedom to look in new places for answers. Be ready to explore new data sources and provide an environment that allows for creativity.
They also need a career path. They are not nerds that will be happy as long as you given them the latest, greatest modeling software. They want to affect change within organizations and to have their work and skills be valued.
Do You Understand Your Current Situation?
Data science means different things to different organizations. If you are in a big well established organization getting a data scientist requires an understanding of communication, technology and process changes. The daily processes within the business are most likely already well formed and the value of data science can be found through increased profit margins and making the many little decisions even better – or even totally automating them. Innovation can still be driven by data science for sure but it will happen differently depending on your current stage. Be a realist – but one that is ready to rock the boat.
Consider this scenario: Waiting in an airport you are in the middle of watching the latest episode of your favorite TV series on your phone. Suddenly, a message arrives from your provider telling you’re are about to run out on your data plan. However, instead of asking you to call to extend your data plan, they send you an offer that you can accept right then and there. That is what one large mobile phone provider now does after implementing data science to automatically create a personalized offer and deliver the offer at just the right moment. Customers could have automatically opted to extend their data plans earlier, but with the relevancy and urgency of an offer just in the nick of time, the wireless provider has seen a significant increase in conversion rates.
I don’t want to ever discourage companies from hiring data scientists; such hires offer tremendous business benefits. But to get the most out of a data scientist hire you need to be ready to change processes and culture. “Decision making the way we’ve always done it” needs to be changed to “Decision making the way we used to do it.”
About the author: Sascha Schubert is the Director of Advanced Analytics Product Marketing at SAS. Schubert joined SAS in 1997. He sets strategic directions and defines the global market needs for the SAS advanced analytics product portfolio, with a specific focus on data mining, machine learning, predictive model management and high-performance analytics. Schubert has worked in various technical, product management and sales support roles in which he closely engages with customers all over the world. Schubert holds a PhD in Statistical Climatology from Humboldt-University Berlin.