Data is now said to be the new oil, making data scientists indispensable. However, it is still a fairly new field and many non-technical professionals do not have a clear understanding of what data scientists do, why they do it and how to work with them.
When building a company that monetizes consumer behavior data, I knew we would need to involve data scientists. But as non-technical founder with a business background, I did not know when or how to hire them, or even how to present the problems we needed to solve. As often happens when non-technical people delve into the world of technology, there are plenty of unknown unknowns. How do you find an answer if you are not quite sure what the question is?
I spoke to Susie Sun, data scientist at WhatsApp, about what data science is, what data scientists do and how business functions should work with them. Sun understands both the commercial and the technical sides of the equation well, having started her career on the business side and completed an MBA at INSEAD, before transitioning to data science.
Sun’s definition of data science is “using the accidental output of computing – i.e. data – this new(ish) field uses statistics and coding to do previously difficult things, from understanding customer behavior, to making predictions, to copying human-like ‘intelligence.’”
Much like the term engineer, the term data science can encompass a broad spectrum. Sun suggests thinking about data scientists on the following spectrum:
Giving an e-commerce business as an example, Sun presents the following divisions:
- Data Analysts answer questions like “Given this customer funnel data, where are my customers dropping off?” The output is data.
- Data Scientists answer questions like “Given all my data, how can I improve profitability?” The output of their work is insight.
- Machine Learning engineers answer questions like “Given that I want to increase my customer basket size, how can I build or improve my recommendation engine?” The output of their work is a model.
Continuing with the e-commerce example, Sun says that data scientists can use information such as past sales transactions, customer details and demographic data to understand who the company’s most valuable customers are. If you know this, you can adjust your marketing to target the people who are likely to spend the most with your business.
With machine learning and predictive analytics businesses can take these insights even further. If you know who your most valuable customers are based on the past interactions, you can build a model to spot similar customers early and tailor the e-commerce experience for them.
However, it is worth remembering that predictive analytics are based on past data, which does not make them entirely future proof. This is where data scientists, creatives and marketers can work together to combine data with instinct.
Hiring data scientists
If you are considering hiring a data scientist, think about what questions your business is facing. Examples of questions that data scientists can answer include:
- How can I detect fraud before a purchase goes through?
- How do I match users to advertising in order to maximize my revenue?
- How can I better predict sales at each location so that we do not run out of stock?
An important point to consider is whether you have enough data for the data scientist to provide you real insights. This is why companies that collect their own data need exist for a while before getting a data scientist involved.
Working with data scientists
To have the most productive relationship with a data scientist, present the problem and ask them to find a solution, rather than presenting your own. This is the same advice I would give about working with developers in general. It is not the business person’s responsibility to understand the fine points of writing back end code, but it is up to them to gather customer feedback and work with their technical teams to make sure the product works.
As every company becomes enabled by technology, and traditional businesses acquire technology companies, learning how to work with data scientists and other technical professionals is a necessary skill for a successful career.
Data may be the new oil, but if you do not know how to use it and work with people who do, it is just a collection of meaningless facts.